R.N. Batra

Download as pdf or txt
Download as pdf or txt
You are on page 1of 367

STUDIES IN THE PURE THEORY OF

INTERNATIONAL TRADE
STUDIES IN THE
PURE THEORY OF
INTERNATIONAL
TRADE

Raveendra N. Batra

Palgrave Macmillan
©Raveendra N. Batra 1973
Softcover reprint of the hardcover I st edition 1973 978-0-333-13367-5
All rights reserved. No part of this publication may be
reproduced or transmitted, in any form or by any means,
without permission.

First published 1973 by


THE MACMILLAN PRESS LTD
London and Basingstoke
Associated companies in New York Dublin
Melbourne Johannesburg and Madras

SBN 333 13367 6

ISBN 978-1-349-01425-5 ISBN 978-1-349-01423-1 (eBook)


DOI 10.1007/978-1-349-01423-1

Phototypeset by Oliver Burridge Filmsetting Ltd,


Crawley, Sussex
TO SHRI ANANDAMURTI JI
Contents

PREFACE xi

I INTRODUCTION I
l.l The Production Function I
1.2 Technical Progress 6
1.3 The Transformation Curve II

2 STRUCTURE OF PRODUCTION IN AUTARKY 18


2.1 Introduction 18
2.2 Assumptions and the Model 18
2.3 Equations of Change 22
2.4 Commodity and Factor Prices 27
2.5 Factor Endowment and Commodity Prices 32
2.6 Technical Change and Commodity Prices 34
2.7 Other Properties of the Model 35
2.8 Autarky Equilibrium 40
2.9 The Community Indifference Curve 40
2.10 Summa.ry 44

3 THE BASIS OF INTERNATIONAL TRADE 46


3.1 Introduction 46
3.2 Conditions for the Absence of Trade 47
3.3 The Ricardian Theory 47
3.4 The Heckscher-Ohlin Theory 58
3.5 A Comparison 67
3.6 Equilibrium in International Trade 68

4 GAINS FROM TRADE 81


4.1 The Optimality of Free Trade 84
4.2 Higher Trade Intervention versus Lower Trade
Intervention 89
Vlll Contents

4.3 Trade Intervention versus No Trade 94


4.4 Tariffs, Subsidies and Taxes 97
4.5 The Terms of Trade and Welfare 99
4.6 Discriminatory Tariffs: The Theory of Customs
Union 101

5 THE THEORY OF NOMINAL TARIFFS 110


5.1 The Analytical Framework 110
5.2 Tariffs and the Terms of Trade 113
5.3 Tariffs and the Domestic-Price Ratio 118
5.4 Tariffs and the Distribution of Income 122
5.5 The Optimum Tariff 123
5.6 The Symmetry between the Import and Export
Taxes 126

6 ECONOMIC EXPANSION AND THE TERMS OF TRADE 129


6.1 Introduction 129
6.2 Economic Expansion 131
6.3 Immiserising Growth 136
6.4 Factor Accumulation 139
6.5 Technological Change 145

7 INTERMEDIATE PRODUCTS: THE INTER-INDUSTRY


FLOWS 154
7.1 The Model with Inter-Industry Flows 155
7.2 The Transformation Curve 158
7.3 Some Standard Trade Theorems 165
7.4 Technical Progress 167
7.5 The Gains from Trade 169
7.6 Appendix 174

8 PURE INTERMEDIATE PRODUCTS 180


8.1 The Model with Pure Intermediate Goods 181
8.2 The Structural Relations 185
8.3 Some Properties of the Model 186
8.4 The Standard Trade Theorems 187
8.5 Technical Progress in the Final Good 188
8.6 Technical Progress in the Intermediate Product 189
8.7 Trade in Intermediate Products 192
8.8 Appendix 199
Contents ix

9 THE THEORY OF EFFECTIVE PROTECTION 202


9.1 The E.R.P. Concept and the Partial Equilibrium
Approach 204
9.2 Negative Value-Added 208
9.3 Effective Protection in General Equilibrium 210
9.4 Effective Protection and Real Wages 215
9.5 The E.R.P. and Resource Allocation 218
9.6 Effective Protection and the Gains from Trade 231

I0 FACTOR MARKET IMPERFECTIONS 240


10.1 The Model with the Wage Differential 241
10.2 Factor Intensities and the Wage Differential 243
10.3 Factor and Commodity Prices 246
10.4 The Slope of the Transformation Curve 247
10.5 The Price-Output Response 249
10.6 The Shape of the Transformation Curve 250
10.7 The Heckscher-Ohlin Theorem 251
I0.8 The Implications of the Wage Differential 254
10.9 The Wage Differential and Welfare 257
10.10 The Wage Differential and the Terms of Trade 259
I0.11 The Wage Differential and the Gains from
Trade 263
10.12 The Wage Differential and Immiserising Growth 270
10.13 Other Types of Factor Market Distortions:
Factor Immobility and Factor-Price Rigidity 274
10.14 Concluding Remarks 279

11 PRODUCT MARKET IMPERFECTIONS: THE THEORY OF


MONOPOLY IN GENERAL EQUILIBRIUM 281
11.1 The Model with Pure Monopoly 282
11.2 The Consumer-Producer Equilibrium 284
11.3 The Price-Output Response and the Nature of
Autarky Equilibrium 290
11.4 The Standard Trade Theorems 293
11.5 Monopoly and the Gains from Trade 296
11.6 Concluding Remarks 301
11.7 Appendix 302

12 NoN-TRADED GooDs 305


12.1 The Model with Non-Traded Goods 305
X Contents

12.2 The Price-Output Response of Traded Goods 309


12.3 The Stolper-Samuelson Theorem 312
12.4 The Rybczynski Theorem 313
12.5 The Gains from Trade Once Again 315

13 INTERNATIONAL INVESTMENT 322


13.1 Capital Mobility as a Substitute for
International Trade 322
13.2 Optimal Tariffs and Taxes 325

14 INTERNATIONAL TRADE IN A DYNAMIC ECONOMY 334


14.1 Assumptions and the Model 334
14.2 The Uniqueness of Momentary Equilibrium 336
14.3 The Stability of the Long-Run Equilibrium 342

AUTHOR INDEX 351

SUBJECT INDEX 353


Preface

Writing a textbook, especially on as vast a subject as the theory of


international trade, requires a considerable amount of intellectual
investment and time. Perhaps for this reason alone, the author
should be able to provide justification not only for the prospective
reader and reviewer but also for himself. My interest in writing this
book, which is meant to be a textbook for graduate and advanced
undergraduate students interested in trade theory, stems from the
revolutionary strides which the theory of international trade has
achieved in the last decade. What we have inherited from the classical
economists and the neo-classicists have by now become standard and
indispensable tools of analysis. No book can hope to proceed with-
out these tools. However, given the unavoidable lag that charac-
terises the popularisation of most new theories, concepts and
developments, general books in trade theory have not as yet
accorded comprehensive and systematic treatment to topics such
as the implications of factor market imperfections, the theory of
imperfect product markets, the concept of effective protection, the
repercussions of the introduction of intermediate goods, etc., topics
that have by and large attracted the attention of the trade theorist in
the last ten years. A large portion of this book is allocated to the
development of these topics.
The level of analysis in general is quite simple. Although the
analysis gives the impression of being predominantly algebraic,
geometry has not been slighted. In fact a very liberal use of diagrams
is made to supplement the results derived from the mathematical
formulae the derivation of which requires only a knowledge of
simple differentiation. Throughout the book I have used the mathe-
matical tools developed and popularised by Jonest in his article in

t Jones, R. W., 'The Structure of Simple General Equilibrium Models', Journal of


Political Economy, LXXIII (Dec 1965) 557-72.
xu Studies in the Pure Theory of International Trade

the Journal of Political Economy ( 1965). I sincerely believe that the


use of Jones's approach, suggested by activity analysis, considerably
simplifies the mathematics and makes the results that much more
intelligible. A case in point is the dynamic model of international
trade developed in a voluminous article by Oniki and Uzawa.t Their
results are discussed in the last chapter of this book. Anyone already
familiar with their work and its recent related developments can
readily appreciate how simple the model becomes when we utilise
the Jones-type approach rather than the one used by Oniki and
Uzawa.
This book is not comprehensive, in the sense that the historical as
well as the empirical aspects of trade theory have been by and large
ignored, mainly because, I think, these subjects belong to another
genus of international economics and for that reason constitute the
subject-matter of another book.
In dealing with each topic I have deliberately tried to be precise
and to the point. Few words are wasted on huge introductions or the
history of a subject, except where I thought it was absolutely neces-
sary to do so. The stress all along has been to provide the reader with
a fresh look at most of the standard topics and to examine in detail
some of the more recent developments.
Most of this book was written while I was at the University of
Western Ontario from 1970 to 1972. I wish to express my special
gratitude to my friend and former student, Francisco R. Casas, for
very helpful discussions, suggestions and extensive reading of the
earlier drafts of this book. Some of the material appearing here is a
by-product of our joint articles. Thanks are also due to J. R. Melvin,
V. S. Rao, J. C. Leith, P. C. Ip and C. Y. Kuo, who during my stay
at Western Ontario contributed much to the clarity of my own
thinking. I should also like to express my gratitude to Bun Song Lee
for careful reading of the penultimate draft. Chapter 6 of this book
is a condensed version of my dissertation submitted to Southern
Illinois University in 1969. There I received very helpful guidance
and criticisms from my adviser, AlbertY. Badre, and from Charles
Stalon and Thomas Martinsek.
Finally, I should like to acknowledge my great debt to Jagdish
Bhagwati, who as my teacher at the Delhi School of Economics in

t Oniki, H., and Uzawa, H., 'Patterns of Trade and Development in a Dynamic
Model of International Trade', Review of Economic Studies, xxxn (Jan 1965) 15-38.
Preface xiii

1965 greatly stimulated my interest in the pure theory of inter-


national trade. Last but not least I wish to express my gratitude to my
wife Diane, who for several months had to live through the tedium
of drawing my diagrams.

July 1972 RA VEENDRA N. BATRA


1 Introduction

Although the pure theory ofinternational trade has much in common


with the micro-economic theory, for obvious reasons trade theory
should be and has been treated as a separate discipline. Nevertheless,
a clear comprehension of the micro-economic concepts is essential
for the exploration of various issues that arise in the field of inter-
national trade. Accordingly, the present chapter is devoted to a brief
examination of mathematical concepts, definitions and geometric
tools which have been utilised quite extensively in the literature and
which as an aid to exposition will be used on numerous occasions in
subsequent chapters.

1.1 The Production Function


At the heart of all studies in the theory of international trade lies the
concept of the production function which attempts to describe a
certain technical process, as of a given technology, through which
productive inputs like capital and labour are transformed into
output. A number of restrictions are usually imposed on the function
that provides this information.
To begin with, we assume that the production process can be
represented by a sufficiently smooth function
X= F(K,L) (1.1)
which shows that the output flow X is obtained from given amounts
of homogeneous factors K (capital) and L (labour). This production
function is assumed to possess the following properties:
(i) Both factors are indispensable in the sense that
F(O,L) = F(K,O) = 0.
(ii) F(K,L) is homogeneous of degree one, so that
).X = F(,l.K,,l.L) (,1. > 0)
2 Studies in the Pure Theory of International Trade

that is, if both inputs are changed in some proportion A., the
output also changes in the same proportion.
(iii) The first partial derivatives of F(K,L), denoted by FK and FL,
are positive, but the second derivatives, FKK and FLL• are
negative. This implies that all marginal productivities are
positive but diminishing:
F. _ oF(K,L) O F. _ oF(K,L) O
K- oK > ' L- oL > '
82 F(K,L) 82 F(K,L)
FKK = oK2 < 0, and FLL = oL2 < 0.

A production function which satisfies these properties is usually


called a neo-classical production function.
The second property- of the production function enables us to
write it in an intensive form. Suppose A. = 1/L; then (1.1) can be
written as
X= LF(~, t) = Lf(k) (1.2)

where k equals the capital/labour ratio. When the production func-


tion is presented in this form, it can be shown that all marginal
products depend on the capital/labour ratio only; that is,
ax o(Lf) ok
FK = - = - - = Lh.- = fk
oK oK oK
ax o(Lf) ok
FL = oL = ~ =f+Lfk aL

= f-kfk ( f =f(k) and fk =----ak


of(k))
. (1.3)

Evidently, both FK and FL depend solely on k. Furthermore,


F. - oh. - r iJk - fkk
KK - aK - J kk oK - L

and
au- kh.) ok ok ok k2
FLL = oL = fk oL -kfkk oL -h. oL = L fkk·
The negative signs of FKK and FLL imply that Ak =
o.t,;/ok < 0,
which, in turn, implies that the marginal product of capital declines
Introduction 3

(rises) as a result of a rise (decline) in the capital/labour ratio.


Simultaneously, the rise (decline) in k leads to a rise (decline) in the
marginal product of labour. This can be appreciated by differen-
tiating FL with respect to k to get
dFL
dk = -kfu > 0.

The magnitude of the change in the two marginal products is deter-


mined by the elasticity of substitution between capital and labour,
which is defined as
dk w
(1=-·- (1.4)
dw k
where a is the elasticity of factor substitution and w the ratio between
the marginal productivities of labour and capital:
FL f-kh.
w=-=--
FK fk
so that
dw fh.k
dk - !l
and hence
-fk(f-k.f,)
(1 = --=--'7"-:-:--_.::_:_"-
kf.f,k
which, in view offu < 0, is positive. Ifwe assume that the entrepre-
neur seeks to minimise the unit cost of production in order to maxi-
mise his profits, then for a given set offactor prices, w, the wage rate,
and r, the rental rate of capital, prevailing in perfect factor markets, t
t LetT= rK + wL be the total cost of production, and X = F(K,L) the fixed level
of output. The problem is to minimise T subject to the constraint imposed by the
production function. For this purpose, form a Lagrangian
H = (rK+wL)+1XdX-F(K,L ))

where IX 1 is the Lagrangian multiplier. The first-order conditions for the minimisation
of T for given w and r are then given by
aH aH
oK = r-IX,FK = oL = W-IX 1 FL = 0.

Solution of these conditions yields

Note that this condition, associated with the minimisation of the unit cost, must be
valid so long as there is perfect competition in factor markets, irrespective of con-
ditions in the product markets.
4 Studies in the Pure Theory of International Trade

FL w
w = - = -. (1.5)
Fx r
Thus, so long as there is perfect competition in factor markets, the
ratio of the two marginal products equals the wage/rental ratio.
From (1.3) follows another well-known result: namely, the total
product is fully exhausted by the contribution by the two factors:
LFL + KFx = L(f- kfk) + Kfk = X.
There is yet another way in which the production function can be
expressed. With the marginal products of both factors determined
exclusively by the factor proportions, the entire information con-
cerning the production surface in the two-dimensional capital and
labour space can be summed up in the unit isoquant, representing
oneunitoftheoutpu t. Supposd = 1/X. Then(l.l)can be written as
1 = F(Cx,Cd
where Cx = K/X and CL = L/X are the input-output coefficients.
From this and (1.5) the definition of the elasticity of substitution
given in (1.4) becomes

(J
K.*-L*
= ------,--
(K.*-X*)-(L*-X* ) q-q
w*-r* w*-r* w*-r*

where the asterisk indicates the proportionate change, e.g. K.* =


dK/K.
If, in addition to facing a competitive factor market, the entre-
preneur produces his output under conditions of perfect competition,
then the commodity price faced by him is fixed. Facing a given price
and the neo-classical production function such as ( 1.1 ), the producer
is assumed to select that combination of factor inputs which maxi-
mises his profits (n):
n = pF(K,L)-wL-rK. (1.6)
This maximisation is accomplished by setting the derivatives of(l.6)
to zero; that is,
an an
= pFL-w = - = pFK-r = 0
-
aL aK ·
Evidently, the competitive producer employs the two factors up to
the point where the value of their marginal products equals their
Introduction 5

given prices:

and (1.7)

Further comprehension of the rudimentary production analysis


presented above may be gained from Fig. 1.1, where X1 is the unit
isoquant, which is a locus of the various combinations of K and L

L
Figure 1.1

necessary to produce one unit of X. Different points on X1 represent


different levels of the capital/labour ratio and hence different
marginal productivities for the two factors of production. For
example, at point E the capitaljlabour ratio equals the slope of the
ray 0 E, the marginal product of capital associated with this capital/
labour ratio equals 1/0A, and the marginal product of labour equals
I jOB, where points A and Bare obtained by drawing AB tangential
to the unit isoquant at E. The demonstration of these results proceeds
as follows. From the product exhaustion theorem stated earlier,
OD. FL +DE. FK = xl = l. Moreover, the slope of xl at E is
given by
0A DE I:!K I:!Kj !:!X
-=-=-=
OB DB I:!L I:!Lji:!X
6 Studies in the Pure Theory of International Trade

Substituting, we get
1
FL(OD+DB) = 1, or FL = OB'
Similarly, it can be shown that FK = 1/0A.
Now suppose that AB also represents the isocost line, so that the
slope of AB equals the wage/rental ratio. PointE then furnishes the
capital/labour ratio (k = DE/OD), the capital/output ratio (CK =
DE) and the labour/output ratio (CL = OD), which are uniquely
determined, provided the entrepreneur seeks to minimise his unit
cost. Thus for any given wjr ratio, and a given technology repre-
sented by the neo-classical production function, the equilibrium
values of k, CK and CLare uniquely determined. Furthermore, with
production taking place under conditions of perfect competition,
the commodity price from (1.7) can be expressed in terms of the
cost of capital or labour alone. That is, p = wjFL = r/FK, or, in
termsofFig.l.l,p = w.OB = r.OA.

1.2 Technical Progress


Until now we have assumed that the transformation of productive
inputs into output expressed by the production function occurs at a
given level of technology. What happens if the level of technology
itself changes over time? The production function in the general form
may then be written as
X= F(K,L,t) (1.8)
where t denotes the state of technology. The function described by
( 1.8) is assumed to possess the same properties as those characterising
(1.1) for a given state of technical know-how. In addition, F, > 0.
We have already seen that, with the neo-classical production func-
tion, the marginal factor productivities depend solely on the factor
proportions. Since technical progress usually results in a rise in the
marginal productivity of one or both factors, the latter is also
affected by t. In other words,
FL
w = FK = w(k,t). (1.9)

Total differentiation of (1.9) yields


I
w* = Ft-fi. = -k*+a
(J
(1.10)
where a = (Ijw)(owjot) dt exhibits the incidence of technical pro-
gress on the two marginal productivities; the partial derivative
notation used in obtaining a shows that k is kept constant.
Introduction 7

Equation ( 1.1 0) enables us to explain what is widely familiar as the


Hicks [I] measure of technical progress. According to Hicks,
technical progress is neutral if, at the pre-technical change capital/
labour ratio, the marginal productivity of both factors rises in the
same proportion; technical progress is labour-using (or capital-
saving) if the proportionate rise in the marginal productivity of
labour exceeds the corresponding rise in the marginal productivity
of capital; it is capital-using (or labour-saving) if the greater pro-
portionate rise in the marginal productivity occurs for capital rather
than labour. Since the capital/labour ratio has been kept constant in
obtaining tX, the sign of tX determines the nature of the Hicksian
technical improvement. Specifically, for the neutral technical change,
tX = 0; for the labour-using type of technical improvement, tX > 0;
and under the capital-using type, tX < 0. Although technical pro-
gress normally leads to a rise in the marginal productivity of both
factors, it is at least conceivable that the factor-using type of tech-
nical improvement could be compatible with the absolute decline
in the marginal productivity of one factor.
In the literature on economic theory, the Hicks technical progress
has been defined in still another way. Although Hicks originally
emphasised the productivity-raising role of technical innovations,
his definitions imply other ways of stating the nature of improve-
ments where this role is not so explicit. Under the assumed condition
of cost minimisation, the equilibrium capital/labour ratio has already
been shown to be determined solely by the wage/rental ratio which
in turn equals the ratio between the marginal productivity of labour
and capital. It follows, therefore, that there exists a dual relationship
between the marginal factor productivities and the capital/labour
ratio; the latter affects the former because of our assumption of
constant returns to scale but diminishing returns to factor propor-
tions, whereas the former influences the latter because of our
assumption of profit maximisation, or alternatively cost minimisa-
tion, on the part of the producer. When technical progress is intro-
duced, the capital/labour ratio is also affected by the nature of the
improvement. Thus the dual of (1.9) is written as
k = k(w,t) (l.ll)
so that
k* = aw*+P (l.l2)
where p= (l/k)(okfot) dt represents the effect of technical change
8 Studies in the Pure Theory of International Trade

on the capital/labour ratio; the partial derivation notation used in


writing p implies that the factor-price ratio is kept constant; and it
is the sign of P which crucially determines the nature of technical
inventions. With the wage/rental ratio kept constant, it is not difficult
to see that p = 0 implies Hicks-neutral improvement, whereas
p ~ 0 represent capital-using and labour-using improvements,
respectively.
The Hicks definitions of technical inventions can also be illus-
trated in terms of simple diagrams. Consider Fig. 1.2, where, as
before, X 1 is the unit isoquant and AB represents the unit cost
associated with the production of one unit of X. Since normally
the effect of technical progress is to raise the marginal productivity
of one or both factors, the unit isoquant X 1 shifts towards the origin
to X~, thereby showing that smaller quantities of capital and/or
labour are now required to produce one unit of X. In order to
determine the nature of the improvement, the change in the
marginal productivity of the two factors must be evalutated at the
original capital/labour ratio given by the slope of OE or OE',
where E' lies on the rayO£. Draw GH tangential to X~ at£'. If GH
is parallel to AB, then technical progress must be Hicks-neutral.
For the proportionate rise in the marginal productivity of capital
equals
(1/0G)-(1/0A) AG
-~Fx = ----,----,----
FK (1/0A) OG

whereas the proportionate rise in the marginal productivity of


labour is given by HB/OH, so that with the Hicks-neutral improve-
ment requiring (AG/OG) = (HB/OH), GH must be parallel to AB
to ensure this equality. Notice further that GH parallel to AB
implies that the wage/rental ratio has remained constant at the old
capital/labour ratio. If technical progress is labour-using, then GH
will be steeper than AB to show that the proportionate rise in the
marginal productivity of labour exceeds the corresponding rise in
the marginal productivity of capital at the original capital/labour
ratio. This case is depicted in Fig. 1.3, where HB/OH > AG/OG.
The alternative way of defining the labour-using (or capital-saving)
improvement is also clear; at the same wage/rental ratio given by
the slope of MN (parallel to AB), we find that the capital/labour
ratio has declined to that represented by the slope of OQ.
Introduction 9

Figure 1.2

Figure 1.3
10 Studies in the Pure Theory of International Trade

Figure 1.4

The case of capital-using (or labour-saving) technical progress is


depicted in Fig. 1.4. Here HB/OH < AG/OG, so that at the same
capital/labour ratio the marginal productivity of capital has risen
in a greater proportion than that of labour, or equivalently, at the
same wage/rental ratio reflected by the slope of M N, the capital/
labour ratio has risen to the slope of OR.
The.stage has now been set for defining the character of technical
progress in terms of input-output coefficients. Under neutral tech-
nical progress, the capital/output (CK) and the labour/output (Cd
coefficients decline in the same proportion at the old wage/rental
ratio. For instance, in Fig. 1.2, CK has declined from OCto OC'
and CL from OD to OD' as a result of the technical improvement,
and it can be easily seen that, with GH parallel to AB, ( CC'/OC) =
(DD'/OD). With capital-saving (or labour-using) improvement, CK
declines in a greater proportion than CL at the old wage/rental ratio,
as is the case in Fig. 1.3, where (CC'/OC) > (DD'/OD). The
opposite holds for the labour-saving (or capital-using) improve-
ment, as in Fig. 1.4, where (CC'/OC) < (DD'/OD).t
From this discussion and that in the previous section we conclude
that each input-output coefficient depends on the wage/rental ratio

t With non-neutral improvements, it is, of course, possible that one of the input-
output coefficients may rise at the original wage/rental ratio. Figs. 1.3 and 1.4 can be
modified to portray this case.
Introduction 11

as well as on the character of technical improvements; that is,


C; = C;(w,t). ( 1.13)
This relationship will be extensively used in the an lysis of succeeding
chapters.

1.3 The Transformation Curve


One geometric tool of analysis, the use of which in trade theory has
probably surpassed the use of every other diagrammatic device, is
well known by the name of the production possibility curve or
simply the transformation curve. In an economy where all factors
of production are always fully employed in the production of two
commodities, commodity I and commodity 2, the transformation
curve simply represents the various combinations of the two out-
puts or, more specifically, the maximum possible output of one
commodity, given the output of the other. In other words, the
transformation curve is a locus of efficient production points,
production efficiency being defined in the sense of what is well
known as the Pareto optimality criterion, namely, that a combina-
tion of two outputs is efficient if every other feasible reallocation of
inputs diminishes the output of at least one commodity.
The efficiency criterion is readily established geometrically in
terms of the Edgeworth- Bowley box diagram for the case where the
two commodities, with outputs designated as X1 and X 2 , are pro-
duced with the aid of two factors of production, capital (K) and
labour (L). The dimensions of the box in Fig. 1.5 represent the total,
inelastically supplied quantities of capital and labour, so that any
point in the box reflects a certain allocation of inputs between the
two commodities, whose outputs are measured by reference to the
two origins 0 1 and 0 2 . If we assume that production functions in
both commodities are neo-classical, then the level of two outputs
can be measured by the distances of their isoquants from the re-
spective, origins. Here the diagonal 0 1 0 2 , whose slope with respect
to the 0 1 0 axis equals the overall capital/labour ratio in the
economy, plays a useful role. Since each point on an isoquant
represents the same level of output, the output of each commodity
can be measured by the distance between the respective origin and
the point of intersection between its isoquant and the diagonal.
For example, the output of the first commodity, X1 , represented
by isoquants x 1 and x'1 , is given respectively by 0 1 a 1 and 0 1 a2 •
Similarly, the output of the second commodity, X 2 , with isoquants
12 Studies in the Pure Theory of International Trade

•'
t
>(
:! 0
·a .;a.
"
<.J ;
0

- - Output of X 2
Figure 1.5

x2 and x2, equals 0 2 b 1 and 0 2 b 2 • Clearly, then, a movement away


from 0 1 towards 0 2 is equivalent to a rise in the output of the first
commodity and a decline in the output of the second commodity.
Conversely, a movement away from 0 2 towards 0 1 signifies an
increase in the output of the second commodity at the expense of the
output of the first commodity.
Now according to the Pareto optimality criterion, maximum
efficiency in the allocation of resources is attained at points such as
A and C, where the isoquants of the two commodities are tangential
to each other. The locus of all such points, the curve 0 1 AC0 2 , is
called the contract curve, or simply the efficiency locus. All points
on the contract curve represent the maximum attainable output of
one commodity, given the output of the other. For example, suppose
that the output of X 2 equals 0 2 b 1 . Then the maximum possible
Introduction 13

output of X 1 , equal to 0 1a 2 , is obtained at C. At any other point


the output of X 1 will be less than that at C. At point B, for instance,
X 2 still equals 0 2 b 1 , but X 1 , equal to 0 1 a 1 , is less than 0 1 a 2 .t
Hence B is inferior to C.
The location of the contract curve in the box reflects a certain
assumption concerning the capital/labour ratio in two commodities.
The contract curve in the diagram lies below the diagonal because
of our implicit assumption that under all circumstances X 1 is labour-
intensive relative to X 2 . For example, at point A the capital/labour
ratio in X 1 , equai to the slope of 0 1 A with respect to the 0 1 0 axis,
is less than the capital/labour ratio in X 2 , which is given by the slope
of 0 2 A with respect to the 0 2 0' axis. If, instead of X 2 , X 1 was the
capital-intensive commodity, then the contract curve would lie
entirely above the diagonal.
We are now in a position to derive the transformation curve from
the contract curve. It has already been established that various
points on the diagonal representI different levels of output of each
commodity. This measuring scale can be easily transformed to the
vertical and horizontal scale represented by the origin 0. For
instance, the output levels of X 1 , given by points a 1 and a 2 on the
diagonal, can be projected towards the 00 2 axis to points a and a',
respectively. Since the production functions are homogeneous of the
first degree, the distance 0 1 a 2 exceeds 0 1 a 1 in exactly the same pro-
portion as Oa' exceeds Oa. Analogously, the output levels of X 2 ,
given by points b 1 and b 2 , can be projected to obtain the respective
points band b' on the horizontal axis 00 1 • In this fashion, following
the technique developed by Savosnick [2], the 00 2 and 00 1 axes
can be used as output scales.
The next step is to derive points in the commodity space, repre-
senting the levels of each output along the contract curve. The output
of X 1 and X 2 corresponding to point A, for instance, is given respec-
tively by Oa and Ob'. The output combination given by points a and
b' is then furnished by point P in the commodity space. In other
words, the output combination associated with A, when transformed
into output scale, is given by P. Similarly, the output combination at
C is represented by Q in the output space. Thus we see that to each
point on the contract curve corresponds a unique point in the
tHere and elsewhere, X1 is identified with thejth commodity as well as its output.
Furthermore, the words 'commodity', 'industry' and 'sector' will be used synony-
mously.
14 Studies in the Pure Theory of International Trade

commodity space. The locus of points such asP and Q is called the
transformation curve, which in Fig. 1.5 is given by 0 1 PQ02 • The
transformation curve is thus a mirror-image of the contract curve.
One may observe that the curve 0 1 PQ0 2 is wholly concave towards
the origin, 0.
What is the slope of the transformation curve in equilibrium?
To answer this question, we need information about the type of
markets in which production takes place. The transformation curve
is the locus of efficient points and its curvature depends on market
conditions in which the firms operate. The answer turns out to be
very simple and elegant, if we assume that all markets are perfect.
In the presence of market imperfections, of course, it is possible
that the transformation curve, as a locus of efficiency points, may
cease to exist, simply because production efficiency cannot be
attained under such imperfections. For the time being, however, we
gloss over this question, but we shall examine it in detail in the
chapters concerning market imperfections.
Under the assumption of perfect competition in all markets, the
reward of each factor in equilibrium equals the value of its marginal
product and is the same in both commodities. Let M Pu be the
marginal productivity of the ith factor in the jth commodity and Pi
be the commodity price (i = K,L;j = 1,2). Then
w = P1MPL1 = P2MPL2
and

From these relations,


P1 MPL2 MPK2
P2 MPLI MPK1
or
P1 !!X2jl!L 2 !!X2/l!K 2
P2 !!Xd!!L 1 !!Xd!!K 1.
Given our assumption that factors are fully employed and inelasti-
cally supplied, !!L 2 = - !!L 1, and !!K2 = - !!K1 • Substituting, we
get
!!X2 Pt
!!Xt P2
Now !!X2j!!X1 furnishes the slope of the transformation curve,
and is usually called the marginal rate of transformation. In other
Introduction 15

words, in competitive equilibrium the slope of the transformation


curve reflects the negative of the commodity-price ratio.
This equilibrium condition is diagrammatically depicted in Fig.
1.6, where TT' is the transformation curve drawn concave to the

x,
Figure 1.6

origin. Lines such as RS and M N are the budget lines which measure
the level of national income in the economy at different production
points. The slope of each budget line reflects the relative price of X 1 ,
ptfPJ. The equilibrium production point is obtained by drawing the
given budget line tangential to the transformation curve. Thus, with
RS, the production point is given by P, where the slope of TT' equals
the relative price of X 1 • The concavity of TT' towards the origin
implies that a rise in the output of a commodity calls forth a rise in
its relative price. This point is illustrated by the fact that the pro-
duction point associated with MN, showing a higher relative price
of X1 , is Q, which shows that the output of X 1 rises from Oa to Oa'
and that of X 2 declines from Ob' to Ob.
What is the reasoning behind the concavity of the transformation
curve, or, stated differently, what is the economic explanation behind
the phenomenon of greater output requiring a greater relative price,
16 Studies in the Pure Theory of International Trade

even though returns to scale have been assumed to be constant?


The reason lies in our assumption, implicit in the construction of the
efficiency locus, that the capital/labour ratio in each commodity is
different under all circumstances. Because of unequal capital/labour
ratios in the two commodities, it can be shown that the output level
is positively related to the unit cost, which, under perfect competi-
tion, equals price. For instance, consider the movement from Q to
P along the transformation curve TT'. At P, the output of X1 is
lower than at Q, so that a fraction of capital and labour employed
in xl is released in a certain proportion, only to be absorbed by x2.
Since X2 employs more units of capital per unit of labour than X1 ,
the quantity of labour released from xl is too great to be fully
employed in x2' whereas the quantity of capital released by xl is
insufficient for the req'uirements of X 2 . The result is that at the
existing capital/labour ratios in the two commodities there is excess
supply of labour and excess demand for capital. This disequilibrium
situation, which is inconsistent with the full-employment equili-
brium, can be corrected only by a rise in the price of capital along
with a decline in the price oflabour. However, since X 2 is the capital-

Figure 1.7
Introduction 17

intensive commodity, the rise in the price of its intensive factor


boosts up its unit cost, whereas for X1 the decline in the price of
labour, its intensive factor, lowers the unit cost. Finally, the full-
employment equilibrium is restored through unidirectional changes
in the two capital/labour ratios consequent upon the change in
factor prices. However, the commodity whose output has risen is
the one where the unit cost and hence the price has increased, and
conversely.
From this discussion, it follows that if there exists only one factor
of production, or if the capital/labour ratio in each commodity is
the same, the transformation curve will be a straight line, as shown
in Fig. 1. 7. The slope of TT' would still reflect the relative price of
X1 , but the point of production could lie anywhere on TT' for a
given commodity-price ratio, because the unit cost in any commo-
dity would remain constant for all levels of output.

REFERENCES
[I] Hicks, J. R., The Theory of Wages (London: Macmillan, 1932).
[2] Savosnick, K. M., 'The Box Diagram and the Production-Possibility Curve',
Ekonomisk Tidskrift, LX (Sep 1958) 183-97.
2 Structure of Production
in Autarky

2.1 Introduction
One of the elemental, though most important, questions in inter-
national trade concerns the determination of a country's pattern of
trade. Stated differently, why are countries induced to import goods
if they can produce them at home? The answer to these questions,
however, requires a prior comprehension of the structure of produc-
tion in a closed economy or autarky. It is this latter question with
which the present chapter is concerned. Specifically, we wish to deter-
mine the factors which influence commodity prices in the absence
of trade.

2.2 Assumptions and the Model


Unless otherwise specified, the following assumptions will be main-
tained in every chapter:
I. There are two commodities, X1 and X 2 , which in the process of
production require two primary factors of production, capital
(K) and labour (L).
2. Returns to scale are constant, but returns to factor proportions
are diminishing.
3. Factors of production are fully employed but inelastically
supplied.
4. Factors are perfectly mobile between sectors, and factor prices
perfectly flexible. The latter assumption, of course, ensures the
maintenance of full employment at all times.
5. Each commodity utilises a different capital/labour ratio.
Furthermore, the inter-commodity capital-intensity relation is
non-reversible, or, stated more explicitly, a commodity is inten-
sive in the use of the same factor at all factor prices.
6. There is perfect competition in product as well as factor markets,
and factor prices are the same in both industries.
Structure of Production in Autarky 19

The development of the model proceeds with the specification of


the full-employment equations. Let Ki be the amount of capital and
Lithe amount of labour employed in thejth commodity. Then with
full employment,

and
K 1 +K2 = K.
These equations can be transformed in a manner which, by means of
the concept of input-output coefficients, will determine the output
level of each commodity. Let Cii be the quantity of the ith factor used
in the production of one unit of the jth commodity. For example,
eLl = LtfXl, and CKl = KtfXl, so that Ll = CLlxl and Kl =
CK 1 X 1 , and so on. The full-employment equations may then be
written as

cLlxl + cL2x2 = L (2.1)

cKlxl +CK2X2 = K. (2.2)

The input-output coefficients can also be used to specify equations


for the price of each commodity. Under competitive conditions,
each commodity price equals the unit cost, which in turn is given by
the ratio between total cost and total output. With the first com-
modity, for example,
wL 1 +rK 1
xl
where, as before, Pi is the price of the jth commodity, w the wage
rate and r the rental of capital. Each commodity price may then be
described as

(2.3)

(2.4)

Next, the input-output coefficients must be determined. As


established in the previous chapter, each input-output coefficient
depends on the wage/rental ratio as well as the character of technical
improvements; that is to say,
20 Studies in the Pure Theory of International Trade

(i = L,K;j = 1,2). (2.5)

Equation (2.5) contains four independent relations for the four


input-output coefficients. Thus there are eight independent relations,
four described by (2.1)-(2.4) and four by (2.5), containing eight
variables, X 1 , X2 , w, rand four Cii's, in five parameters, L, K, p 1 ,
p 2 and t. In other words, the production side of the model is deter-
minate. To close the system, equations expressing demand for each
commodity must be introduced. Let Di stand for the demand for
thejth commodity, p for the commodity-price ratio, p 2 /p 1 , and Y
for national income. Then under autarky

(2.6)
and
(2.7)

Equations (2.6) and (2. 7) state that the demand for each com-
modity depends on the commodity-price ratio and national income,
which in turn may be written as

(2.8)

that is, national income simply equals the total value of the two
outputs or the payments made to the factors of production. With
this equation, the specification of our two-commodity, two-factor
autarky model is complete. We have now added five equations in
four more variables, D 1 , D 2 , p and Y, but from Walras's Law only
one of the two equations, Di = Xi, is independent, so that the system
continues to be determinate. One may have noticed the differential
treatment accorded to p on the demand and the production sides
of the model. On the demand side,p is treated as a variable. However,
on the production side p is treated as a parameter, because the
competitive firm takes the commodity price as given.
The significance of the assumption of differing capital/labour
ratios in the two commodities for the unique determination of out-
puts may now be shown. For a given t = t, the economy's state of
technology is completely defined by the input-output coefficients,
which can be collectively expressed by a technology matrix [CJ,
that is,
Structure of Production in Autarky 21

The columns of the C matrix describe the state of technology behind


the production function in each commodity, because it is the matrix
of the production coefficients associated with X1 and X 2 in equations
(2.1) and (2.2). The determinant of [ C] is given by

where ki is the capitaljlabour ratio in thejth commodity. Equations


(2.1) and (2.2) may be solved with the aid of Cramer's Rule to yield
L(k 2 -k)
(2.1 0)
CL1(k 2 -k 1 )
and

(2.11)

where k = K/ L.
As far as the price equations (2.3) and (2.4) are concerned, the
matrix of the input-output coefficients associated with w and r is
defined by the rows of the C matrix, or simply by its transpose.
However, the sign of the matrix is still given by the sign of ICJ. The
solution of (2.3) and (2.4) gives us

(2.12)

One glance at (2.1 0) and (2.11) suggests that if the capital/labour


ratio is the same in both commodities, that is, if k 2 = k 1 , the two
outputs are indeterminate. As stated in the previous chapter, the
transformation curve in this case becomes a negatively inclined
straight line, and any point on the line could be a production point
at a given commodity-price ratio. Another important point is that
the production of both commodities, or what may be properly called
incomplete specialisation, requires that k lie between k 2 and k 1 • This
is because if the capital/labour ratio in any commodity is equal to the
overall capital/labour ratio, the output of the other commodity
22 Studies in the Pure Theory of International Trade

reduces to zero. For example, in (2.10) if k 2 , the capital/labour ratio


in the second commodity, equals k, the overall capital/labour ratio,
the output of X 1 , the first commodity, is zero. Similarly, if k 1 = k
in (2.11), the output of X 2 equals zero.
The factors affecting the wage/rental ratio are described in (2.12),
which shows that w is determined by the production coefficients and
the commodity-price ratio, p. However, since the production co-
efficients themselves are determined by w, the latter is influenced
only by p. Thus there exists a one-to-one relation between p and w.
Here again the condition of incomplete specialisation is relevant.
For in the presence of complete specialisation w is no longer related
top simply because p is then undefined. Consider, for instance, the
case where k 1 = k, so that only X 1 is produced in the economy.
Since X 2 reduces to zero, p is undefined. The wage/rental ratio is
then determined simply by the overall capital/labour ratio via the
elasticity of substitution in the first industry, u 1 . In other words, w
is no longer related to p,in the absence of incomplete specialisation.

2.3 Equations of Change


In order to understand the internal working of the model, it is neces-
sary to convert the system of equations presented above into
equations of change. In other words, we wish to examine the
comparative-statics properties of the model by considering the effects
of a change in the parameters on the variables of the system. For
this will enable us to see whether or not, and under what conditions,
all the variables are uniquely determined.
Let an asterisk indicate the rate of change. Thus L* = dLjL.
Differentiating equations (2.1)-(2.4) totally, we obtain

A.Llxt +A.L2X! = L*- [A.Llql +A.L2q2] (2.13)


A.K1Xi+A.K2X! = K*-[A.Klql +A.K2q2] (2.14)
0Llw*+0Klr* = pj-[OL1C!1 +0Klql] (2.15)
eL2w*+OK2r* = P!-[OL2C!2+0K2ck2] (2.16)

where A.ii is the proportion of the ith factor employed in the jth
commodity and eii is the share of the ith factor in the total earnings
in the jth commodity (i = L,K; j = 1,2). For example, A.Ll =
(CL 1XdL) = LdL equals the proportion of labour employed in the
first commodity, whereas eLl = (wCLl/p 1) = wLifp 1X 1 equals the
Structure of Production in Autarky 23

relative share of labour in the total value of output of the first com-
modity. By very definition, A.il +A.i 2 = I and (}Li+(}Ki = I. Let [A.]
and [ (}] respectively denote the matrices of A. coefficients and (} co-
efficients in equations (2.13)-(2.16); that is,

[A_] =[ALl
AK!
AL2],
AK2
[(}] =[(}Ll
(}L2
(}K!J.
(}K2
The signs of the determinants lA. I and 1e1 are the same as the sign of
the determinant Iq, because
,
IA. I = A.LIAK2-A.K!A.L2 L 1L 2
= LK (k2-kl)

wrL 1 L 2
1e1 = eLleK2-eL2eK! = x x (k2-kl). (2.17)
P1 1P2 2

Furthermore, since each row in IA.I and 1e1 adds to unity, the deter-
minants lA. I and 181 are also given by
IA.I = A.Ll-A.Kl = A.K2-A.L2
I(}I = (}Ll- (}L2 = (}K2- (}Kl (2.17*)
Clearly, the next step in the solution of the system is to obtain
expressions for Cij. Equation (2.5) can be totally differentiated to
obtain
(2.18)
where Aij = (l/Cii)(oCufow) dw is the change in the input-output
coefficient that occurs as a result of a change in the wage/rental
ratio (the partial derivative notation in writing Aij signifies that
technology remains unchanged) and Bij = (- 1/Cii)(oCii/ot) dt is
a measure of technical progress that results in a change in Cii; the
partial derivative notation in Bij shows that the wage/rental ratio is
kept constant. Since technical improvements ordinarily involve a
reduction in input-output coefficients, Bij is defined to be non-
negative. Simple expressions for Aij can be derived by introducing
the definition of the elasticity of factor substitution and taking into
account the implications of the unit cost minimisation. Let ai be the
elasticity of factor substitution in the jth sector. As defined in the
previous chapter,

(2.19)
24 Studies in the Pure Theory of International Trade

w 8k2 AI2- A!2


(2.20)
(j2=--=
k2 aw w*-r*
.

The unit cost in the first commodity equals wCLl + rCKl· For
given factor prices in competitive factor markets, the unit cost in
the first commodity is minimised by setting the first derivative
(wCL 1A! 1 + rCK 1At 1) equal to zero.t Dividing through by Pt> we get

(2.21)

An analogous procedure can be followed for cost minimisation in


the second commodity to derive
(2.22)

Equations (2.21) and (2.22) can be solved in combination with (2.19)


and (2.20) to furnish the effect of a change in the wage/rental ratio
on the input-output coefficients. For example, from (2.19) and
(2.21), and (2.20) and (2.22), we get

Ali = - (}Kirriw*- r*) () = 1,2)


Ati = 8Lp/w* -r*) (J = 1,2).
These relations for the Ai'j's can then be substituted first into (2.18),
and then the resultant solutions for Ci'j can be fed into equations
(2.13)-(2.16) to obtain

AL1X!+A.L2X! = L*+llL+fh(w*-r*) (2.23)


AK1X!+A.K2X! = K*+llK-f1K(w*-r*) (2.24)
(}Llw* + eKlr* = P! + nl (2.25)
OL2w*+OK 2r* = p!+ll 2 (2.26)
where
PL = A.LleKlal +A.L2eK2a2
PK = A.KlOLlrrl +A.K20L2a2
lli = A.ilB{1 +A.i2B{2 (i = L,K)

t Since at present we are trying to assess the effect of a change in won the input-
output coefficients as of a given technology, the notation used for dC,;/C,; is A;j
instead of C;j. which actually denotes the rate of total change in c,i.
Structure of Production in Autarky 25

and
llj = (}LjBfj + (}KjB:j (J = l ,2).
Before we proceed further with the analysis, a few explanatory
remarks concerning /3;, ll; and nj are in order. If production co-
efficients are fixed, crj = 0, in which case /3; also equals zero. In the
general case of variable coefficients, however, /3; reflects the percent-
age change in the use of the ith factor per unit of output that occurs
in both commodities as a result of a change in the wage/rental ratio
alone. By contrast, ll; represents the percentage reduction in the ith
factor that occurs owing to the occurrence of technical progress in
both commodities when the wage/rental ratio is kept constant; for
example,

Consequent upon technical improvements, the percentage reduction


in the use of labour per unit of output in the first commodity equals
Bt 1 , but AL 1 Bt 1 gives the total saving in the use of labour, made
possible by the improvement, to produce the original output
of X 1 • Similarly, ).L 2 Bl2 represents the saving in the use of labour
required to produce the pre-technical change level of X 2 • When
such savings in the use of labour in each commodity are added
up, we get nL, which represents the overall labour-saving character
of technical progress. Similar explication also applies to nK 0
On the other hand, nj is a measure of the rate of technical advance
in the jth commodity; for instance,

The incidence of technical improvement is to lower the unit cost of


production; 8L 1 B! 1 then reflects the reduction in the labour cost,
whereas 8K 1 ~ 1 exhibits the reduction in the capital cost in the pro-
duction of one unit of X 1 . The total reduction in the unit cost of
production is therefore given by 0 1 . Similar remarks apply to 0 2 .
It is worth pointing out that ll; and njare defined to be non-negative.
Turn now to equations (2.6) and (2. 7). Differentiating them
totally, we get
D! = ex 1(p! - p!) + '11 Y* (2.27)
D! = -cx 2(p!-p!)+'72Y* (2.28)
26 Studies in the Pure Theory of International Trade

where tx 1 = (p/D 1 )(oDJiop) and tx 2 = -(p/D 2)(oD 2/op) are the


price elasticities of demand, and 17 1 = (Y/D 1 )(oDJioY) and
17 2 = ( Y/ D 2)(oD 2/o Y) are the income elasticities of demand for the
first and the second commodity, respectively. In the absence of
inferior goods, '1i > 0.
For expository purposes, another relation may now be introduced
on the demand side, namely, the elasticity of demand substitution,
av, which is defined as

av = P
(DdD2)
• o(DdD2) =
op
p[-1 oD 1_. oD 2]
1_

D1 op D2 op
= (XI +tx2 > 0.

Subtracting equation (2.27) from (2.28) gives us

By now we have collected the full complement of ingredients needed


to derive general equations showing the effects of a change in all
parameters on the variables of the system, so that the implications
of a change in any particular parameter can be analysed as a special
case of the general solution. By subtracting equation (2.23) from
(2.24), and (2.26) from (2.25), utilising equations (2.21), (2.22) and
the 121 and 181 determinants from (2.17*), we have

(2.31)

The substitution of (w*- r*) from (2.31) in (2.30) yields

where a. = (X! - Xt)/(p~ - pt) = (p K + pL)/(IA.JIO I) is the elasticity of


substitution between commodities on the supply side, that is, it is
the elasticity of commodity substitution along the transformation
Structure of Production in Autarky 27

curve. t The mutual interaction of demand and supply in a closed


economy requires that Dj = Xj. The right-hand sides of equations
(2.29) and (2.32) are therefore the same in equilibrium. Making use
of this equality then gives us the expression for the change in the
commodity-price ratio:
* * (K*-L*)+(llK-llL)
P! -p2 = IA.i(uv+u.)
u.(n2-nl) Y*(,l-'12)
+ ( + . (2.33)
u 0 + u.) (u 0 + u 5 )

2.4 Commodity and Factor Prices


Equation (2.31) exhibits a relationship among the commodity-price
ratio (p), the wage/rental ratio (w) and the rate of technical advance
in the two commodities. This equation may be rewritten to obtain
(2.31*)
Consider first the case where technical progress is absent in both
industries. The relationship between p and w is then determined
solely by 101, which, from (2.17), is positive if k 2 > k 1 , but negative
if k 2 < k 1 . In other words, in the absence of technical progress a
decline in p leads to a rise in w if the first commodity is labour-
intensive, but to a decline in w if the first commodity is capital-
intensive. Furthermore, the relationship between p and w is mutually
dependent. In other words, a change in w results in a change in p,
and conversely.
This result can be very simply derived from the unit isoquant dia-
gram presented in the previous chapter. The only difference now is
that the diagram will have to include unit isoquants in both commodi-
ties. This is accomplished in Fig. 2.1, where X 2 and X 1 are respectively
the unit isoquants of the second and the first commodity; the slope
of AB displays the wage/rental ratio and, corresponding to AB, the

t This elasticity was first introduced by Jones [2]. Along any transformation
curve, factor supplies and the level of technology are given, so that K* = L * = n L =
ll 1 = ll 2 = 0. Then, from equation (2.32),
(X!-X!)
= (15.
P!-p!
In the absence of market distortions, u,, which also equals [(PL + PK)!(jA.JJOJ)], is
positive, because PL > 0, PK > 0 and JA.J and JOJ always possess the same sign.
28 Studies in the Pure Theory of International Trade

Figure 2.1

equilibrium capital/labour ratio in X 2 is given by the slope of Ob 1


and in X 1 by the slope of Oa 1 . As shown in the previous chapter, the
price of each commodity can be expressed exclusively in terms of
the cost of labour or capital. Suppose capital is selected to be the
numeraire. Then p 1 = P2 = r.OA, so that p = p 2 /p 1 = I. It is not
difficult to see that the reason for a unit value of plies in the fact that
both unit isoquants lie on the same isocost line. Now suppose there
is a decline in co, so that AB is replaced by the two parallel lines CD
and EF, both of which exhibit a smaller wage/rental ratio. Two
changes take place. First, the capital/labour ratio in each industry
declines- in X 2 from the slope of Ob 1 to Ob 2 and in X1 from the
slope of Oa 1 to Oa 2 • Second, P2 = r. OC and p 1 = r. OE, which
means that p, now equal to OC/OE, has risen above its previous
level of unity. It may be observed that the diagram depicts X 2 to be
capital-intensive relative to X 1 at all wage/rental ratios, as is evident
from the fact that the slope of Ob 1 and Ob 2 exceeds that of Oa 1 and
Oa 2 , respectively. Inherent in this construction also is the fact that
factor intensities are non-reversible. This, as we shall see later, is
Structure of Production in Autarky 29

n
R

\ '•-----------+--:.......o<------,r

Figure 2.2

D
R

p 0 k

Figure 2.3

attributable to the fact that the unit isoquants intersect only once.
A similar diagram can be drawn for the case where k 2 < k 1 .
The 'one-to-one' relationship between w and p on the one hand,
and between w and ki on the other, is depicted in Figs. 2.2 and 2.3.
30 Studies in the Pure Theory of International Trade

In the first quadrant is depicted the positive relationship between w


and ki. In Fig. 2.2, k 2 < k 1 at any w, whereas k 2 > k 1 for all win
Fig. 2.3. The second quadrant displays the unique relationship
between p and win terms of the RR curve. When k 2 > k 1, RR is
negatively inclined, whereas with k 2 < k 1 , RR exhibits a positive
slope. However, not all points on RR are consistent with the con-
straint imposed by the full-employment condition. Suppose the
overall capital/labour ratio (k) in the economy is given by Ok. Then
the range for variations in w is given by w 1w2 . For, as established in
the previous section, if, at any w, k comes to equal the capital/labour
ratio in any commodity, the output of the other commodity falls to
zero. If w is given by Ow 1 , then k = k 1 , so that both factors will be
completely employed in the first industry. On the other hand, if w is
given by Ow 2 , k = k 2 , so that both factors will be fully employed in
the second industry. Any w above Ow 1 in Fig. 2.3 is clearly beyond
the approach of an economy with a constant level of k. Hence the
range of variation for w is determined by the economy's overall
capital/labour ratio in conjunction with the Ok 1 , Ok 2 curves. It
follows that, since w must vary between w 1 and w2 , p must vary
between R 1 and R 2 • Any commodity-price ratio lying outside the
range R 1 R 2 will otherwise result in complete specialisation in any
one commodity. Thus the 'operational' range of the RR curve is
given by the dotted portion, w 1abc. The diagram, of course, is in
conformity with the result derived in section 2.2, namely, wand p
are no longer related in the presence of complete specialisation.

Factor-Intensity Reversals
Until now we have assumed that factor intensities are non-reversible.
The analysis becomes more complicated and vexing if factor inten-
sities are reversible. Symbolically, the problem is one of confronting
the multiple signs of I01 at different levels of w. For if k 2 ~ k 1 at
different w's, 101 ~ 0. Evidently, then, the relationship betweenp and
w will no longer be unique. Geometrically, the problem is illustrated
in Fig. 2.4. The only difference between Figs. 2.4 and 2.1 lies in the
presence of two points of intersection between the two unit isoquants
in Fig. 2.4, as against the single intersection point in Fig. 2.1. As
before, we commence with the isocost line AB and a unit commodity-
price ratio (p). As the wage/rental ratio declines, we find that X 2
becomes labour-intensive relative to X1 because Ob 2 is less steep
than Oa 2 , although in the initial situation X 2 was capital-intensive
Structure of Production in Autarky 31
K

L
Figure 2.4

p
Figure 2.5

relative to X 1 • In other words, factor intensities have been reversed


in the new situation. Moreover, in contrast to the situation depicted
in Fig. 2.l,p has now declined to the level given by OE/OC < I.
The implications of the factor-intensity reversals for the com-
modity-price ratio are pictured in Fig. 2.5, where, as before, the first
quadrant portrays the unique relationship between w and ki, but
32 Studies in the Pure Theory of International Trade

where, unlike the previous diagrams, factor intensities get reversed at


points such as a and b. If w lies between 0 and w 1 , k 1 > k 2 ; if it lies
between w 1 and w 2 , k 1 < k 2 ; and beyond w 2 , k 1 > k 2 again. In
the presence of the factor-intensity reversals, the slope of the RR
curve is also altered at points such as c and d corresponding, respec-
tively, to a and b. It can be easily seen now that the relationship
betweenp and w is no longer unique. For example, to one price ratio
OR 1 there correspond three values of w (w', w" and w).

2.5 Factor Endowment and Commodity Prices


The relationship between the commodity-price ratio and the overall
capital/labour ratio (k) can be ascertained from equation (2.33).
Here again, we assume that the level of technology is given, so that
nK = nL = n2 = n1 = 0. With this simplification, equation
(2.33) reduces to
K* -L* Y*(171 -172)
-p* = PT-p! ..---:-.---,--------,-+
-----,--__:_:_::__..:..=._ (2.33*)
l,l.l(cr 0 + cr.) (cr v+ cr.)
Since cr v and cr. are both non-negative, the sign of (p!- p!) de-
pends crucially on the signs of I.A.I, and (17 1 -17 2). Further simpli-
fication can be achieved if we assume that the community preferences
are homothetic, so that both income elasticities of demand are
identically equal to unity.t Under these circumstances, a rise in
k(K*-L* > 0) will lead to a decline in p(p!-PT < 0) if 1;.1 > 0,
which from (2.17) means that k 2 > k 1 • On the other hand, if
I.A.I < 0, so that k 2 < k 1 , k and pare positively related. These two
cases are diagrammatically depicted in Figs. 2.6 and 2. 7. However,
if factor intensities are reversible, so that 1;.1 is positive for some
levels of w but negative for others, the relationship between k and p
will no longer be unique.
What happens if community preferences are not homothetic, so
that neither 17 1 nor 17 2 equals unity? For expository purposes, let us
assume that the change ink is effected by a change inK alone, without
being accompanied by a change in L, so that L* = 0. This implies
that k* and Y* are positively related. A rise in k shifts the trans-
formation curve away from the origin, resulting thereby in a higher
t By homothetic preferences we mean that the communities' taste pattern is
described by a homothetic social utility function, so that the commodity-price ratio
is exclusively determined by the ratio between the output levels of the two com-
modities.
Structure of Production in Autarky 33
k

0 p

Figure 2.6

Figure 2.7

level of national income. The introduction of the possibility ofnon-


homothetic preferences, however, may or may not alter the com-
plexion of the analysis. If IA. I < 0, that is, if the second commodity
is labour-intensive relative to the first commodity, the relationship
between k andp is reinforced, provided '7 2 > '7 1 • Stated differently,
a rise in k will now result in a greater rise in p than the case would
be if '7 1 and '7 2 were equal. The reason is not difficult to pinpoint.
If the relative price of the second commodity rises as a result of a
rise in k in the presence of homothetic preferences, the rise in p will
34 Studies in the Pure Theory of International Trade

be even greater if the demand change due to the income change was
biased in favour of the second commodity, that is if '7 2 > 'It. The
same reasoning applies to the case of a decline in k.
The relationship between k and p continues to be unique if lA. I > 0,
that is, k 2 > k t, and '7 2 < 'It· Here a rise in k will result in a larger
decline in p than before, and conversely.
The picture becomes blurred if the terms on the right-hand side
of (2.33*) possess opposite signs. This will be the case if (i) lA. I < 0
but '7 2 < 'It• or (ii) IA.I > 0 but '7 2 > 'It· To say the least, there-
lationship between k and p is now weakened. Furthermore, the
previous relationship between k and p may be reversed; that is to
say, if k and p were positively related when '7t = 17 2 , they may be
negatively related when the influence of differing income elasticities
of demand is taken into account, and conversely.

2.6 Technical Change and Commodity Prices


The implications of technical progress for the commodity-price
ratio can be explored as a special case of the general solution pro-
vided by equation (2.33). For the sake of convenience, let us assume
that factor endowments are given, so that K* = L* = 0. Equation
(2.33) then simplifies to

-p* =Pi-p!= IA.llK-llL (02-llt) f*('lt-172)


I(uv+u.) + 0' (uv+u.)
5
+ (uv+u.) . (2.33')

At first glance, equation (2.33') is not amenable to simple and un-


ambiguous interpretation. Even if we assume that preferences are
homothetic so that the last term in the equation disappears, myriad
possibilities suggest themselves, and all may be equally plausible on
a priori grounds. Drastic simplification is needed if we wish to derive
unambiguous results of some kind. The analysis will become more
manageable if we rewrite the expressions for ll; and nj presented in
section 2.3. Thus
ll; = A.oBrt +A.;2Br2 (i = L,K)
llj = ()LjBLj + ()KjBkj (j = I ,2).
If we assume that technical progress is Hicks-neutral, then
Bi.j = Bkj·
Using this relationship and expression (2.17*), we can
show that
Structure of Production in Autarky 35

Substituting this in equation (2.33'), we obtain

(2.34)

As before, matters are simplified if we assume that preferences are


homothetic, so that 17 1 = 17 2 = 1. The change in the commodity-
price ratio is then simply determined by nj, the rate of technical
advance in thejth industry. If technical progress occurs at a uniform
rate in both industries, that is, n2 = n1, the commodity-price ratio
remains unaltered, for then p! = p!. On the other hand, if n 1 > ll 2 ,
that is, if the first commodity enjoys a greater rate of technical
improvement than the second commodity, the commodity-price
ratio rises (p! < p!), and conversely.
When income effects are introduced, the relationship between p
and n tfll 2 may be reinforced or weakened. If 17 2 > 17 1 , the relation-
ship is strengthened; if 17 2 < 17 1 , the relationship may be either
weakened, or actually become negative.

2.7 Other Properties of the Model


Sections 2.4 and 2.5 in the foregoing analysis of this chapter have
been concerned, respectively, with the implications of changes in the
factor-price ratio and the overall capital/labour ratio for the com-
modity-price ratio in a closed economy. The equations of change
that were used to analyse these effects can also be used to examine
some other related theorems, of great intrinsic interest and import-
ance, which deal directly with the impact of a change in any
particular parameter on the individual variables rather than their
ratios.

The Stolper-Samuelson Theorem


Consider first the implications of a change in commodity prices for
real factor rewards. A categorical analysis of this problem was first
provided by Stolper and Samuelson in a pioneering article [5].
According to Stolper and Samuelson, a rise in the price of a com-
modity results in a rise in the real reward of its intensive factor and a
decline in the real reward of its unintensivefactor, and vice versa. For
a demonstration of this result we go back to equations (2.25) and
36 Studies in the Pure Theory of International Trade

(2.26), which in conjunction with (2.17*) can be solved for given


factor endowments and technological levels to furnish
• • oKj<Pt-pn
w -pj = IOI (2.35)

r*-pt =
-fh·<Pt-Pn
J IOI u = 1,2). (2.36)

Clearly, the impact of a change in the commodity-price ratio on the


factor rewards in terms of any commodity price is determined by the
sign of 101. Suppose the first commodity is the labour-intensive of
the two, so that k 2 > k 1 and I01 > 0. Then a rise in the relative price
of the first commodity, implying that (pt- pn > 0, promotes a
rise in the wage rate, which is the reward of its intensive factor, and
a decline in the rental of capital, which is the price paid to its un-
intensive factor. Since the wage rate increases and the rental rate
declines in terms of both commodity prices, the real reward oflabour
increases and the real reward of capital decreases unambiguously.
If 101 were negative, that is, if k 2 < k 1 , then labour would un-
ambiguously suffer but capital would benefit as a result of a decline
in p, and conversely. It should be obvious by now that this result
derives directly from the one-to-one relationship established in
section 2.4 between the wage/rental ratio and the commodity-price
ratio. Here, as before, incomplete specialisation is a necessary con-
dition for the full validity of the Stolper-Samuelson theorem.
For a geometrical proof of this theorem, we revert to Fig. 2.1,
which is drawn under the assumption that k 2 > k 1 • It may be recalled
that the relative price of the second commodity associated with the
wage/rental ratio given by the isocost line AB equals unity, but that
associated with the lower wage/rental ratio furnished by the slope
of CD, parallel to EF, equals OC/OE > I. This of course brings us
back to the result that a rise in p has resulted in a decline in w. In
addition, however, the real wage rate declines, but the real reward
of capital rises. For prior to the rise in p, real reward of labour in
both industries was given by 1/0B, and the real reward of capital
by I/O A, but after the rise in p the real wage rate has declined to
1/0F in terms of the first commodity and to 1/0D in terms of the
second commodity, whereas the real rental has risen to I/OE in
terms of the first and to I/OC in terms of the second commodity.
The effects of a decline in p are symmetrically opposite.
Structure of Production in Autarky 37

The Rybczynski Theorem


Another remarkable theorem is associated with T. M. Rybczynski
[3], who showc:d that a rise in the supply of a factor at constant
commodity price:. promotes expansion of the commodity which utilises
the expanding factor relatively intensively and contraction of the other
commodity. The economic explanation of this theorem is rather
straightforward. Suppose there is an increase in the supply of capital
alone. At constant commodity prices, factor rewards and hence factor
proportions in each industry are unaltered. Under the full-employ-
ment constraint the additional capital stock must be absorbed in
the economy, and this, with the constant capital/labour ratio in each
industry, requires an expansion in the output of the capital-intensive
commodity. Since both factors are indispensable in the production
process, the additional labour needed by the capital-intensive com-
modity in order to utilise the additional capital stock must, with the
constant labour endowment, be withdrawn from the other com-
modity, which means that the output of the labour-intensive good
must decline.
A simple algebraic demonstration of this phenomenon follows
directly from the solution of (2.23) and (2.24) and the use of (2.17*),
so that
(2.37)

n - AuK*-A.KlL*
(2.38)
2 - 121
With a rise in the supply of capital alone, K* > 0 but L* = 0. The
proof of the Rybczynski theorem then follows directly by observing
that
Xi§ 0,
A simple diagrammetic derivation of the Rybczynski theorem can
be accomplished by utilising the box diagram presented in the previ-
ous chapter. Consider Fig. 2.8, where the factor supplies are
measured along the axes merging in the origins 0 1 and 0 2 and the
outputs by the distance of the production point from the respective
origins. Suppose A is the initial production point lying on the con-
tract curve 0 1 A0 2 , so that the capital/labour ratio in each industry
is given by the slope of 0 1 A and 0 2 A, reflecting that the first com-
modity is labour-intensive relative to the second commodity, and
38 Studies in the Pure Theory of International Trade
Labour
r----------------------------,r--------,0~
I
I
I

AI/
,-;~
I
/1 ><"
0
'SQ.
...
:I
0

0 o"
------Output of x2
Figure 2.8

the output of X 1 and X 2 equals 0 1 A and 0 2 A, respectively. The


supply of capital and labour constraining the contract curve 0 1 AO 2
is given, respectively, by 0 1 0' and 0 1 0. Now suppose the supply of
labour increases by 00" to 0 1 0". Since the capital/labour ratio in
each commodity is unaltered at constant commodity prices, the new
production point is given by A', which lies on the extension of 0 1 A
and O~A', drawn parallel to 0 2 A. The new production point,
which -lies on the new contract curve (not drawn), shows that the
output of the labour-intensive commodity, X 1 , rises to 0 1A' and
the output of the capital-intensive commodity, X 2 , declines to O~A'.
The diagram can be easily adapted to the case where the supply of
capital, instead of labour, has increased.

Duality and the Magnification Effect


The formulation of the model in terms of equations (2.1 )-(2.4)
serves to reveal the fundamental duality that exists between the
Stolper-Samuelson and the Rybczynski theorems. Moreover, the
effects of variations in commodity prices on factor rewards and in
factor endowments on individual outputs at constant commodity
prices are magnified to show that a change in the price of a com-
modity alters the factor rewards in a greater proportion, whereas a
change in the supply of any factor gives rise to a more than pro-
Structure of Production in Autarky 39

portionate change in the two outputs. This so-called magnification


effect, which further emphasises the dual relationship between the
theorems described in this section, becomes vivid when equations
(2.35)-(2.38) are written in the following manner:
w* {)KI
PT I +TOT (2.35*)

r*
PT
,_w
eu (2.36*)

X! ;.L2
L* l+g (2.37*)

X! Au
L* '-m· (2.38*)

Equations (2.35*) and (2.36*) are derived under the condition that
only the price of the first commodity changes, e.g. P! = 0, whereas
the condition behind (2.37*) and (2.38*) is that K* = 0. It may
now be evident that the link provided by I01 between commodity
prices and factor rewards is symmetrical to the link provided by IJ.I
between factor endowments and the individual outputs. Further-
more, each equation serves to show the magnification effect. For
example, suppose that k 2 > k 1 so that both IA.I and 101 are positive.
Then a rise in the price of the labour-intensive commodity (implying
that PT > 0) stimulates, at constant k, a more than proportionate
rise in the reward of labour, whereas a rise in the supply of labour
alone at constant p results in a more than proportionate increase in
the output of the labour-intensive commodity (see equations (2.35*)
and (2.37*)). On the other hand, ifk 2 < k 1 , i.e. ifiJ.I and 101 are
negative, the magnification and the dual effects are reflected in
equations (2.36*) and (2.38*). t
t Any remaining doubts concerning the dual relationship may be dispelled by
observing that, if P! = K* = 0, then the substitution of (2.17) in (2.35) and (2.37)
yields
dw dX 1
dp 1 dL
On the other hand, if p! = L* = 0, then
dr dX 2
dp, dK CL,(k,-k!)
40 Studies in the Pure Theory of International Trade

2.8 Autarky Equilibrium


By definition, the equilibrium under autarky is reached when de-
mand and supply for all goods are equal. If we assume that the
community behaves like a single consumer, the problem may be
formulated as one of maximising an aggregate utility function U
subject to the constraint imposed by available factor supplies, a
constraint inherent in the construction of the transformation curve
which describes that X 1 = X 1(X2 ). If the community welfare is a
function of the aggregate consumption of the two commodities, then
the aggregate utility function may be written as
U = U(D 1,D2).
The problem is then to maximise U subject to
Dj =xi
and
X1 = X1(X2).
This is accomplished by forming a Lagrangian H as H =
U(D 1,D 2)-A. 1[D 1-X1(X2)]-A. 2(D 2-X2), where A. 1 and A. 2 are
the Lagrangian multipliers. The first-order conditions for an interior
maximum are
iJH iJH iJH iJX1
an1 = u~-A.~ = an2 = u2-A.2 = ax2 = A.l ax2 +A.2 = o
where ~ = iJU/iJDi is the marginal utility of the jth commodity.
The solution of these equations furnishes
U2 A.2 ax1
u1 = ...tl = - ax2.
Under perfect competition in all markets we know that - iJXtfiJX2 =
U2 /U1 = p. Hence welfare is maximised whenp equals the slope of
the transformation curve as well as the community indifference
curve, which is a geometrical counterpart of the aggregate utility
function. This is achieved in Fig. 2.9 at S, the point of self-sufficiency
equilibrium.

2.9 The Community Indifference Curve


In the preceding section, the concept of community indifference
has been used to illustrate how the closed-economy equilibrium is
Structure of Production in Autarky 41

x,

Figure 2.9

attained. The uniqueness of the autarky equilibrium in Fig. 2.9 is


attributable to the existence of a well-behaved, non-intersecting
community indifference map which ensures that there is only one
point such as S where a community indifference curve is tangential
to the transformation curve. In other words, it is only when the
community indifference map is a national counterpart of the indivi-
dual indifference map that it is possible to obtain a unique equili-
brium point. This procedure., however, gives rise to several vexing
questions, a definite answer to which, contrary to the popular
impression, is not yet in sight. If all individuals constituting a com-
munity or nation possess identical tastes and incomes, there is no
problem in deriving the aggregate utility function that will have the
properties of individual utility functions. But to postulate this con-
dition is to beg the question entirely. Where in the world can we
find a society enjoying such a 'pure' form of communism? The
assumption that individual preferences are identical is certainly not
outrageous, but the assumption that people have identical incomes
certainly is.
Let us first see how a community indifference curve is constructed.
A community indifference curve, as defined by Scitovsky [ 4], is a
locus of quantities demanded at various commodity prices and a
42 Studies in the Pure Theory of International Trade
x,
u,

u,

0
Figure 2.10

given constant distribution of utilities among individuals. Consider a


community of individuals exchanging among themselves two com-
modities, X 1 and X 2 , which are respectively measured along the
abscissa and ordinate in Fig. 2.10, and suppose that the initial con-
sumption point, for given tastes and incomes, is given by C. Now
suppose that the price of X1 rises, inducing members of the com-
munity to forgo a certain quantity of it, and move vertically down
to the consumption point given by C 1 • In principle, we can ascertain
how much extra X 2 will be needed to make every member of the
community just as well off as he was prior to the price change. The
new consumption point, if this compensating substitution does
occur, will lie to the south-east of C. Suppose C2 is such a con-
sumption point. By varying the amount of the initial price change,
an infinity of points such as C 2 can be derived. The locus of these
points, Scitovsky asserts, makes up a community indifference curve
such as U1 U1 in Fig. 2.1 0, which has the same geometrical properties
as the underlying individual preference maps.
The important fact to remember about Scitovsky's construction
is that it assumes a fixed utility (or income) distribution throughout.
In general, as recognised by Scitovsky himself, an infinite number of
community indifference curves pass through any point in the com-
Structure of Production in Autarky 43

modity space, each one corresponding to a different initial income


distribution. Thus in order to obtain a well-behaved, non-intersecting
community indifference map such as the one depicted in Fig. 2.9,
the initial underlying income distribution must be kept constant.
This is not an easy task, especially in an economy with more than
one factor of production where different goods utilise factors in
different proportions. If there was only one factor of production, or
if the factor proportions were similar in all activities of production,
the income distribution would be invariant as we moved along the
resultant straight-line transformation curve in order to obtain the
autarky equilibrium point. The community indifference map would
then be just as well behaved as the individual indifference maps.
However, the picture is more complicated and batHing when factor
proportions differ in each industry. Here the movement along the
transformation curve would induce changes in commodity prices,
which in turn, from the Stolper-Samuelson theorem, would alter
the complexion of income distribution, say between wage-earners
and owners of capital. The situation becomes particularly serious if
the tastes of the wage-earners are biased towards the consumption
of the labour-intensive good and those of the rental earners are biased
towards the consumption of the capital-intensive good. In order to
find a way out of this dilemma, recourse must be made to two condi-
tions succinctly described by Chipman ([1], p. 695):

To summarise, we may say that if utility functions of all people


are positive homogeneous, and if either (1) all people have
identical tastes, or (2) all people have a distribution of resources
proportionate to the aggregate distribution, then their behaviour
can be represented by a single utility function. Since these two
conditions are independent, and each one of them is sufficient,
obviously neither one of them is necessary in itself.

As recognised by Chipman, these are sufficient, though not necessary,


conditions for the existence of an aggregate utility function. More
important, however, is the fact that if only one of the two conditions
(l) and (2) holds, the resulting community indifference curves will
describe merely the consumption behaviour of the society, but will
have no welfare connotation. In other words, aggregate demand
functions can be derived from the indifference curves, although we
cannot say that the community is indifferent among the various
44 Studies in the Pure Theory of International Trade

output bundles along any one of these curves as the latter are not
uniquely related to a single distribution of utilities among indivi-
duals. However, if individual utility functions are homothetic and
if conditions (I) and (2) are simultaneously satisfied, then the com-
munity indifference curves obtained will constitute a genuine
counterpart of each individual's indifference curves both in the
behavioural and in the welfare sense.

2.10 Summary
The purpose of this chapter was to analyse factors that affect the
commodity-price ratio in a closed economy. We studied how the
commodity-price ratio changes under the thrust of changes in factor
prices, factor endowments and levels of technology in the two
industries. Using a two-commodity, two-factor model with perfect
competition, neo-classical production functions, full employment
and non-reversible factor intensities, etc., the following results were
derived:
I. At constant factor endowment and technology, there exists a
unique relationship between the commodity-price ratio and the
factor-price ratio. This relationship is, however, no longer
unique if factor intensities are reversible. Furthermore, a rise in
the price of a commodity leads to a more than proportionate
rise in the reward of its intensive factor and a decline in the
reward of its unintensive factor, and conversely. The latter
result is usually called the Stolper-Samuelson theorem.
2. Under unchanged technology and homothetic preferences,
there exists a unique relationship between the overall capital-
labour ratio (k) and the commodity-price ratio (p). However,
when income effects are introduced, that is, when community
preferences are non-homothetic, or when factor intensities are
reversible, the relationship between k and p may no longer be
monotonic, in which case there arises a strong possibility of
multiple equilibria.
3. At constant technology and commodity prices, a rise in the
supply of a factor results in a more than proportionate rise in
the output of the commodity utilising the expanded factor in-
tensively at the expense of the output of the other commodity,
and vice-versa. This is the so-called Rybczynski theorem.
4. For any k and homothetic preferences, there exists a monotonic
Structure of Production in Autarcky 45

relationship between changes in p and relative rates of Hicks


technical advance in the two industries. However, this relation-
ship may not be monotonic when income effects are also taken
into account.
5. There exists a dual relationship between the Stolper-Samuelson
and the Rybczynski theorems.
These are some of the properties of the model which have been
extensively utilised in the current literature on trade theory. It may
be stated without any fear of exaggeration that the model yields
conclusive, elegant and straightforward theorems. These properties
will be utilised again and again in the analysis of subsequent
chapters. For the time being, however, we wish to emphasise that
the elegant properties of the model depend crucially upon the
assumptions made at the beginning of this chapter, assumptions the
true significance of which will be revealed only as we proceed with
the analysis in succeeding chapters.

REFERENCES
[1] Chipman, J. S., 'A Survey of the Theory of International Trade: Part 2, the Neo-
Classical Economy', Econometrica, XXXIII (Oct 1965) 685-760.
[2] Jones, R. W., 'The Structure of Simple General Equilibrium Models', Journal
of Political Economy, LXXIII (Dec 1965) 557-72.
[3JRybczynski, T. M. 'Factor Endowment and Relative Commodity Prices',
Economica, XXII (Nov 1955) 336-41.
[ 4] Scitovsky, T., 'A Reconsideration of the Theory of Tariffs', Review of Economic
Studies, IX (summer 1942) 89-110.
[5] Stolper, W. F., and Samuelson, P. A., 'Protection and Real Wages', Review of
Economic Studies, IX (Nov 1941) 58-73.
3 The Basis of International
Trade
3.1 Introduction
In the last chapter we studied how changes in the commodity-price
ratio occur under the thrust of variations in the factor-price ratio,
the overall capital/labour ratio and the relative rate of technical
advance in the two industries. Any theory attempting to explain
the basis of international trade must always commence with the
theory of resource allocation and production in a closed economy.
This task having been accomplished in the previous chapters, we
are now in a position to pinpoint the factors that determine a
country's pattern of trade. The issue is what goods a country will
export and import. Stated differently, is it possible to predict a
country's configuration of exports and imports just by examining
the characteristics of a closed economy? Seeking a clear-cut answer
to this query constitutes the subject-matter of this chapter.
The traditional answer to the question of why a particular country
exports a particular commodity is simply that, owing to the handi-
work of nature or man, it is able to produce that commodity at a
lower comparative cost than the rest of the world, usually taken to
represent a single entity in the community of nations. International
trade occurs because countries stand to benefit by the exchange of
goods that are produced at dissimilar relative costs and hence at
different relative prices. It is desirable, therefore, to go deeper and
unearth the causes responsible for inter-country disparities in the
comparative cost or the commodity-price ratios. The reasons,
quite ostensibly, must be sought in the international differences in
factor endowments, tastes or levels of technology, because, as
established in the preceding chapter, these are the forces which
influence the commodity-price ratio in autarky. However, before
we get involved in a detailed examination of the theories identifying
themselves with these forces, it is imperative that the necessary
conditions preventing the occurrence of international trade be
expressly spelt out.
The Basis of International Trade 47
3.2 Conditions for the Absence of Trade
If countries were identical in all respects in the absence of the
opportunity to trade, the commodity-price ratios would be similar
everywhere and there would be no incentive to trade even if such an
opportunity was available. Obviously, this situation requires the
fulfilment of the following conditions:
I. Production functions are similar internationally.
2. Factor endowments are the same internationally.
3. Tastes are similar in all countries.
4. Production functions are neo-classical.
5. All markets are universally characterised by perfect
competition.
This set of conditions is sufficient to generate inter-country
identity on both the demand and the supply side, resulting thereby
in similar autarky prices and, as a consequence, precluding any
possibility of mutually profitable trade. Clearly, then, the relaxation
of any of these five requirements could provide a basis for inter-
national trade. In other words, there are five different ways in which
an 'international economy' could be simulated by the trade
theorist. However, from the viewpoint of propositions concerning
the pattern of trade, the two most widely known simulations are
those of Ricardo [11] and Heckscher-Ohlin [3, I 0], who,
respectively, formulated their theorems in terms of relaxing con-
ditions (I) and (2). Needless to say, if one of these conditions is
relaxed in order to forecast the pattern of trade, the remaining
four assumptions must be retained if the trade pattern is to follow
the prediction under all circumstances.

3.3 The Ricardian Theory


The classical economists were concerned primarily with the
demonstration of the welfare proposition that trade is beneficial,
and only occasionally with isolating the crucial variables that could
explain the pattern of trade by means of a formal, determinate
model. Yet a general model of trade and resource allocation can be
formulated directly from the insights of Ricardo [II] and Mill [9].
There is little doubt about the plausibility of the former viewpoint.
None the less, the foundations of most of the early expositions of
the comparative cost doctrine lie in a definite structural model,
however primitive to the modern writer, that ca~ be used to explain
the pattern of international specialisation.
48 Studies in the Pure Theory of International Trade

On this latter interpretation, the Ricardian theory of comparative


advantage can be constructed directly from Ricardo's famous
numerical example concerning unit costs of production in England
and Portugal. According to the Ricardian theorem, a country
exports that commodity which has higher comparative factor pro-
ductivity and imports the commodity which has lower comparative
factor productivity than the other country.
Let us assume that the world consists of only two countries, one
home country (H) and one foreign country (F), each producing
the two commodities, X1 and X2 • Following Ricardo, suppose
further that there exists only one factor of production, say labour,
and that returns to scale are constant. As established in Chapter 1,
the transformation curve under the postulated conditions is given
by a negatively inclined straight line whose slope reflects the ratio
of labour productivities. It can be shown directly from equations
(2.3) and (2.4), presented in the previous chapter, that the autarky
commodity-price ratio (p = p 2 /p 1 ) is determined exclusively by the
output-input ratios, provided labour (or capital) is the only
factor of production. Hence, with CK 1 = CK 2 = 0, p, from (2.3)
and (2.4), is given by
CL2 a1
p=-=- (3.1*)
Cu a2

where ai = lfCLi is the average productivity of labour in the jth


commodity. Thus the pre-trade commodity-price ratio in the
Ricardian model is determined neither by factor supply nor by
demand, but only by the state of technology hidden behind the
output/labour ratios. If we postulate an identical model for both
countries, it follows that the reason for the disparity in the two
countries' autarky price ratios and hence for the existence of
international trade must be ascribed to international differences in
the relative output/labour ratios or in the state of technology. Put
another way, the basis for international trade lies in international
differences in production functions - a violation of condition (1)
stated in the previous section. To consider a specific example,
suppose that the comparative labour productivity in the first
industry is higher in the home country; that is,
(3.1)
where the subscripts f and h respectively denote countries F and H.
The Basis of International Trade 49

(In what follows, the subscripts h andfwill be introduced wherever


it is necessary to distinguish between the variables in the two
countries. If these subscripts do not appear, the variables will be
taken to refer to the home country, except when specified other-
wise.) It follows immediately from this relationship that
(3.2)
so that the relative price of the second commodity in country H is
higher than that in country F. It is now relatively simple to deduce
that H will export the first commodity in which her comparative
labour productivity is higher, and import the second commodity
in which her comparative labour productivity is lower than that in
F. This is the proof of the Ricardian theory of comparative
advantage.
Geometry will perhaps throw further light on the issues involved
in the Ricardian theorem. Consider Fig. 3.1, where HH' and FF'
are, respectively, the linear transformation curves in countries Hand

Figure 3.1
50 Studies in the Pure Theory of International Trade

F, with their slopes reflecting the autarky price ratios satisfying


the relation expressed by (3.2). If we assume that there are no
transport costs and no impediments to the free international flow
of goods, there prevails only one commodity-price ratio in the
free trade equilibrium. The reason is that the consumer in each
country consumes the importable goods at the price charged by
producers in the other country. As a result, the commodity-price
ratio is the same in both countries in the free trade equilibrium
with no transport costs. Even so, the common terms of trade
prevailing in the international trade market are normally different
from the autarky price ratios, except in the singular case where the
gigantic size of one of the trading partners prevents the trade with
the relatively small country from having any impact on its own
domestic prices, in which case the common international terms of
trade are given by the large country's autarky price ratio. This latter
possibility will not be pursued here, but will be taken up again in
the chapters on gains from trade. Nor shall we examine the factors
that go into the determination of the international terms of trade.
For the time being it is sufficient to note that, barring the large
country case, the international terms of trade (p') lie between the
two autarky price ratios. Under relation (3.2), it means that
Ph> p' > PJ·
The common free trade price ratio in Fig. 3.1 is given by the
slope of AF' which is parallel to BH. One feature of the Ricardian
model is that, with world prices differing from autarky prices, each
trading partner must end up with complete specialisation in the
exportable commodity, even though in the absence of trade both goods
were being produced in both countries. This phenomenon is
attributable to the constancy in the unit cost of production at all
levels of output for any commodity, so that as the economy is
exposed to different prices resulting from the introduction of trade,
the producers find it profitable to go all the way in expanding the
output of the commodity whose relative price has risen. Conversely,
the producers of the commodity with diminished relative price are
totally thrown out of business, because at all output levels the
price falls short of the fixed unit cost. Since in each country it is the
importable commodity which suffers a decline in price, the transi-
tion from autarky to free trade lands both trading partners in a
situation where only their respective exportable good is produced.
The Basis of International Trade 51

The self-sufficiency equilibrium point in the two countries is given


by Shand S1 , but the free trade production points are given by H in
the home country and F' in the foreign country.

The Role of Demand


We have seen above that the factor supply and commodity demand
conditions play no role in the determination of the autarky price
ratio. This perhaps led to the long-held belief that the demand
conditions do not enter into the determination of the trade pattern
in a Ricardian world. The fallacy of this view has been recently
pointed out by Bhagwati [I]. His proof consists in the demonstra-
tion that, even if the autarky price ratios were everywhere the
same, international trade in the Ricardian simulation of the world
economy could still occur because of the possibility of multiple
production equilibria stemming directly from the prevalence of the
linear transformation curve. A neat exposition of this point
emerges from Fig. 3.2, where HH' and FF' are the transformation
curves belonging, as before, to H and F, but unlike the case in
Fig. 3.1 HH' and FF' are parallel, showing that the autarky price

x,

Figure 3.2
52 Studies in the Pure Theory of International Trade

ratios are similar in both countries. Suppose that country H selects


consumption at Ch and country Fat C1 . For the sake of visual
clarity, points C1 and Ch are vertically lined. Owing to the constancy
of the unit production cost, the production point in each country
may lie anywhere on its transformation curve. If the home
production point lies at Chand the foreign production point at C1 ,
there is no disparity in the demand-supply position in both
countries, so that there will be no need for international trade.
Suppose, however, that the production point in the home country
is given by Ph whereas in the foreign country it is given by P1 , such
that the distances RCh and QP1 are equal. The domestic needs of
each country can be satisfied only through trade. For in the
absence of trade the demand-supply equilibrium may never be
achieved without some kind of government intervention. Under
constant prices and incomes, consumers certainly have no reason
to alter their demands. Similarly, in the face of given prices,
producers have no incentive to alter the composition of their
output, unless, of course, the government requires them to do so.
The only solution to this impasse may then be the introduction of
international trade. In the example just presented, the home country
exports RPh of X1 in exchange for QP1 of X 2 •
However, the trade pattern in this type of situation need not be
uniquely determined. If the production point were given by Pl. in
the home country and Pj in the foreign country, such that R'PI.
equalled Q'C1 , the trade pattern would be reversed: H will now
export R'PI. of X 2 in return for Q'Pj of X1 from F. The congruent,
dotted triangles exhibit the possibility of the export of X 2 and the
import of X1 by the home country, whereas the other set of identical
triangles, the striped ones, display the possibility from the view-
point of the home country of exporting X1 and importing X2 • Thus
trade in the Ricardian model may occur even if autarky price ratios
are similar everywhere simply because the presence of the linear
transformation curve permits the multiplicity of production
equilibria. However, the volume and the pattern of trade are just
as indeterminate as the autarky production equilibria. In this sense,
demand does play an important role. This analysis, of course,
leaves unanswered the question as to what happens if the production
in each country, different from the consumption point, is such that
the export-import offers do not match. Presumably some sort of
governmental intervention would then be unavoidable.
The Basis of International Trade 53

Bhagwati [I] provides yet another demonstration where the lack


of restrictions on demand conditions may give rise to the
invalidity of the Ricardian theorem even if autarky price ratios
between the two countries were different. The villain now lies in
such demand conditions as may result in multiple self-sufficiency
equilibria. The discussion on the concept of community indifference
in the previous chapter made it clear that the multiple self-
sufficiency equilibria could occur in the absence of well-behaved,
non-intersecting community indifference curves. In such a case, it
is possible that there may be no trade in spite of the international
differences in production functions. Consider Fig. 3.3, where the
transformation curve of country F is omitted for the sake of
simplicity. Let us suppose that, in the absence of trade, the self-
sufficiency equilibrium is given by Sh, where the home transforma-
tion curve is tangential to the community indifference curve U1 U1 .
As the closed economy is opened to free trade the commodity-price
ratio is given by the slope of BH and the home production shifts
to point H. Now if the community preference map were well
behaved, the consumption point will shift to a point such as C, the
home country will export HG of X1 and import GC of X 2 , and

Figure 3.3
54 Studies in the Pure Theory of International Trade

the home welfare would improve to the level indicated by V3 V3 ,


which, as can be seen, does not intersect V 1 V 1 • However, if the
community preference map is not well behaved, the new consump-
tion point could lie anywhere on BH including the point H. This
latter case is depicted in the diagram, where the price line BH
touches the community indifference curve V 2 H at H and there is
no possibility of any trade, because even at the price ratio given by
BH, production and consumption are the same. Thus the Ricardian
theorem may not be valid if certain restrictions are not placed on
demand conditions. Fortunately, these restrictions transpire to be
insignificant. Since income distribution, as established in the
previous chapter, remains constant in a single-factor economy, all
WI! need to assume, in order to secure well-behaved community
indifference curves, is that the individual indifference curves are
well behaved, a restriction certainly not serious if we remember
the formidable difficulties that stand in the way of obtaining the
well-behaved community preference map in a multi-factor economy
where it may be impossible to keep the initial income distribution
unaltered. An alternative and equally lenient restriction could be
to assume that both goods are demanded in each country at all
commodity prices, so that the extremities of the transformation
curve as possible self-sufficiency equilibria are ruled out by
assumption. Under this later restriction, the trade pattern will
follow the Ricardian dictum even if the community indifference
curves were to intersect. The gist of all this discussion is that some
restrictions on demand conditions are desirable if we wish to
demonstrate the full validity of the Ricardian theorem.

The Two-Factor Economy


The extension of the Ricardian theorem to a two-factor model is
a rather simple matter if we assume that factor proportions in each
industry are the same. In fact the introduction of another factor
does not modify the results at all. The autarky price ratio is still
determined exclusively by the comparative factor productivity of
any one of the inputs. This observation becomes apparent if we
rewrite the price equations (2.3) and (2.4) from the previous
chapter; thus
wCLI +rCKl = P1 (2.3)
wCL2 +rCK2 = P2 (2.4)
The Basis of International Trade 55

Dividing (2.4) by (2.3), we obtain

p = weL 2[l +(rfw)(eK 2feL2)]


--~7---~~-=~~~ (3.3)
weLl [I +(r/w)(eKtfeLl)]
so that with
(eKtfeLl) = (eK2feL2) (3.4)
eL2 a1
p=-=- (3.5)
eLl a2

which is exactly the same as equation (3.1 *) derived in the single-


factor case. From (3.4), (3.3) can also be written as
eK2 b1 eL2 al
p=-=-=-=- (3.6)
eKl b2 eLl a2
where bi denotes the average productivity of capital in the jth
industry U = 1,2). Equations (3.5) and (3.6) make it clear that if
each industry possesses the same capital/labour ratio, as is implicit
in (3.4), the autarky price ratio (p) is determined solely by the
comparative factor productivity of either input. Thus the Ricardian
theory of comparative advantage remains intact in a two-factor
economy, provided both commodities employ the two factors in
the same proportion.
The complexion of the analysis changes substantially if factor
proportions in each commodity are dissimilar, for then (atfa 2 ) is
no longer equal to (btfb 2 ), so that it may be impossible to achieve
consistent comparisons between comparative factor productivities
in the two countries. Furthermore, the autarky price ratio will
now be governed by factor supplies and demand conditions also.
Therefore, in order that inter-country technological differences
retain their prominence as the basis of international trade, we must
now assume that factor endowments and demand conditions are
similar internationally. Matters may be further simplified by assum-
ing that production functions concerning X 2 are similar inter-
nationally but those concerning X1 differ. This situation is then
equivalent to the one where technical progress in X1 has occurred
in one country alone. More specifically, suppose that technology in
the foreign country is unchanged, but that Hicks-neutral technical
improvement occurs in X1 in the home country. Under these
circumstances we can show that the home country will export X1
56 Studies in the Pure Theory of International Trade
x,

Figure 3.4

and import X 2 from the foreign country. This can be established


readily from an examination of Fig. 3.4, where FF' is the trans-
formation curve in the foreign country, U1 is its highest attainable
community indifference curve under autarky and S1 is its self-
sufficiency equilibrium point, with the autarky price ratio given by
the slope of AB. If the home country were identical to the foreign
country in all respects, it would possess the same factor endow-
ments, the same technology in all industries, the same transforma-
tion curve, the same community indifference curves and the same
autarky price ratio. There would then be no basis for trade.
Instead, assume that in the home country X1 alone comes to enjoy
a Hicks-neutral technical improvement, so that its transformation
curve shifts to HF', which lies outside of FF' at all production
points except F' where no X1 is produced. At the same commodity-
price ratio given by the slope of DE, which is parallel to AB, the
production point in the home country is given by P. Now if the
consumption pattern is the same internationally, then in both
countries the two goods are consumed in the same proportion at the
same commodity-price ratio. This, of course, implies that community
preferences in the two countries are not only identical, but that they
The Basis of International Trade 57

are also homothetic.t Under these circumstances, the consumption


point in the home country would be given by C, if the commodity
prices were to remain unchanged. But then the home economy
would not be in self-sufficiency equilibrium, for C is not identical
with P. In other words, under the postulated conditions and the
absence of trade, the commodity-price ratio in the home country
cannot be the same as that in the foreign country. Since at the price
ratio given by the slope of DE the home country produces
relatively less of X 2 than it consumes, the relative price of the
second commodity must rise to bring about the demand-supply
equilibrium. This also implies that the self-sufficiency equilibrium
in the home country must lie on HF' to the right of P but some-
where to the north-west of C. In Fig. 3.4 the home self-sufficiency
equilibrium point is given by Sh and the home autarky price ratio is
given by the slope of RS, which, since it is steeper than AB, shows
that
Ph> P1
which means that the home country will export X1 and import X2
from the foreign country. It is also clear that international trade
occurs only because of international differences in production
functions.
Now the kind of Hicks-neutral efficiency differences in inter-
national production functions assumed in our analysis ensures that
the comparative productivity of both factors in the first commodity
is higher in the home country than in the foreign country. The
Richardian theorem may then be restated for the case of a two-
factor economy. A country exports the commodity which possesses a
higher comparative productivity of all factors and imports the
commodity where the comparative productivity of both factors is
lower than that in the other country.
Until now we have assumed that community preferences are
homothetic and similar internationally. What if the homothety
assumption is dropped? If the two countries have different levels of
income, then non-homothety of preference may seriously com-
promise the validity of the Richardian theorem even if community
preferences in the two countries are similar. This possibility will

t It was not necessary to make this assumption in the simple Ricardian formula-
tion because factor supply and demand conditions played no role in the determina-
tion of the autarky price ratio.
58 Studies in the Pure Theory of International Trade

not be pursued here, but its significance will become clear later when
we examine the validity of the Heckscher-Ohlin theorem.

3.4 The Heckscher-Ohlin Theory


The theory explaining the basis of international trade which has
earned literally universal acceptability among modern trade theorists
is popularly known as the Heckscher-Ohlin theory. Associated
with the names of two distinguished writers, Heckscher [3] and
Ohlin [10], this theory asserts that a country exports the commodity
which uses intensively its relatively abundant factor and imports the
commodity which is intensive in the use of its relatively scarce factor.
Two definitions of factor scarcity (or abundance), namely, the price
and the physical definition, have gained wide currency in the
literature on trade theory. According to the price definition due
initially to Heckscher and Ohlin, country H is capital-abundant
(or labour-scarce) relative to country F if
(3.7)
where it may be reminded that w denotes the wage/rental ratio.
According to the physical definition due initially to Leontief [7],
H is capital-abundant relative to F if
kh > kf (3.8)
where k = K/ L. If the Heckscher-Ohlin dictum is true, then the
capital-rich country H will export the capital-intensive commodity
and import the labour-intensive commodity from the labour-rich
country F. For the full validity of the Heckscher-Ohlin (henceforth
called H.O.) theory, it is necessary to assume that production
functions for each commodity are different within a country (so
that factor proportions differ in each industry) but similar inter-
nationally. We _also assume that factors are perfectly mobile
internally but completely immobile externally.

The Price Definition


The proof of the H.O. theorem in terms of the price definition is
rather simple and requires the use of the one-to-one relationship,
established in the previous chapter, that exists between p and w in
the absence of factor-intensity reversals. Consider Fig. 3.5, which
reproduces parts of Figs. 2.2 and 2.3 presented in the previous
chapter. The first quadrant of Fig. 3.5 portrays a negative
The Basis of International Trade 59

R'
R

R'
R
p 0 p

Figure 3.5

relationship between p and w in terms of the R' R' curve under the
condition that k 2 > k 1 at all w, whereas the second quadrant
depicts the positive relationship between p and w in terms of the
RR curve drawn for the case where k 2 < k 1 at all w. Owing to the
international similarity of production functions, RR and R' R'
curves describe the nature of the relationship between p and w in
both countries. Owh exceeds Ow1 to reflect the inter-country factor-
endowment relationship expressed in (3.7), so that if
k 2 < k1, then Ph > p1 (3.9)
which implies that the home country will export the capital-
intensive commodity X1 and import the labour-intensive commodity
x2 from the foreign country, and if
(3.10)
and the home country will export x2 and import xl, but x2 is now
capital-intensive relative to X1 . Thus, whatever the factor-intensity
relationship between the two commodities, the relatively capital-
abundant country exports the relatively capital-intensive commodity
and imports the relatively labour-intensive commodity from the
60 Studies in the Pure Theory of International Trade

relatively labour-abundant country. It may now be observed that


the converse of the theorem is also valid. A country's relatively
abundant factor is the one utilised relatively intensively by its export-
able good.
What if factor intensities are reversible? The problem is studied
in terms of Fig. 3.6, where the relationship between p and w is given

0 ph pf ph p

Figure 3.6

by RaR', a being the factor-intensity reversal point. The Ra


portion of the curve indicates that k 2 < k 1 , whereas the R' a
portion shows that k 2 > k 1 over the corresponding ranges of w.
If the inter-country factor-abundance relationship is given by wh
and wi, we are operating in that portion of the RR' curve where
k 2 < k 1 and
Ph> PI
which is the same as (3.9), so that the H.O. theorem still holds. On
the other hand, suppose that the autarky w in the home country is
given by w~, then we find that for the home country k 2 > k 1 , but
for the foreign country k 2 < k 1 . Furthermore,
p~ <PI·
The Basis of International Trade 61

It is evident that the H.O. theorem continues to hold in the case of


the home country, but not in the case of the foreign country, which,
despite being a relatively labour-abundant country, exports its
capital-intensive commodity X1 . From this discussion we may con-
clude that if the autarky wage/rental ratios of the two countries lie
on one side of the factor-intensity reversal point, the H.O. theorem
is valid; but if they lie on different sides, at least one of the countries
violates the H.O. dictum. The H.O. logic is demolished because of
the fact that the same commodity is intensive in the use of different
factors in the two countries.

The Role of Demand


One may be tempted to conclude that the H.O. theorem is valid in
the absence of factor-intensity reversals. At least, this was the
popular belief until the recent appearance of two studies, one by
Bhagwati [ 1] and the other by lnada [ 4]. Both have shown that
the validity of the H.O. theorem in terms of the price definition
may be compromised if suitable restrictions on community
preferences are not placed. We have already shown in the context
of the Ricardian model that if the community indifference curves
are not well behaved, there may be multiple self-sufficiency
equilibria which may preclude the possibility of trade even if
production functions were non-identical internationally. However,
in the H.O. two-factor model where the transformation curve is
concave to the origin, the existence of intersecting community
indifference curves may not only prevent the occurrence of inter-
national trade in spite of inter-country disparities in factor
endowments, but may also lead to a pattern of trade contradicting
the H.O. theorem.
Consider Fig. 3.7 for a possible demonstration of these results.
Suppose factor intensities are non-reversible and k 2 > k 1 , so that
from (3.10)p1 >ph; suppose further thatp' denotes the free trade
terms of trade such that p1 > p' >Ph· If the H.O. theorem is to
hold, the home country should export X 2 and import X1 . Assume
that the self-sufficiency equilibrium point for the home country is
given by Shin Fig. 3.7 and the autarky price ratio is furnished by
the line PhPh• which is tangential to both the home transformation
curve HH' and its community indifference curve U1 U1 at point Sh.
In the free trade situation, the production point shifts to P where
the p'p' line, whose slope displays the international terms of trade,
62 Studies in the Pure Theory of International Trade
x,

Figure 3.7

is tangential to the transformation curve. If X 2 is to be exported,


the new consumption point should lie on p'p' but to the left of the
production point P. Such a consumption point is given by C', but
note that the corresponding community indifference curve U' U'
does not intersect U 1 U 1 • However, if the community indifference
map is not as well behaved as the one just depicted, the new con-
sumption point could lie anywhere on p'p'. If it lies on P, which
requires that U 2 U 2 be tangential to the transformation curve at
P, there will be no trade, a case illustrated by Bhagwati [1]; if
it lies on p'p' to the right of P, say on C where the line p'p' is
tangential to U3 U3 , the pattern of trade runs contradictory to the
H.O. hypothesis, a result pointed out by Inada [ 4]. With point C,
for example, the capital-rich home country exports the labour-
intensive commodity X, and imports the capital-intensive com-
modity X 2 from the labour-rich foreign country. Thus some
restrictions on demand conditions are needed even if the H.O.
theorem is to be proved in terms of the price definition of relative
factor abundance. It is sufficient to assume that the community
preferences are well behaved, or,· what is the same thing, that each
The Basis of International Trade 63

country behaves like a single rational consumer. This restriction


eliminates the possibility of multiple equilibria and, with it,
ensures the validity of the H.O. theorem.t

The Physical Definition


The proof of the H.O. theorem in terms of the physical definition
of relative factor abundance is slightly more involved, for here, in
addition to the international identity of production functions, it is
necessary to assume the international similarity of consumption
patterns. At the same time, one may regard the physical definition
as the only acceptable definition of inter-country factor abundance.
It may be argued that proving the H.O. theorem in terms of the
price definition borders on truism because the autarky factor
prices themselves, when the theorem is valid, are different because
commodity prices are different. Furthermore, the starting-point
should not begin with factor prices because the latter are determined
through a complicated interaction of many economic forces, like
demand and supply, and not by physical factor endowments, alone,
so that the price and the physical definition may run counter to each
under some circumstances. A more objective definition, it appears,
would take into account the availability of factor supplies in
physical amounts.
At the outset, we assume that community preferences inter-
nationally are not only similar but also homothetic, so that in each
country the income elasticity of demand for each commodity ('1i) is
identically equal to unity. Under these conditions, there exists a
unique relation between the overall K/L ratio (k) and the
commodity-price ratio (p), as established in the previous chapter.
If factor-intensity reversals are ruled out, this relationship is
depicted in Fig. 3.8, which replicates Figs. 2.6 and 2.7 from the
previous chapter. The first quadrant of Fig. 3.8 depicts the case
where k 2 > k 1 and the relation between k and p in both countries
is given by the negatively inclined curve SS; the second quadrant
on the other hand represents the case where k 2 < k 1 so that the

t Some of the autarky equilibria will be unstable. However, the absence of


unstable autarky equilibria is not sufficient to ensure the general validity of the
H.O. theorem. Since income distribution in the H.O. model varies with the
variations in commodity prices, there are numerous autarky equilibria, not all of
which are unstable. In particular, there may be more than one stable autarky
equilibrium. For further details on these points, see lnada [ 4] and Kemp (6].
64 Studies in the Pure Theory of International Trade

s
S'

kh

kf

s
S'
p ph Pf 0 P'h pi p
f
Figure 3.8

relationship between k and p is given by the positively inclined


curve S' S'; kh is placed above k 1 to show that the home country is
capital-abundant relative to the foreign country. In order to prove
the H.O. theorem, we wish to derive relations (3.9) and (3.10) from
the physical definition described by the relation (3.8). The autarky
price configurations corresponding to this factor-abundance speci-
fication are given by
(1) k 2 < k 1 and Ph > p 1
and (2) k 2 > k 1 and p;, < P[
which accord exactly with expressions (3.9) and (3.10), respectively.
This proves the H.O. theorem in terms of the physical definition.
An alternative proof of the H.O. theorem in terms of the
physical definition turns out to be more rewarding. We can
establish the theorem in exactly the same manner as we demonstrated
the validity of the Ricardian theorem in terms of Fig. 3.4. Consider
Fig. 3.9, where FF' is the foreign transformation curve, U1 is its
community indifference curve, S 1 is its self-sufficiency equilibrium
point and AB reflects its autarky price ratio. Suppose now that
The Basis of International Trade 65

x,

Figure 3.9

there is an increase in the supply of capital alone and that k 1 > k 2 •


The transformation curve then shifts out to HH' which lies outside
FF' at all production points, showing that greater output of one or
both goods can be produced with the help of increased factor
supply. However, since X1 is capital-intensive relative to X 2 , and
the factor with the expanded supply is capital, the outward shift of
the transformation curve is biased towards the capital-intensive
good. Let us now suppose that HH' belongs to the home country
which has been assumed to be relatively capital-abundant. At the
price ratio given by the slope of DE (parallel to AB) the home
production point is given by P, but its consumption point is given
by C. Proceeding in the same manner as we did while discussing
Fig. 3.4, we can argue that the home self-sufficiency equilibrium
will be given by a point like Sh, with an autarky price ratio given by
the slope of, say, RS. Since RS is steeper than AB,
Ph> PI
which in tum ensures that the home country will export X1 and
import X2 in accordance with the dictates of the H.O. theorem.
66 Studies in the Pure Theory of International Trade

The presence of factor-intensity reversals is again destructive of


the validity of the Heckscher-Ohlin theorem. Assume that, because
of this reversal, k 1 > k 2 in the foreign country, but in the home
country k 1 < k 2 . Then the home transformation curve drawn for
the case of higher supply of capital will be biased towards the
second commodity, as is depicted in Fig. 3.10. The comparison of
the autarky price ratios in this case suggests that p 1 > ph, so that
the foreign country will export X1 and the home country will export
X 2 . Clearly, then, the foreign country is violating the H.O.
theorem.
x,
E

Figure 3.10

The Role of Demand


Demand conditions assume the most significant role in the deriva-
tion of the H.O. theorem in terms of the physical definition. We
have already seen that international similarity of taste patterns is
crucial to the validity of the theorem. To see the significance of this
assumption, all we have to do is to conceive that each country (or
one country alone) has a strong preference for the consumption of
the commodity that is intensive in the use of its relatively abundant
The Basis of International Trade 67

factor. Going back to Fig. 3.9, let us suppose that the home con-
sumption pattern is strongly biased towards the first good. Then at
the price ratio reflected by DE (and AB), the home consumption
point would lie on DE to the left of C; but so long as this consump-
tion point lies between P and C, it is easy to see that the point Sh
would lie on HH' to the right of P, so that the H.O. theorem would
continue to hold. However, if the home consumption point at the
commodity-price ratio reflected by DE lies to the left of P, the point
Sh would lie on HH' to the left of P, and the relationship between
the autarky price ratios would be reversed, so that the H.O. theorem
would not hold. Thus, in the absence of international similarity of
consumption patterns, the H.O. theorem may not be valid even if
all other requirements are satisfied. As argued before, for the inter-
country similarity of consumption patterns it is not sufficient that
the community preferences be identical; it is also necessary that in
each country they be homothetic. The significance of this de-
sideratum in the case where national incomes differ has been
demonstrated by Romney Robinson [12]. The tenor of his
argument is quite simple. If the community preference map of each
country is identical but not homothetic, the income elasticities of
demand for the goods differ, but the countries will still consume the
two goods in the same proportion at a given commodity-price
ratio if the national incomes are the same everywhere. Otherwise
the proportion in which the two goods are consumed by each
country at any price ratio will differ even if their community
indifference curves are non-intersecting; and this, as we have seen
above, could seriously compromise the validity of the Heckscher-
Ohlin theorem.

3.5 A Comparison
In the last two sections we have examined in detail the Ricardian
and the Heckscher-Ohlin theories explaining a country's pattern of
trade. This places us in a strong position to combine and collate
the various issues involved in the establishment of the two
theorems.
Perhaps the only similarity between the two theories consists in
the obvious fact that both tend to explain the basis of trade in terms
of international disparities in comparative costs. Otherwise the
two theories are poles apart, because of the following points of
difference :
68 Studies in the Pure Theory of International Trade

1. The Ricardian theory relies on international differences in


production functions in order to explain the causality of
international trade, whereas the H.O. theorem explicitly
assumes the international similarity of production functions.
2. The original Ricardian formulation assumed the existence of a
single factor of production, and this along with the presence of
constant returns to scale ensured the constancy of unit costs
along the transformation curve. By contrast, the H.O.
theorem postulates two productive factors, and this along
with the inter-commodity dissimilarity of production functions
gives rise to increasing unit costs along the transformation
curve. Another implication of this difference is that demand
conditions play a much more important role in the H.O.
logic, expounded in terms of any definition of relative factor
abundance, and as a consequence necessitate much more
stringent restrictions than those needed in the Ricardian
world.
3. As a result of the difference enumerated above, the introduc-
tion of trade in the Ricardian model leads the trading
partners to complete specialisation, whereas in the H.O.
framework trade may or may not result in complete specialisa-
tion, as will be clear from a glance at Fig. 3.7.
These, then, are some of the substantive differences, as far as the
positive aspects are concerned, that exist between the Ricardian
and the H.O. simulations of the world economy. These points
perhaps conceal some of the similarities that can be shown to exist
when the two theorems are applied to the 'normative' analysis of
international trade. For instance, in terms of both frameworks it is
possible to show that free trade is beneficial. A detailed exploration
of these points is, however, beyond the domain of this chapter.

3.6 Equilibrium in International Trade


The next important question, after a country's pattern of trade has
been determined, concerns the characteristics of equilibrium in
international trade. Here we assume that there is no interference in
the free inter-country flow of commodities and that returns to scale
are constant. Equilibrium in international trade simply signifies a
situation where the value of a country's exports equals the value of
its imports and the commodity prices thus obtained furnish the
The Basis of International Trade 69

equilibrium terms of trade. In order to facilitate comprehension


of how this equilibrium is reached, it is necessary to derive what is
well known as a country's offer curve, which may be defined simply
as the locus in two-dimensional space of various quantities of a
commodity offered by a country in exchange for the varying offers
of another commodity by its trading partner. Because it simul-
taneously indicates the export supply that reciprocates for the
import demand, the offer curve is also called the reciprocal demand
curve.

Figure 3.11

The derivation of a country's offer curve proceeds as follows.


Consider Fig. 3.11, where the home country's exports of the first
commodity, given by £ 1 = X 1 -D 1 , are measured along the
abscissa and its imports of the second commodity, given by
£ 2 = D2 - X2 , are measured along the ordinate. The slope of any
ray through the origin represents the ratio between the international
prices of exportables and importables. For example, the slope of
OA furnishes, from the home country's viewpoint, a certain ratio
between the export price and the import price, and as the ray from
the origin becomes steeper, the terms of trade become more
favourable to the home country. Now at a certain price ratio the
70 Studies in the Pure Theory of International Trade

home country is in self-sufficiency equilibrium, that is to say, its


consumption and production points lie identically on its trans-
formation curve. Suppose such a price ratio is indicated by the ray
0 A; clearly, the starting-point of the offer curve is the origin, where
no trade occurs. At another price ratio, reflected by the slope of
OB, the home country produces more but consumes less of the first
commodity and at the same time produces less and consumes more
of the second commodity. Thus at this new price ratio the home
country faces excess supply of the first commodity and excess
demand for the second commodity. The demand-supply equi-
librium requires the export of the first commodity equal to its
excess supply and import of the second commodity equal to its excess
demand. Suppose, then, the quantity of exports and imports
accompanying the terms of trade given by OB is represented by
point G. At other levels of terms of trade exhibited respectively by
the slopes of OC and 0 D these points are given by Q and R. The
locus of all such 'exchange' points is OH, which depicts the home
country's offer curve.
The shape of the offer curve is determined by the transformation
curve and the community's preference function. The export supply
is positively related to its relative price, but only up to a point such
as Q, because export supply is an excess supply, the difference
between domestic production and domestic demand. The response
of the demand for exportables to improvements in the terms of
trade (which is the same thing as the rise in the relative price of the
exportable good) can be split into a pure substitution effect, which
is always negative, and an income effect, which may be positive or
negative. In the case of non-inferior goods the rise in income
resulting from an improvement in the terms of trade tends to raise
the domestic demand for the exportable good, and if this effect is
very pronounced, the export supply could decline as a result of the
rise in its relative price. Such a possibility is depicted by the RQ
segment of the offer curve. In the extreme case it is possible that
the offer curve may bend backwards, as shown by the HR segment
of the offer curve. But this would in tum imply that the imported
commodity is a Giffen good, for as a result of a decline in its
relative price the home consumers would be consuming less of it.
A question related to the shape of the offer curve concerns the
elasticity of demand for imports by the home country (ah), where
by definition ah = -(dE2 /dp)(p/E2 ) and where p = p 2 /p 1 = Etf£2 ,
The Basis of International Trade 71

so thatt

- [dE /d(E /E )] (£ /£ 2 ) = - £ 1 d£2


2 I 2 I 2 E2dEI - El d£2
-dE2/dE1

after differentiation. Now dE2/dE1 is the slope of the offer curve at


any point, say G, and equals GN/TN, and (£2 /£1 ) = GN/ON, so
that ah = ON/OT > 1, because, as may be observed, the offer
curve has a positive slope at G. Similarly, at Q where the offer curve
has just begun to change its slope, ah = 1, and at R where the offer
curve is negatively inclined, ah = OF/OM < 1.
In a similar fashion, one can construct the offer curve of the
foreign country. The volume and the terms of trade in equilibrium
are determined by the point of intersection between the two
countries' offer curves, OH and OF, as depicted in Fig. 3.12 where
the equilibrium terms of trade are indicated by the slope of OE,
and where the home country exports OQ amount of the first
commodity in exchange for EQ amount of the second commodity
from the foreign country.

Stability of Equilibrium
How is one to ensure that the international trade equilibrium
schematised in Fig. 3.12 is stable? For example, if the home offer
curve was given by OH' instead of OH, there would be three points
of intersection and hence three possible equilibria, E, E' and £".
However, it can be shown that equilibrium at E is unstable if the
offer curves are given by 0 F and 0 H', but stable if the home offer
curve is given by 0 H. The demonstration consists in differentiating
the balance-of-payments (B) equation
B=E1 -pE2

t Since each point on the offer curve represents by definition a certain level of
export supply or import demand, at every point p 2 E2 = p 1E 1 • However, the latter
equality should not be misconstrued as the equilibrium condition at this point. It
merely suggests what a country is willing to import in return for some exports at
certain terms of trade. The equality becomes an equilibrium condition only when
one country's import demand is exactly matched by the other country's willingness
to meet that demand at certain terms of trade.
72 Studies in the Pure Theory of International Trade
H

E,
Figure 3.12

with respect top and in observing that the stability of equilibrium


requires that an improvement in a country's terms of trade, which
is equivalent to a decline (rise) in the relative price of importables
(exportables) in the foreign trade market, must lead to a deficit in
its balance of payments, that is, dBjdp > 0. Carrying out the
differentiation of B with respect to p yields
dB dE1 dE2
- =--p--E2.
dp dp dp
If, initially, trade is in balance, so that B = 0,

-dB = E2 [ -p
dp El
·dE
-1- -p -dE2
dp E2 dp
-1 J= E2 (a 1 +ah-1)

where E2 > 0, and a1 and ah are respectively the foreign and home
elasticities of demand for imports. Evidently, stability is ensured if
a1 + ah > 1, that is, if the sum of the two elasticities of import
demand exceeds unity.t Reverting to Fig. 3.12, it is now easy to see

t Quite frequently, this condition is referred to as the Marshall- Lerner condition


for stability.
The Basis of International Trade 73

why E is. unstable and E' and E" stable when the two offer curves
are given by OF and OH'. For example, with E', if the terms of
trade shift in favour of the home country from OE' to OZ, its
balance of payments runs into a deficit of D 1D 2 in terms of the
imported commodity, which in turn means that dB/dp > 0. How-
ever, with E, a similar favourable shift in the home country's terms
of trade to 0 W brings about a balance-of-payments surplus of
D 1 D 3 , so that dB/dp < 0. Here, then, the equilibrium is unstable.
The stability of E" can be established analogously. This analysis
simply echoes Marshall's view [8] that an unstable equilibrium
point must be flanked by points of stable equilibria, and that the
number of equilibrium points must be odd. It follows then that a
unique (single intersection) equilibrium must be stable.
It is possible to derive the stability condition in terms of only one
commodity and in the process obtain more information. Since the
export of a commodity by one country must in equilibrium equal
the import demand of the other country,
E1 =and E 2 = E 1f
Elf
where Elf and E1f are respectively the quantities imported and
exported by the foreign country. The balance of payments in terms
of one commodity can be written as
B =E 1 -Elf= (X1 -D 1 )-(Dlf-Xlf)
= (X1 +X1f)-(D 1 +Dlf) = x 1 -d1
where x 1 and d 1 are respectively the world supply and demand for
the first commodity. Any discrepancy between x 1 and d 1 causes
disequilibrium and sets into motion changes in the terms of trade.
We are interested in finding conditions which cure this situation of
disequilibrium. Stability in terms of the world market of the first
commodity requires that dB/dp < 0, that is, a rise in the world
relative price of the first commodity causes an excess supply, or a
rise in B, and conversely. Differentiating B and remembering that
with initial trade equilibrium x 1 = d 1 , we get
dB x1
-d = -(bl +c~)-
'P p
where b 1 = -(p/x 1 )(dxddp) and c 1 = (p/d 1 )(ddddp) are respect-
ively the elasticities of world supply and demand for the first
commodity. Similarly,
B = (D 2 +D 2f)-(X2 +X2f) = -(x 2 -d2 )
74 Studies in the Pure Theory of International Trade

and
dB x2
dp -(b 2 +c 2 ) -
p
where b 2 = (p/x 2 )(dx 2 /dp) and c 2 = -(p/d2 )(dd2 /dp). Stability
conditions expressed in terms of the second commodity again
require that dB/dp < 0, that is, a fall in the world relative price of
the second commodity must lead to a situation of excess demand,
and vice versa. It may be observed that, in the absence of decreasing
costs and Giffen goods, dB/dp < 0 in terms of both markets.
However, if either the opportunity costs are decreasing or the goods
are of the Giffen variety, the equilibrium may be unstable. It is
interesting to observe that, in the presence of positively sloped
supply curves, the necessary condition for instability is that the
commodities be Giffen goods. Unless otherwise specified, the
equilibrium point will be assumed to be stable.

Equilibrium in the Ricardian World


In the Ricardian simulation of the world where unit costs are
constant, the mechanism of attaining equilibrium and the stability
conditions remain the same, except that the shapes of the offer
curves differ. Instead of the smooth curves depicted in Figs. 3.11
and 3 .12, the offer curves in the Ricardian economy contain a kink
at the point of complete specialisation. This is pictured in Fig. 3.13,
where the home and the foreign offer curves exhibit kinks at K and

Figure 3.13
The Basis of International Trade 75

K' respectively. The existence of these kinks is attributable to the


fact that, until the point of complete specialisation, each country
can satisfy the additional demand for its exportables by just
shifting its production point without any increase in the price of
the exportable good. However, after complete specialisation has
been attained, further increases in export surplus call forth a
decrease in local demand and hence a rise in the price of the
exportable good. In other words, after the attainment of complete
specialisation, the offer curves assume the shape and properties of
those depicted before, so that the equilibrium is reached at E, the
point of intersection between the two offer curves.

Factor-Price Equalisation
One striking property of the Heckscher-Ohlin model is that under
certain conditions factor prices in the free trade equilibrium get
completely equalised in the trading countries. Despite undertones
of unrealism, the so-called factor-price equalisation theorem has
held remarkable fascination with trade theorists, if only because
the theorem is in fact valid under some admittedly stringent con-
ditions. Although the controversy over the theorem has raged for
more than twenty years, the underlying logic of the theorem is
very simple.
We have already shown that in a two-commodity, two-factor,
constant returns to scale model, there is a unique relationship
between the commodity-price ratio and the wage/rental ratio,
provided factor intensities are non-reversible and both commodities
are produced. Now under free trade and the absence of transport
costs there prevails only one commodity-price ratio in the trading
countries, so that if production functions are internationally
similar, there will be only one wage/rental ratio (ro) prevailing in
both countries, which in turn implies that each commodity will
utilise the same capital/labour ratio everywhere. It follows then
that the marginal product and hence real reward of each factor
must under free trade be the same in both countries.
It is thus seen that the underlying logic of the factor-price
equilisation theorem is very simple. Yet the long controversy that
this theorem has sparked belies this impression. There are two
issues here, which, in view of the earlier debate, deserve examination
in detail. What are the roles played by (i) incomplete specialisation
and (ii) the non-reversibility of factor intensities in the attainment
76 Studies in the Pure Theory of International Trade
K

Figure 3.14

Figure 3.15

of international factor-price equalisation? A satisfactory answer


to this question requires recourse to Figs. 3.14 and 3.15.
The Basis of International Trade 77

Consider any equilibrium international price ratio. By the


normalisation rule, such units of traded quantities can be selected
as yield the unit price ratio. If at this unit price ratio both
commodities are to be produced in a country, then two conditions
must simultaneously be satisfied. First, there must exist an isocost
line tangential to the unit isoquants of both commodities,t and
second, the overall capital/labour ratio must lie between the
capital/labour ratios of each commodity.t All this is depicted in
Fig. 3.14, where the isocost line AB is tangential to the unit
isoquants X1 and X 2 at C and D respectively, and where OH is the
endowment ray, the slope of which indicates the overall capital/
labour ratio in the home country. Rays OC and OD form what
Chipman [2] calls the 'cone of diversification', which in the
diagram is given by the cone COD. Thus if the endowment ray OH
lies within the cone of diversification, as it clearly does in Fig. 3.14,
then at the wage/rental ratio reflected by the slope of AB both
commodities are produced and both factors can be shown to be
fully employed.
Let us now introduce the endowment ray of the labour-abundant
foreign country, OF. If OF lies within the cone of diversification,
there will be incomplete specialisation in both countries, the cone
of diversification will coincide, and factor prices will be the same.
If OF lies on OD, factor prices will still be similar internationally,
although the foreign country will be specialised completely in the
production of the labour-intensive commodity X 2 • However, if OF
lies outside the cone COD, as it does in the diagram, not only will
the foreign country be completely specialised in X 2 , but factor
rewards will also be dissimilar. This is because the wage/rental
ratio required to effect full employment in the foreign country will
be given by the slope of the isocost line RS, which is different from
the wage/rental ratio prevailing in the home country.
What if the relationship between p and w is not unique, that is,
what if the factor intensities are reversible as in Fig. 3.15, where,
because of multiple intersections of the isoquants, two isocost lines
are shown to be tangential to the unit isoquants, giving rise to two

t For the sake of illustration, suppose that the isocost line is tangential only to one
unit isoquant and that the other unit isoquant lies above it. Evidently, then, only
one commodity will be produced, because the unit cost of the other commodity
exceeds its price.
t This was established in section 2.2 of the previous chapter.
78 Studies in the Pure Theory of International Trade

cones of diversification, COD and TOU. Clearly, then, if both


the endowment rays lie in one of the two cones, factor-price
equalisation is inevitable. However, if they lie in different cones,
equalisation is impossible in spite of the prevalence of incomplete
specialisation in both countries. With OH flanked by COD and OF
by TOU, one can see that the isocost lines in the two countries
possess different slopes and hence reflect unequal wage/rental
ratios. Thus we conclude that the necessary and sufficient condition
for the validity of the factor-price equalisation theorem, among others,
is that the two endowment rays lie within or coincide with the same
cone of diversification, irrespective of whether or not the factor
intensities are reversible. In other words, the unique relationship
between p and w is neither necessary nor sufficient for factor-price
equalisation; nor is incomplete specialisation in any country a
necessary condition.t
What is the effect of international trade on factor rewards in the
event of the violation of this condition? There is no categorical
answer to this question, except in the case where factor intensities
are not reversible and the autarky equilibria are unique, which is,
of course, one of the sufficient conditions for the validity of the
Heckscher-Ohlin theorem. The introduction of trade then at least
moves the international factor prices closer together than before
the opening of the trade. If factor intensities are reversible, this
effect is indeterminate. As shown by Johnson [5], the answer is
determined in part by the number of intersections between the unit
isoquants, whether two, three, or more and whether this number is
odd or even.

Arbitrary Number of Commodities and Factors


A good deal of the literature in trade theory is concerned with the
validity of the factor-price equalisation theorem in the realistic
case where there are more than two commodities and two factors.
An illuminating discussion of this issue has been provided by
Chipman [2] and Kemp [6] among many others. It turns out that
if the other sufficient conditions upholding the Heckscher-Ohlin
theorem are maintained, the general rule for equalisation established

t Suppose the endowment ray of the home country coincides with OC and that
of the foreign country coincides with ODin Figs. 3.14 or 3.15. Then both countries
will be completely specialised but factor-price equalisation will still prevail, for the
wage/rental ratio in each country will be given by the slope of AB.
The Basis of International Trade 79
above applies regardless of the number of commodities and factors.
Suppose that there are r factors and n commodities, something of
which is produced in the world, though not necessarily in each
country. Given a vector of n world equilibrium commodity prices,
what are the conditions which will ensure that each of r factors of
production receives the same reward in each country?
As before, given the price vector, commodity units can be so
chosen that all price ratios will be equal to unity. Corresponding
to these unit product-price ratios, we can define n unit isoproduct
surfaces, which may or may not possess a common tangent isocost
plane. If a common tangent plane does not exist, factor-price
equalisation, as before, is impossible. If such a plane exists,
equalisation is possible but not necessary. For' factor prices to be
equalised internationally, it is necessary that the endowment ray of
each country lie within the cone of diversification specified by n
points of tangency. For otherwise the relative factor prices
associated with the common tangent plane may not be consistent
with the full employment of all factors.
What about the uniqueness of the common tangent plane? In
the two-by-two case, this uniqueness implies the non-reversal of
factor intensities. In the n-by-r case, where it may be extremely
difficult, if not impossible, to define relative factor intensities, one
of the necessary conditions for uniqueness is that n = r. If n =I= r,
uniqueness is impossible, for then the number of equations does not
match the number of unknowns. However, we have shown in the
two-by-two case that uniqueness is not a necessary condition for
the international equalisation of factor rewards. This is also true
with the n-by-r case. All that is necessary and sufficient to ensure the
equalisation of factor prices in the two countries is that each
country's endowment ray be flanked by the cone of diversification
defined by the points of tangency between the common tangent isocost
plane and the unit isoproduct surfaces. This rule applies to all cases
defined by n ~ r.

REFERENCES
[1] Bhagwati, J., 'The Proofs of the Theorems on Comparative Advantage',
Economic Journal, LXXVII (Mar 1967) 75-83.
[2] Chipman, J. S., 'A Survey of the Theory of International Trade: Part 3, The
Modem Theory', Econometrica, XXXIV (Jan 1966) 18-76.
80 Studies in the Pure Theory of International Trade
[3] Heckscher, E. F., 'The Effect of Foreign Trade on the Distribution of Income',
in Readings in the Theory of International Trade, ed. H. S. Ellis and L.A. Metzler
(Philadelphia: Blakiston, 1949).
[4] lnada, K., 'A Note on the Heckscher-Ohlin Theorem', Economic Record, XLIII
(Mar 1967) 88-96.
[5] Johnson H. G., 'Factor Endowments, International Trade and Factor Prices',
Manchester School of Economic and Social Studies, xxv (Sep 1957) 270-83.
[ 6] Kemp, M. C., The Pure Theory of International Trade (Englewood Cliffs, N.J. :
Prentice-Hall, 1964).
[7] Leontief, W. W., 'Domestic Production and Foreign Trade: The American
Capital Position Re-examined', Proceedings of the American Philosophical
Society, XCVII (Sep 1953) 332-49.
[8] Marshall, A., The Pure Theory of Foreign Trade (London: London School of
Economics and Political Science, 1949).
[9] Mill, J. S., Principles of Political Economy with Some of their Applications to
Social Philosophy, 3rd ed. (London: Parker & Co., 1852).
[ 10] Ohlin, B., Interregional and International Trade (Cambridge, Mass.: Harvard
U.P., 1933).
[II] Ricardo, D., On the Principles of Political Economy and Taxation (London: John
Murray, 1817).
[12] Robinson, R., 'Factor Proportions and Comparative Advantage', Quarterly
Journal of Economics, LXX (May 1956) 169-92.
4 Gains from Trade

The theory of gains from trade is almost as old as economics itself.


Early contributions in trade theory were motivated by the desire to
demonstrate the speciousness of protectionist policies advocated
by the Mercantilist School. Adam Smith was among the first to
point out the beneficial effects of international trade because it
provided expanded opportunities for 'division of labour', or for
what the modern writers call 'specialisation'.
In this chapter we are interested in a rigorous demonstration and
evaluation of some of the welfare propositions that have been
advanced by various economists under the standard assumptions
which are necessary to ensure the existence of the Paretian optimum,
at least within each trading country. Such cases where Pareto
optimality conditions cannot be satisfied will be examined in sub-
sequent chapters.
The first prerequisite in any discussion of welfare propositions
calls for a prior agreement on the welfare criterion to be used in
evaluating the merits of the suggested policy devices. Unfortunately,
here we have to traverse very slippery grounds because of the nature
of the analysis itself. For the introduction of trade is bound to cause
a change in commodity prices, and, from the Stolper-Samuelson
theorem established in Chapter 2, we know that some factors will
lose and some will gain. On what basis then can we say that the
community as a whole is better off in one situation than in the other?
Furthermore, what type of income distribution is desirable and,
maybe, socially acceptable?
The discussion on this question in welfare economics is volumi-
nous, and we shall not explore its subtleties and intricacies, if only
because no satisfactory technique commanding universal concur-
rence has as yet been devised. Some progress, of course, has been
achieved by the discovery of what is now familiar as the compensa-
tion principle. According to this principle, a policy change is
82 Studies in the Pure Theory of International Trade

recommendable if, in the new situation, gainers can more than com-
pensate the losers, so that everyone can be better off than before.
The question of whether the compensation should actually take
place is an ethical one for which the economist has no pronounce-
ment. The important point is that some 'desired' type of income
distribution can always be maintained if the state is ready to adopt
a policy of lump-sum transfers. In this spirit we assume throughout
our analysis of the gains from trade the existence of a community
welfare function which is a function of the quantities of various
products consumed by the community and which possesses proper-
ties that are parallel to individual utility functions. t Under this
approach the desired level of income distribution becomes inherent
in the choice of the utility function. Most of the assumptions made
in Chapter 2 are retained, except that, in the interest of generality,
we assume any arbitrary number of commodities and factors of
production. Thus production functions are still assumed to be
homogeneous of the first degree, all goods are non-inferior and fac-
tors are fully employed and inelastically supplied. The supply curves
for all commodities are upward-sloping; all markets are perfect and
factors of production are fully mobile internally but immobile be-
tween countries ;t furthermore, unless otherwise specified, the
country under analysis is assumed to be a small country. This last
assumption implies that international prices are unaffected by the
policies followed by the country under question, simply because the
trade volume of this country constitutes a tiny fraction of the total
volume of world trade in any commodity. The foreign offer curve
in the two-good setting then becomes a ray from the origin, so that
whatever the position of the home offer curve, the terms of trade are
unaltered.
Under this setting the economy can be described by the following
equations:

(4.1)

t The exact conditions under which the community utility function enjoys the
properties of the individual utility functions have been established by Samuelson [II].
t This assumption is not necessary and will be relaxed in the chapter on inter-
national investment. Furthermore, the assumption of inelastic factor supplies is
made only for simplifying the exposition.
Gains from Trade 83
(4.3)

(4.4)

(4.5)

(4.6)

(4.7)

(4.8)
and
Xt+PX2+p3X3+ .. ·PnXn = Dt+PD2+p3D3+ .. ·PnDn (4.9)

where U = utility, D = demand, X= output, K;, L; and N; are


respectively the factors of production -capital, labour and land -
used in the production of the ith commodity, a, b and e are re-
spectively the marginal productivities of capital, labour and land,
and the subscript i denotes the ith commodity.t The price of each
commodity is expressed in terms of the first. Thus p is the world
price of the second commodity in terms of the first, p 3 is the world
price of the third commodity in terms of the first, and similarly for
p 4 , p 5 , . • • , Pn· The community's utility function is described by
(4.1), its production functions by (4.2), and its full-employment
relations by (4.6)-(4.8). Under perfect competition, the value of the
marginal product of each factor is the same in all industries. This
equilibrium situation is described by (4.3)-(4.5), where factor prices
are expressed in terms of the first commodity. Equation (4.9) states
the economy's budget constraint in terms of foreign prices, namely,
the value of production expressed in terms of the first commodity
equals the value of consumption. In the absence of trade, D = X,
whereas in the presence of international trade the demand for some
commodities differs from their local supplies. Note that it is not
necessary that all goods be traded; it is sufficient if at least two are.
For non-traded goods, Dj and Xj simply cancel out from (4.9), where
j denotes the jth non-traded commodity (j ~ n- 2).

t For the sake of expository convenience only three factors are explicitly shown in
the production function. However, the number of factors, as we shall see later, is
immaterial.
84 Studies in the Pure Theory of International Trade

In the two-good framework, we demonstrated that (dXddX2 )


- p, or dX1 + pdX2 = 0. In the general setting, we can show that
(4.10)

4.1 The Optimality of Free Trade


To begin with, we demonstrate that under the stipulated conditions
free trade is the optimal policy, where the criterion for optimality is
the maximisation of social welfare. Although several protagonists of
this view can be found in the eighteenth and nineteenth centuries,
the first rigorous demonstration of the theorem was provided by
Samuelson in 1939 [ 1OJ. Free trade is defined as a situation where the
local and the world prices of all traded goods are the same, assuming,
of course, the absence of transport costs. Our method of analysis in
proving this theorem is this. We seek an expression for the change in
welfare resulting from a slight deviation from the initial situation of
/aissez-faire. If this expression is non-zero, then an increase in wel-
fare could be secured by introducing a suitable policy; otherwise
the initial free trade situation is optimal.
Differentiating (4.1) totally, we obtain

dU = U[dD + ~: dD + ~: dD + ... ~: dD"J


1 1 2 3 (4.11)

where U; = iJU/iJD; is the marginal utility of the ith commodity.

t For simplicity, we derive this expression for the three-good, three-factor case;
the extension to arbitrary numbers of goods and factors, not necessarily equal, follows
the same procedure.
Differentiating (4.2) totally, we have
dX1 = a1dK1+b 1dL 1+e1dN1•
Similarly, from (4.6)-(4.8),
dK 1 +dK2 +dK 3 = dL 1 +dL 2 +dL 3 = dN 1 +dN2+dN3 =0
because K, L and N are all inelastically supplied. Taking this and (4.3)-(4.5) into
account, the expression for dX1 can be written as
dX 1 = a 1 dK 1 +b 1 dL 1 +e 1dN 1 = -p[aidK 2 +dK 3)+b 2(dL 2+dL 3)
+eidN2 +dN3)]

-p[ dX2 + ;- (a 3dK 3+b 3dL 3 +e 3 dN3 )]

-p[dx2 +; dx3J.

Hence dX 1 + pdX2 + p 3dX3 = 0.


Gains from Trade 85
Assuming the absence of satiation in consumption, U; > 0. Since
U2 /U 1 = p, U3 /U 1 = p 3 and so on, (4.11) becomes

(4.12)

Under the small country assumption, p, p 3 , .•. , Pn are all given.


Therefore, from (4.9),

which from (4.1 0) equals zero. Hence dU = 0 under the initial


situation of laissez-faire. In other words, no other policy can lead to
an improvement in welfare. Stated differently, free trade is the
optimal policy.
When the country in question is not small, but is large enough to
influence world prices by manipulating the volume of its trade, that
is, when the country possesses monopoly power in international
trade, t the optimality of free trade cannot be established, but, as
Kemp [7] has shown, free trade still turns out to be superior to no
trade, a situation where the state intervention is so high that no
commodity can be imported and hence exported. The economic
explanation of the superiority of free trade over no trade is this. By
restricting its trade, the country enjoying monopoly power in trade
can cause an improvement in its terms of trade and thus secure
higher welfare. This will be rigorously established in the next
chapter. But when trade is completely prohibited, this benefit
evaporates, because there are no longer any foreign terms of trade.
Hence the complete prohibition of trade leads merely to a mis-
allocation of resources without any compensating benefits. In what
follows, we retain the small country assumption and assume that
the international prices are unaffected by the home country's com-
mercial policy.
A geometrical demonstration of the optimality of free trade turns
out to be very rewarding in terms of clarity and exposition. Let us
assume that the home country produces only two commodities, X 1
and X 2 , each utilising two factors in the production process. In other
words, we are reverting to the transformation curve geometry that

t Note that this situation does not imply monopoly or imperfect competition at
home.
86 Studies in the Pure Theory of International Trade

we have worked with in previous chapters. Consider Fig. 4.1, where


HH' is the home country's transformation curve and the slope of
FP reflects the exogenously given free trade terms oftrade, so that the
production point is given by P.
FP

x,
Figure 4.1

In demonstrating the superiority of free trade over no trade, it is


not necessary to show just how this FP line is determined. Whether
the country in question is a price-taker or possesses monopoly power
in trade, the availability (or the consumption possibility) frontier
under free trade is given by the foreign price line, because now the
country has the choice to select any consumption point along FP.
Without trade, however, the availability frontier is given by the
transformation curve HH'. Since the free trade availability frontier
lies uniformly outside the no-trade frontier (except for the border-
line case at P), the consumption bundle available with free inter-
national exchange of goods is superior to the one attainable in the
absence of trade. Note that, in arriving at this result, we have not
Gains from Trade 87

made use of the small country assumption. It is a relatively simple


matter now to demonstrate that the community welfare under free
trade is superior to the one without trade.
As before, we assume that the home country imports the second
commodity, so that its social indifference curve lies at Con FP to the
north-west of P, with the level of welfare in free trade given by U2 .
Now suppose trade is completely restricted by means of, say, a
prohibitive tariff which raises the domestic price of the importable
good to such an extent that its local consumption falls and is en-
tirely satisfied by the increased local production. At this tariff,
nothing need be imported and hence exported. In the diagram, DP
indicates the tariff-inclusive domestic-price ratio, self-sufficiency
equilibrium is at Sand the level of welfare is at U0 which lies below
the free trade level, U2 • Hence free trade is necessarily superior to
no trade.
The total gain in welfare resulting from the introduction of free
trade can be divided into two components. First, the switch from
no trade to free trade enables the consumers to consume the im-
portable good at lower relative prices, and this results in what
Johnson [6] calls the 'consumption gain'. This gain arises purely
from the international exchange of goods and does not take into
account any secondary change in welfare which may occur owing
to the change in the production point. In terms of Fig. 4.1, the con-
sumption gain is given by the improvement in welfare from U0 to
U'. Second, the switch from no-trade to free trade prices calls for a
shift in the production point. Any gain arising from this shift is
termed by Johnson the 'production gain'. In Fig. 4.1 this gain is
given by the improvement in welfare from U' to U2 • Thus the total
change in welfare resulting from the introduction of free trade is the
sum of the consumption and the production gains.
Next, we show that for a small country free trade is necessarily
superior to restricted trade, which may be effected by a consump-
tion tax on the importable, or a production subsidy to the im-
portable, or a non-prohibitive tariff. Actually, free trade can be
shown to be superior to any kind of trade intervention which may
reduce the volume of trade or promote its expansion.
Consider Fig. 4.2, where the incidence of the consumption tax on
X2 is considered. It is important to remember that a consumption
tax under the small country assumption changes the commodity-
price ratio to the consumer but leaves it unchanged to the producers.
88 Studies in the Pure Theory of International Trade

Figure4.2

Consequently, only the consumption point is affected, and the


production point is not. In Fig. 4.2, as a result of the consumption
tax the relative price of X 2 facing the consumers is indicated by DP',
the consumption point shifts from the initial free trade point C to
C', the production point remains unchanged, and welfare declines
to u2.
By contrast, a production subsidy changes the commodity-pri ce
ratio to the producers but leaves it unaltered to the consumers. The
implications of the production subsidy to the importable good are
examined in Fig. 4.3, where the commodity-pri ce ratio facing the
producers changes to DP', production shifts toP', consumption to
C' which lies on FP' parallel to FP, and welfare declines to U1 •
A tariff, on the other hand, alters the commodity-pri ce ratio to
both producers and consumers. In Fig. 4.1 a non-prohibitive tariff
results in a domestic-price ratio exhibited by the slope of DP', pro-
duction shifts toP', consumption to C', and welfare declines to U1 •
It may be observed that with a tariff, unlike with a consumption tax
or a production subsidy, both the marginal rate of substitution and
the marginal rate oftransfo,rmati on are different from their free trade
levels. In all three cases we find that free trade is superior to restricted
Gains from Trade 89

DP 1

Figure 4.3

trade. Similarly, it can be shown that free trade is superior to any


other type of trade intervention, even if it promotes the expansion of
trade. Hence free trade is the optimal policy for a small country.

4.2 Higher Trade Intervention versus Lower Trade Intervention


From the practical and policy viewpoints, the comparison between
various levels of intervention in free trade is the most important. For
free trade seldom prevails in the real world, and the architects of
commercial policy are frequently faced with the choice between a
higher versus a lower tariff or a higher versus a lower consump-
tion tax and so on. Kemp [7] has shown that less restricted trade
is superior to more restricted trade, where trade restriction is
measured by the divergence between international and local prices.
Kemp limits his discussion to tariffs, quotas and exchange controls.
However, it can be shown that any form of trade intervention would
give rise to his results regardless of whether the intervention is
trade-increasing or trade-decreasing. A rigorous demonstration of
90 Studies in the Pure Theory of International Trade

the proposition that the higher trade intervention is inferior to the


lower level of intervention is presented below.t
Suppose the home country interferes with the free operation of
markets by imposing a tariff or a consumption tax on the imports
of the second commodity, or by granting a production subsidy to its
local production. Here we assume that all goods entering into the
budget constraint are traded. The problems arising from the non-
traded goods will be dealt with in a subsequent chapter. Lett, c and
s denote respectively the tariff, the consumption tax and the produc-
tion subsidy. Then the local relative price of the second commodity
equals p(l + t) with the tariff, that facing the consumers alone equals
p(l +c) with the consumption tax, and the one facing only the pro-
ducers is given by p(l + s). The marginal rate of substitution between
the second and the first commodity must then reflect either the
tariff-inclusive price or the price ratio facing the consumers,
depending on whether it is the tariff or the consumption tax that is
the subject of analysis. Similarly, the marginal rate of transforma-
tion must either reflect the tariff-inclusive prices or the price ratio
facing the producers. It must be noted, however, that with the
production subsidy, for example, the marginal rate of substitution
remains unchanged. Suppose now the level of trade intervention is
slightly increased. Differentiating the utility function totally with
respect to t, we obtain

Repeating the similar differentiation with respect to c, we get

With the production subsidy, however,

(4.16)

t This is a more general version of Kemp's theorem and is free from any trade-
restricting connotations.
Gains from Trade 91

Using the budget constraint (4.9), (4.14)-(4.16) can be expressed in


terms of the outputs; (4.1 0) can then be utilised to obtain simple
expressions for a change in welfare resulting from a rise in the level
of trade intervention. With the consumption tax (4.1 0) is unaltered;
but with the tariff or the production subsidy, pin (4.10) is replaced
either by p(l + t) or by p(l +s). Keeping this discussion in view, and
following the procedure outlined above, (4.14)-( 4.16) can be written
as

_I_. dU = tp [dD 2 _ dX2 ] = tp d£2 (4.17)


U1 dt dt dt dt

l dU dD 2
-·- = ep-- (4.18)
U1 de de
and

u; .----;};
dU
=
dX2
-sp~. (4.19)

In the absence of inferior goods, an increase in t or e leads to a


decline in the demand for imports or for the importable commodity,
so that both d£2 /dt and dD 2 jde are negative. With the supply curve
rising, dX2 /ds > 0. It follows from this discussion that all of the
dUjdt, dUjde and dUjds are negative. In other words, a slight in-
crease in trade intervention of any form culminates in a decline in
welfare. t Equations (4.17) and (4.18) also confirm Bhagwati's
results [I] that, in the presence of inferior goods, the higher tariff
(or consumption tax) may be superior to the lower tariff (or con-
sumption tax).
The geometric presentation of these results in the two-good case
follows the technique developed before. In Fig. 4.4 the domestic-
price ratio with the initial tariff is indicated by DP. The increase in the
tariff shifts it to DP', production moves from P toP', consumption

t It is worth noting here that in the discussion of the consumption tax or the pro-
duction subsidy, it is not necessary to specify whether a commodity is exported or
imported. Thus whether D 2 in (4.18) and X 2 in (4.19) are respectively the consump-
tion and the production of the importable or the exportable good is immaterial to the
theorem in question. The theorem permits easy extension to other forms of inter-
vention in free trade, including the consumption tax or subsidy and the production
tax or subsidy on the exportable good.
92 Studies in the Pure Theory of International Trade
FP'

Figure 4.4

Figure 4.5
Gains from Trade 93

from C to C', and welfare declines from U2 to U 1 • Note that the


foreign-price ratio is unchanged at FP parallel to FP' and the im-
port volume declines as a result of the shift from C to C'. Note
further that if the exportable is the inferior good, the import
volume may rise with the tariff, consumption may switch to C", and
welfare may improve to u3.
In the case of the consumption tax, we can go back to Fig. 4.2. As
the consumption tax is raised, the domestic-price ratio facing the
consumers shifts from DP' to DP", consumption from C' to C",
production is unchanged, and welfare declines from U2 to U1 • How-
ever, if the export good is an inferior good, the import demand may
increase with the consumption tax and welfare may rise. (This is
not shown in the diagram.)
With the production subsidy, the marginal rate of substitution
remains unchanged as in Fig. 4.5, where the higher subsidy to X 2
moves production from P toP', consumption shifts from C to C',
and welfare declines from U2 to U1 • It is worth noting perhaps that,
since trade takes place along the international prices, each con-
sumption point must lie on FP.
However, the proposition that the higher trade intervention is
necessarily inferior to the lower level of intervention in the absence
of inferior goods does not hold under realistic situations where
tariffs or other forms of intervention may exist on the imports of
several commodities. The theorem transpires to be valid only when
the intervention occurs in the international market for only one
commodity. If, say, tariffs exist on the imports of two or more goods,
the theorem is no longer tenable.
In this spirit, assume the imports of, say, the third good are also
subject to a tariff t 3 , so that its local price expressed in terms of the
first good is given by p 3 (1 +t 3 ). Equation (4.14) is then superseded
by

and (4.10) by
94 Studies in the Pure Theory of International Trade

Differentiating the budget constraint and using (4.21), (4.20)


becomes

I dU dE2 dE3
--=tp-+t3P3- (4.22)
ul dt dt dt

which means that unless dE3/dt ~ 0, the sign of dUjdt is ambiguous.


Normally, an increase in the tariff on one commodity, with un-
changed tariffs on others, will raise the import demand of the other
commodity, so that dE3 /dt > O.t In other words, there is clearly no
a priori reason to believe that a higher tariff on one imported com-
modity is inferior to a lower tariff when tariffs on other importables
are constant even in the absence of inferior goods. The same result
applies to other forms of trade intervention.t

4.3 Trade Intervention versus No Trade


It is a relatively simple matter now to compare the welfare implica-
tions of trade intervention with the complete absence of trade.
Kemp [7] subscribes to the view that some trade is superior to no
trade at all, whereas Bhagwati [I] has recently demonstrated that
this proposition is valid only if trade intervention takes the form of
tariffs, quotas or exchange control, policies which give rise to the
divergence between foreign prices and domestic prices facing both
the consumers and the producers. Stated differently, if the inter-
ference in the free inter-country flow of goods is such as to disrupt
the equality between foreign prices and the domestic prices facing
either the consumers or the producers, but not both, Kemp's
theorem that some trade is necessarily superior to no trade may no
longer be valid. In other words, Kemp's theorem is tenable only for

t This is necessarily true if all imports are gross substitutes.


t With the corresponding consumption tax on the third commodity (e 3 } already
in existence, we have
I dU dD 2 dD 3
-- = ep-+e3p3-.
u! de de de
Similarly, in the presence of a production subsidy to the third commodity (s 3 } a rise
in the production subsidy to the second good yields

__!___ dU = -[sp dX2 +s 3 p 3 dX3].


U 1 ds ds ds
The reader may derive his own conclusions from these expressions.
Gains from Trade 95

the class of the trade restrictions he considered, namely tariffs, etc.,


but it cannot be generalised to all forms of trade intervention.
That the tariff-ridden trade is superior to no trade is apparent from
one glance at Fig. 4.1, Where U 1 , the level of welfare associated with
a non-prohibitive tariff, is superior to U0 , the welfare level pre-
vailing in the presence of a prohibitive tariff. This is a simple
demonstration of Kemp's viewpoint.

H'
Figure 4.6

A simple counter-example when trade intervention consists in the


imposition of a consumption tax on the importable commodity, X2 ,
is presented in Fig. 4.6. The domestic-price ratio facing the con-
sumers at a low level of the consumption tax is indicated by DP;
consumption is at C and welfare at U1 • A higher consumption tax
which shifts DP to DP' not only lowers welfare to U' but lowers it
below the autarky level U0 . It may be noted that some trade exists
even with the higher consumption tax, because the consumption
point, C', differs from the production point, P. The necessary con-
dition for this result is that the consumption of importables after
the introduction of the consumption tax be less than the correspond-
ing level in autarky.
96 Studies in the Pure Theory of International Trade
FP

Figure 4.7

A similar demonstration in the presence of a production subsidy


to importables appears in Fig. 4.7, where Pis the production point,
C the consumption point, and U 1 the welfare level associated with
the initial low level of production subsidy to the importable com-
modity, X 2 • Suppose now that this subsidy is increased so that
production shifts to P', consumption to C' and welfare to U'. This
latter welfare level is not only below U1 but also below the autarky
level given by U0 . The necessary condition for this result is that the
local production of importables associated with the higher subsidy
be higher than the production of importables in autarky.
The trade intervention depicted in these two counter-examples
tends to restrict the volume of trade, as can be observed from
Figs. 4.6 and 4. 7. However, these results are valid even if the inter-
vention tends to expand the quantum of trade. For example, it can
be shown that increasing doses of production subsidy to the ex-
portable commodity (X1 ) would continue to move the production
point towards the H' extremity of the transformation curve, promote
the expansion of trade, and lead to a continuous decline in social
welfare which, at some level of subsidy, might decline even below
the autarkic level.
Gains from Trade 97
4.4 Tariffs, Subsidies and Taxes
The analysis of the previous sections can be used to compare the
welfare implications of tariffs with those of production subsidies and
consumption taxes. Corden [ 4] has shown that a tariff is inferior to
an equivalent production subsidy, where equivalence is defined here
in terms of the equality between the protective effect generated by
the two policies on the production of the importable commodity.
Suppose trade intervention occurs in the world market for the
second commodity only. Then the equivalence between the tariff
and the subsidy is defined by

s = t, ds = dt, and

that is to say, we compare those rates of the tariff and the production
subsidy which result in an equal rise in the output of the second
commodity.
Corden's theorem can be demonstrated simply by subtracting
(4.19) from (4.17), so that

G =_I [dU _ dUJ = p[t d£2 +s dX2 ]


ul dt ds dt ds
= p[t{dD 2 _ dX2 }+s dX2 ]·
dt dt ds
For Corden's theorem to be valid, G < 0. With s = t and dX2 /dt
= dX2 /ds under the definition of equivalence,

dD 2
G = tpdt < 0

in the absence of inferior goods, for then a rise in the tariff rate would
lead to a decline in the local demand for the importable good.
The two-commodity case permits a simple geometrical demon-
stration of this proposition. In Fig. 4.8, P is the initial production
point and P' the production point as a result of the introduction of
the tariff which results in a domestic-price ratio indicated by the
slope of DP. An equivalent production subsidy which also shifts the
production point from P to P' furnishes a welfare level of u•. How-
ever, with the tariff, the welfare level is given by U,, which lies below
u•. Hence the tariff is shown to be inferior to the equivalent produc-
tion subsidy. The reason is not far to seek. The introduction of a
98 Studies in the Pure Theory of International Trade

FP

x1
Figure 4.8

tariff creates a divergence between the given foreign-price ratio


(FP) and the domestic marginal rate of transformation and the
marginal rate of substitution, both of which are reflected by DP in
Fig. 4.8, and hence generates distortions on the production as well
as the consumption side. However, a production subsidy gives rise
to an equivalent distortion on the production side only, and hence
results in a lower decline in welfare than the tariff.
Next, we tum to the relative welfare implication of tariffs and
consumption taxes. Bhagwati and Srinivasan [3] have recently
shown that a consumption tax is superior to an equivalent tariff
when the community's objective is to restrict the consumption of the
importable good at a certain level. Equivalence is here defined by
tdD 2/dt = edD 2jde.
Subtracting (4.18) from (4.17), we have

Q = _1 [dU _ dU] = p[t d£2 -e dD2]


ul dt de dt de

= p[t{dD 2 _ dX2 }-e dD 2]·


dt dt de
Gains from Trade 99

For the validity of the theorem by Bhagwati and Srinivasan, we


require that Q < 0. With cdD 2 /dc = tdD 2 /dt under the definition
of equivalence,
dX2
Q = -tpdt < 0

because dX2 /dt > 0.


For a geometrical demonstration, consider Fig. 4.9, where wel-
fare with the tariff is given by U1 and that with the equivalent con-
sumption tax by Uc. The equivalence of the two policies is reflected
by the fact that c and c1 are horizontally aligned, so that the con-
sumption of X 2 is the same in both cases. Since Uc lies above flr,
the superiority of the consumption tax over the equivalent tariff
is clear.

FP

x1
Figure 4.9

4.5 The Terms of Trade and Welfare


The classical economists frequently used the terms of trade as
an indicator of a country's. gains from trade. A rigorous demon-
stration to this effect has recently been provided by Krueger and
Sonnenschien [8]. Assuming that the country is a price-taker in the
l 00 Studies in the Pure Theory of International Trade

world markets, what is the effect of an exogenous change in one


foreign-price ratio, with all the other terms of trade remaining con-
stant? For example, suppose there is a change in p alone. Differen-
tiating (4.1) and (4.9) totally with respect top, using (4.10) and
remembering that the initial situation is one of free trade, we obtain
l dU
- -= X2-D2 = -E2 (4.23)
ul dp
which is less than zero, because £ 2 > 0. In other words, there is an
inverse relationship between the level of welfare and p. Now a
decline in p, with all other foreign prices constant, implies an im-
provement in the terms of trade and vice versa. Thus we conclude
that an improvement in the terms of trade results in an improvement
in welfare, and conversely. Note that it was not necessary to assume
the absence of non-traded goods. Note further that the change in
welfare is proportional to the initial volume of trade.
Fig. 4.10 depicts this result diagrammatically. The initial terms
of trade are given by FP, the new favourable ones by FP', produc-
tion shifts from P to P', consumption from C to C', and welfare
improves from U 1 to U 2 •

Figure 4.10
Gains from Trade 101

The relationship between the terms of trade and welfare is no


longer predictable when more than one terms of trade change
exogenously. For example, suppose the first commodity is exported
and the second and the third are imported. Assume further that
both p and p 3 change exogenously. Following the procedure in-
dicated above, we can derive

u:I dU
dp
[ dp 3 ]
= - £2 + £3 dp .
Now if all terms of trade alter in the same direction, there is no
problem. For example, if dp 3 and dp have the same sign, dU/dp is
unambiguously negative. The trouble arises when dp 3 and dp have
different signs, so that the sign of dU/dp is not predictable in the
absence of information concerning the extent of the shifts in each
terms of trade and the initial volumes of the two imports. It is worth
pointing out, however, that the sign of dU/dp may be indeterminate
not because of the presence of more than two traded goods, but
because the two terms of trade may not change in the same direction
and it may then be impossible to speak of an improvement or a
deterioration in the terms of trade.

4.6 Discriminatory Tariffs: The Theory of Customs Union


Some sections of this chapter have been devoted to the examination
of welfare implications of tariffs which were implicitly assumed to be
non-discriminatory. The analysis presented there can be directly
used to explore the welfare implications of discriminatory tariffs,
or what is commonly called the customs union. The tariff dis-
crimination can occur in two ways. A country may impose different
taxes on the imports of the same commodity from different coun-
tries. This is called country discrimination. Alternatively, a country
may impose different rates of duty on the imports of different com-
modities. This is called commodity discrimination. 'The theory of
customs union', in Lipsey's words, 'may be defined as that branch
of tariff theory which deals with the effects of geographically dis-
criminatory changes in trade barriers' ( [9] p. 496). Prior to a pioneer-
ing contribution by Viner [12], the theory of customs union
attracted little attention, the popular belief being that customs
union will always be welfare-increasing simply because it involves
reductions in tariffs and hence constitutes a movement towards
102 Studies in the Pure Theory of International Trade

freer trade. This result was challenged by Viner who demonstrated


that customs union can be welfare-decreasing. His demonstration
consisted in the introduction of the now familiar concepts of trade
diversion and trade creation. In the nature of the anlysis, a customs
union involves at least three countries. In this spirit, then, suppose
that there are three countries, the home country (H), its union
partner, country B, and another country, F, each of which produces
two commodities, X1 and X 2 • Suppose further that F is the least-
cost and H the highest-cost producer of the second commodity the
import of which is prevented by a prohibitive tariff by the home
country. If now the home country forms a customs union with B,
she will be better off than before. This is an example of trade
creation. However, if initially the home tariff was non-prohibitive
and non-discriminatory, she would be importing the second com-
modity from its least-cost producer, country F. Now suppose the
home country enters into an agreement with B, and removes its
tariff on B's goods such that to the home importers F is no longer
the least-cost supplier. This is an example of trade diversion where
the home country switches its imports from the lower-cost producer
to the higher-cost producer. According to Viner, trade-creating
customs union results in a gain in welfare, whereas trade-diverting
union leads to a welfare loss. Viner thus assailed the then tradi-
tional view that customs union was generally welfare-improving.
Although Viner's logic appears impeccable, his main proposition
that trade diversion is necessarily welfare-decreasing is open to
some serious objections. Gehrels [5] and Lipsey [9], accusing Viner
of ignoring the substitution possibilities on the consumption side,
have shown that trade diversion need not lead to a welfare loss when
such possibilities are taken into account. Lipsey, however, ignores
the substitution effects on the production side, and if such effects
are operative, which they must be in the absence of complete spe-
cialisation, Viner's proposition may not be valid, even if the substi-
tution effects on the consumption side are ignored. Lipsey's
rendition of Viner has recently been challenged by Bhagwati [2]
who, while building his case on the substitution possibilities on the
production side, argues that the implicit assumption made by Viner
in demonstrating the welfare-reducing potential of the trade-
diverting union is concerned not with the constancy of the con-
sumption pattern but with the constancy of the level of imports.
Bhagwati's interpretation appears all the more plausible when we
Gains from Trade 103

remember that the absence of substitutability on the consumption


side alone is not a sufficient condition for trade diversion to be
detrimental to social welfare, provided the specialisation is in-
complete. All these points emerge clearly from a formula derived
in the following, which determines the conditions under which trade
diversion can be shown to be beneficial.
The trade-diverting customs union in general gives rise to two
conflicting pulls, one tending to raise and the other tending to lower
welfare. On the one hand, the repeal of the tariff by the home
country on (potential)t imports from B works for an increase in
welfare; on the other hand, the switch from the lower-cost producer
to the higher-cost producer, equivalent to a deterioration in the
terms of trade, serves to lower welfare. In this spirit we may write

where, as before, t is the non-prohibitive, non-discriminating home


tariff on its imports and p is the exogenously determined terms of
trade between the home country and its trading partner. In other
words, the home welfare becomes a function of concomitant changes
in t and p when the home country decides to form a customs union.
Furthermore,

so that
dD 1 dD 2 d£ dp
-+p(l+t)- = t p -2- E2 - .
dt dt dt dt

Differentiating the utility function totally with respect to t, we have

I dU dD 1 dD 2 d£2 dp
- - = -+p(l+t)- = tp--£2-
U1 dr dr dr dr dr
or
(4.24)

t Assuming, of course, that the relative price at which B offers its goods is lower'
than the tariff-inclusive price of the goods which the home country imports in the
absence of the customs union from country F.
104 Studies in the Pure Theory of International Trade

where E1 = (t/ E 2 )(dE2 /dt) is the total home elasticity of demand for
imports with respect to the tariff,t and AP = (Ijp)(dpjdt) is the pro-
portionate change in the terms of trade as the home country switches
its trade from country F to country B.t The home country's agree-
ment to form a customs union implies that dt < 0. The signs of E1
and AP, however, depend on the characteristics of the union. In
general, dp is negative with the trade-creating union, but positive
with the trade-diverting union. However, in our example, where the
small home country initially trades with country F, the trade
creation can be represented by a simple removal of the tariff, with
no accompanying change in the terms of trade. This suggests that
with trade creation dt < 0 and AP = 0, but with trade diversion
dt < 0 and AP < 0. Like the sign of AP, the sign of E1 is also deter-
mined by the nature of the customs union. Under trade creation,
where dp :::; 0, the repeal of the tariff must lead to a rise in the
demand for the importable good. In other words, d£2 /dt and hence
E1 are negative.§ Now for the customs union to be beneficial to
welfare, dU > 0. It is at once clear from (4.24) that with AP = 0 and
E1 < 0, the trade-creating union leads to a rise in social welfare.
The sign of £ 1 is not so apparent when the union is of the trade-
diverting type. Since AP < 0 under trade diversion, (4.24) suggests
that the necessary condition for dU > 0 is that £ 1 < 0. Lipsey
postulated a Ricardian economy where any amount of profitable
trade results in complete specialisation in the exportable good.
Under Lipsey's setting, then, dX2 /dt = 0, so that dE2 jdt reduces to
dD 2 /dt. Thus, if £ 1 is to be negative, dD 2 jdt must be negative. This

t £, is the total home elasticity of demand for imports because it represents the
effects on the home demand for imports of a simultaneous change in both 1 and p
that occurs as a result of the formation of customs union.
t The reader may be reminded here that the change in the terms of trade resulting
from the formation of the trade-diverting union is not because the home country
comes to possess natural monopoly power in trade and ceases to be a small country,
but because of the exogenous and once-for-all shift in the terms of trade consequent
upon the home country's decision to form the union with Band thus accept the given
terms at which B will offer its goods.
§This can be seen immediately by differentiating £ 2 = E 2 (1,p) totally with
respect to 1 to obtain
d£2 o£2 o£2 dp
-=-+-·-
dl a1 ap d1
so that with o£2 /ol < 0, o£2 /op < 0 and dp/dl ~ 0 under trade creation, d£2 /dl and
hence £, are negative.
Gains from Trade 105

means that the formation of the customs union must give rise to a
decline in the relative price of the importable good facing the con-
sumers in the home country. For the unfavourable movement in the
terms of trade that accompanies the trade-diverting union is bound
in the presence of complete specialisation to lower the value of home
production (or national income measured in terms of outputs), and
this factor tends to make dD 2 negative and dD 2 /dt positive. For
dD 2 to be positive, therefore, the necessary condition is that the
trade-diverting union contribute to a decline in the relative price
of the second good. This latter condition can also be seen to be
necessary if the trade diversion is ever to lead to an improvement in
welfare. For if dD 2 > 0 and hence dD 2 /dt < 0, E1 < 0, and if
JE,J > JAPJ, dU may be positive. However, if the substitution effects
on the consumption side are ignored, dD 2 must be negative, dD 2 /dt
and hence E, must be positive, and with A, < 0, dU must be nega-
tive. t In other words, if the goods are consumed in a given propor-
tion, trade diversion must lead to a loss in welfare. This is Lipsey's
interpretation. He accuses Viner of implicitly assuming that the
consumption pattern is constant, for otherwise trade diversion may
also contribute to an improvement in welfare.
Lipsey's critique, however, is open to a serious objection in that he
postulates a Ricardian economy and ignores the substitution possi-
bilities on the production side. Let us then examine these issues
when dX2 =F 0. If the formation of the union results in a decline in
the relative price of the importable good (which, as we have seen

t The validity of this argument becomes vivid if we differentiate D 2 = D2


[p(l + 1), Y] with respect to 1 to obtain
dD 2 oD 2 dp(l + 1) m2 dY
-;:-;c--c:------:-,;·---+-- · - (a)
d1 o[p(l + t) J d1 p(l + 1) d1
where m 2 is the home marginal propensity to consume the second good. We assume
that 0 < m 2 < I. Substituting this expression in (4.24), equating (ljU 1 )(dUjdl) with
dY/dl and remembering that under Lipsey's formulation d£2 = dD 2 and £ 2 = D 2 ,
we obtain
2
d Y [ l -m- 1J [= p
1oD 2 ·dp(l
--- +I) J
- D 2 A dl.
1+ 1 o[p(l + 1) J d1 •
Since m 21/(l +I)< I, dY < 0 if oD 2 /o[p(l +t)] = 0, because under trade diversion
A• and d1 are both negative. In other words, if the substitution possibilities are
ignored on the consumption side, trade diversion in Lipsey's model leads to a decline
in welfare. It may also be seen from expression (a) that with dYjdl > 0 when the
substitution effect is ignored, dD 2 jd1 > 0; this is what has been intuitively argued
in the text.
I06 Studies in the Pure Theory of International Trade

above, is the necessary condition for trade diversion to be beneficial


in Lipsey's formulation), then dX2 < 0, so that dX2 /dt > 0. This
suggests that (dE2 /dt) = (dD 2 /dt)- (dX2 /dt) may be negative even
if dD 2 /dt is positive. Thus even if the consumption pattern is fixed,
so that dD 2 /dt > 0, £ 1 (whose sign is the same as that of dE 2 /dt) may
be negative, and if this effect is sufficiently pronounced, dU from
(4.24) will be positive and trade diversion would lead to a rise in
welfare. Hence Lipsey's contention that Viner implicitly assumed
the constancy of the consumption pattern in order to demonstrate
the inferiority of trade diversion is not tenable, because the alleged
implicit assumption is not sufficient to rule out the beneficial effects
of the trade-diverting union.
Taking these facts into account, Bhagwati [2] has presented a
more convincing interpretation of Viner's analysis. He suggests that
Viner implicitly assumed the constancy of the level of imports, rather
than the constancy of the proportion in which the two goods are
consumed, for under the former assumption the detrimental effects
of the trade-diverting union can be unambiguously demonstrated.
This follows vividly from an examination of (4.24), for if the im-
ports are given, then £ 1 = 0, and with AP < 0 and dt < 0, dU < 0.
In fact, so long as £ 1 2: 0, Viner's proposition that trade diversion
is welfare-decreasing is valid. This serves to confirm Bhagwati's view
that, so long as d£2 ::; 0, the trade-diverting union is welfare-
decreasing. However, although Bhagwati's condition is weaker than
the one suggested by Lipsey, it is not as weak as the one suggested
by our formula (4.24). As a matter of fact our formula also furnishes
a necessary and sufficient condition for trade diversion to be welfare-
increasing. The two factors that determine the welfare implications
of customs union are the home elasticity of demand for imports with
respect to the tariff, and the extent of the deterioration in her terms
of trade resulting from the formation of the customs union. It is
clear from (4.24) that dU ~ 0 if IErl 5: iApi·
These results are illustrated geometrically in Fig. 4.11, where H H'
is the home transformation curve, DP indicates the domestic-price
ratio, FP represents the terms of trade between the home country
and F, and at the initial consumption point given by C, which lies
on the social indifference curve U, the marginal rate of substitution
equals the domestic-price ratio. Clearly, then, the initial situation
describes the case of a non-prohibitive tariff by the home country,
but since F is the least-cost producer of the second commodity, she
Gains from Trade 107

Figure 4.11

is also the only supplier of the home imports. Now suppose the home
country and B form a union which eliminates the importation of the
F goods by any union partner. The new terms of trade facing the
home country are then given by FPb which are less favourable to the
home country than those reflected by FP, ensuring that the union
is trade-diverting. If, following Lipsey, we ignore the substitution
effects on the consumption side, the consumption point in every
case must lie on a given ray from the origin. Suppose this ray is
given by OR and the initial consumption point lies on it. If the sub-
stitution effect on the production side is also ignored, then the
production point remains unchanged in spite of the shift in the
relative prices from DP to FPb. Under· this setting, the new con-
sumption point is given by C' which is clearly inferior to the initial
point C. This is how Lipsey interprets Viner. He argues that if
substitution is allowed on the consumption side, the new consump-
tion point will be different from C' and the level of welfare attained
108 Studies in the Pure Theory of International Trade

will be higher than that associated with C'. In the diagram, FPb is
drawn tangential to the original indifference curve at Cb. It follows
immediately that, by consuming at Cb, the home country can attain
the original level of welfare even with the customs union. Clearly,
then, if the terms of trade between the home country and B were
slightly more favourable than those reflected by FPb, the customs
union would result in a gain in welfare.
If substitution is permitted on the production side also, the pro-
duction point will shift from P to P' where FP~ (parallel to FPb)
touches the transformation curve. If substitution on the consump-
tion side is ignored, the new consumption point C" is superior to the
initial point C in terms of welfare. The welfare gain, of course, would
be higher if commodities are not consumed in a given proportion.
But the point is that even if this is conceded, substitution on the
production side itself may nullify Lipsey's assertion that with a given
consumption pattern trade diversion necessarily leads to a decline in
welfare. The proposition of general validity is that welfare with a
trade-diverting union may decline, rise, or even remain unchanged.
A necessary condition for trade diversion to be welfare-improving
is that the union terms of trade (FPb) must be intermediate to the
domestic-price ratio (DP) and FP. Likewise, a sufficient condition
for trade diversion to tie detrimental to welfare is that the union
terms of trade are such as to raise the output of the importable good
or leave it unchanged. This can be verified easily from the diagram.
In the course of our analysis, the reader may have noticed the
intimate relationship that exists between the propositions estab-
lished in this section and Kemp's theorem that a higher tariff is
inferior to a lower tariff. We have already shown that Kemp's
theorem may not hold when more than one import is subject to the
tariff. It follows then that our customs union results also cannot be
extended beyond the narrow domain of two traded goods.

REFERENCES
[1] Bhagwati, J., 'The Gains from Trade Once Again', Oxford Economic Papers,
XX (July 1968) 137-48.
[2] - , 'Customs Unions and Welfare Improvement', Economic Journal, LXXXI
(Sept 1971) 580-7.
[3] -,and Srinivasan, T. N., 'Optimal Intervention to Achieve Non-Economic
Objectives', Review of Economic Studies, XXXVI (Jan 1969) 27-38.
[ 4] Corden, W. M., 'Tariffs, Subsidies and the Terms of Trade', Economica, XXIV
(Aug 1957) 235-42.
Gains from Trade 109

[5] Gehrels, F., 'Customs Unions from a Single Country Viewpoint', Review of
Economic Studies, XXIV (Jan 1956) 61-4.
[6] Johnson, H. G., 'Optimal intervention in the Presence of Domestic Distortions',
in Trade, Growth and the Balance of Payments (Chicago: Rand McNally, 1965).
[7] Kemp, M. C., The Pure Theory of International Trade and Investment (New
Jersey: Prentice-Hall, 1969) chap. 12.
[8] Krueger, A. 0., and Sonnenschien, H., 'The Terms of Trade, the Gains from
Trade and Price Divergence', International Economic Review, VIII (Feb 1967)
121-7.
[9] Lipsey, R., 'The Theory of Customs Unions: A General Survey', Economic
Journal, LXX (Sep 1960) 496-513.
[10] Samuelson, P. A., 'The Gains from International Trade', Canadian Journal of
Economics and Political Science, v (May 1939) 195-205.
[II]--, 'Social Indifference Curves', Quarterly Journal of Economics, LXX (Feb
1956) 1-21.
[12] Viner, J., The Customs Union Issue (New York: Carnegie Endowment for
International Peace, 1950).
5 The Theory of Nominal Tariffs

Because of its long chronicle, the theory of nominal tariffs has earned
a pivotal place in the literature on the pure theory of international
trade. Historically, the main concern of this theory has been with the
implications of tariffs on the imports of final products, and it is only
recently that the trade theorist has begun to explore the implications
of tariffs on the imports of intermediate products, despite their
existence in practice for centuries. The fast-proliferating literature
concerned with the latter question is surveyed in the chapter on
effective protection; the present chapter is concerned solely with the
issues arising from the imposition of nominal tariffs.
The theory of nominal tariffs figured prominently in the classical
writings of Mill [11], Marshall [8] and Graham [1, 2], among
others, and as with the gains from trade, the theory of tariffs
branched directly out of the controversy over free trade and pro-
tectionism. Some of the normative aspects of tariffs were examined
in the previous chapter. Here our main concern will be with the
positive analysis of tariffs. Although the analysis of tariff dates
back to the classical tradition, interest in the nominal tariffs has
been kept alive by two pioneering contributions, one by Stolper
and Samuelson [13] and the other by Metzler [10], whose basic
analytical framework was cast in the modern Heckscher-Ohlin
mould. The seminal results of Metzler have been more recently
challenged by Sodersten and Vind [12], but Jones [ 4] has effectively
demonstrated the spurious nature of their critique. The analysis in
this chapter parallels closely the illuminating discussion provided
by Jones.

5.1 The Analytical Framework


For our analytical tool-kit, we return to the two-country, two-
commodity, two-factor framework outlined in Chapters 2 and 3.
The generalised multi-commodity, multi-factor framework ex-
The Theory of Nominal Tariffs Ill

pounded in the previous chapter is not suitable for the issues dis-
cussed in the present chapter.
In the presence of tariffs, each country's demand for imports
depends on the international terms of trade, p, and the level of its
tariff. As before, £ 2 = D 2 - X 2 is the import demand by the home
country for the second commodity, and E 1f = D 1f-Xlf is the
foreign country's import demand for the first commodity. Let lh
and If be the tariffs imposed respectively by the home and the foreign
country and let 7;. = (I +I h) and 1f = (1 +If). Then
E2 = E 2(p,7;,) (5.1)
and
(5.2)
Furthermore,
(5.3)
and
(5.4)
where Ph = pi;, and pf = p/1J are respectively the domestic-price
ratios in the home and the foreign country, Y is national income in
the home country and lf is national income in the foreign country.
For our purposes it is the change in the real income affecting the
change in demand that is relevant. If the change in social welfare is
an index of the change in real income, then for the two-commodity
case we may write
dU
- = dY = dD 1 +phdD2 (5.5)
ul
which expresses the change in the home real income in terms of the
first commodity and is free from any cardinal utility restrictions.
Similarly,
(5.6)
In the absence of tariffs, the budget constraint, as in the previous
chapter, simply shows that the value of consumption equals the
value of production. However, in the presence of the tariffs the
complications arising from the distribution of tariff proceeds have
also to be taken into account. Here there are several alternatives.
112 Studies in the Pure Theory of International Trade

The government may just collect tariff revenue without spending it,
in which case the value of consumption expressed in terms of
domestic prices is still restricted by the value of production. On the
other hand, the government may spend the tariff proceeds in the
manner that private consumers would; this procedure also causes
little trouble. However, if the government's spending proclivities are
different from those of the private sector, the analysis becomes quite
complicated, for now we also have to consider separate demand
functions for the government. The simplest way of handling the
problem arising from the disposal of tariff proceeds is to assume that
the government hands the tariff revenue back to the private sector
in a lump-sum fashion. In this last alternative, which is the one
adopted here, the value of consumption expressed in terms of
domestic prices is restricted by the value of production as well as
tariff proceeds.
Another issue that deserves consideration here concerns the way
the tariff revenue is to be expressed. For the home country which
imports the second commodity, the tariff revenue simply equals

However, with the foreign country importing the first commodity,


which has all along been taken to be the numeraire, the expression
is not so simple. If plf is the absolute local price of the first com-
modity in the foreign country and p 1 is the corresponding world
price, then the tariff proceeds are given by

Elf(plf-pl) = tfplElf

but since plf = p 1 (I+ t1 ), the tariff revenue equals

so that when expressed in terms of the first commodity, the tariff


revenue is given by

With all this information at hand, the budget constraint in each


country, expressed in terms of domestic prices, is given by

(5.7)
The Theory of Nominal Tariffs 113

and

Finally, the balance-of-payments constraint is described by


(5.9)
or
(5.10)

5.2 Tariffs and the Terms of Trade


In order to determine the effects of tariffs on the terms of trade, the
system of equations presented above needs to be differentiated. As
before, let an asterisk denote the relative rate of change. The
analytical procedure followed here, and actually to be followed later
in all questions of comparative statics in trade theory, runs parallel
to the procedure employed in the derivation of stability conditions.
First, we examine the impact of any change in the system on the
excess demand for importables at constant terms of trade. If the
change in this excess demand is not zero, then the prices in the
system will have to change to restore the equilibrium situation.
In this spirit, differentiate (5.1) and (5.2) totally to obtain
(5.11)
and
(5.12)

where ah = -(p/E2 )(oE2 /op) and a1 = (p/Elf)(iJElf/op) are the


familiar elasticities of demand for imports, Ah = - (T,/E 2 )(iJE2 /oT,)
and A1 = -(1j/E 11 )(oE11 /o1j) are respectively the home and the
foreign elasticities of import demand with respect to the tariff.
Equations (5.11) and (5.12) state that the change in each country's
demand for imports can be divided into two parts, the first relating
to the change in the import demand along the offer curve for a given
tariff and the second arising from the shift in the offer curve itself as
a result of the change in the tariff for given terms of trade. Ah and A1
have ~en defined to be non-negative to signify that the rise in the
tariff leads to a decline in the demand for imports at a given p. The
justification for this will be provided later.
114 Studies in the Pure Theory of International Trade

For the sake of simplicity and convenience, let us limit our


differentiation to the home country. From (5.3),

(5.13)

where eh = -(ph/E 2 )(oD 2 /oPh) describes the substitution effect of


a change in Ph on D 2 at a constant real income, sh = (ph/E 2 )
(dX2 /dPh) captures the substitution effect on the production side as
output changes along the transformation curve,t and mh =
ph(oD 2 joY) equals the home marginal propensity to consume the
importable good. Differentiating (5. 7) and using (5.5), we have

(5.14)

Differentiating Ph = pT,, we get

(5.15)

and substituting in (5.14), we obtain

(5.16)

Substituting (5.16) and from (5.15) the relation

Pt = P' + Tt
in (5.13), we derive

(5.17)

The comparison of (5.17) with (5.11) suggests that

(5.18)

t The substitution effect on the production side represents the change in the output
of the commodities that accompanies the change in commodity prices. This substitu-
tion effect depends, among other thirrgs, on the elasticities of factor substitution in
the sectors. For further discussion of this effect, see Jones [ 4].
The Theory of Nominal Tariffs 115
and
Ah = (sh+eh) (5.19)
(1-mhth/T,)
Since th/T, < 1, and for non-inferior goods 0 < mh < 1, the de-
nominator of(5.17)-(5.19) is positive; (5.18) furnishes the factors
that influence the elasticity of demand for imports as we move along
the offer curve. In view of the construction of the offer curves
described in Chapter 3, it is not surprising to find that ah depends
upon the substitution effects on the consumption and the produc-
tion side as well as the marginal propensity to consume importables;
in addition, the initial level of tariff also affects the magnitude of the
elasticity of the import demand. The expression (1-mhth/T,), as
Jones ([4] p. 420) has observed, is similar to the Keynesian type of
multiplier, and this indeed is the case. Let us call the term
1/( 1 - mh th/T,) the tariffmultiplier. If initially there is no tariff, so that
th = 0, this multiplier phenomenon disappears. In the presence of
an initial tariff, however, this multiplier, as with the Keynesian
system, has a multiple effect. The major difference is that, unlike in
the Keynesian framework where the multiple effect is on national
income, the multiplier in the barter-trade model has a pronounced
effect on the demand for imports. It affects both ah and Ah. For
example, suppose that with a tariff already existing there is an exo-
geneous increase inp so that £ 2 declines. Now a decline in £ 2 lowers
the community's receipts from the tariff proceeds, which leads to a
further decline in the demand for imports, and so on. Therefore the
final decline in £ 2 equals the initial decline in £ 2 multiplied by the
tariff multiplier 1/(1-mhth/T,), and similarly for the term Ah. For a
given p and an initial tariff, any further rise in the tariff serves to
lower the import demand by raising the domestic production of the
importable good as well as by lowering its domestic consumption.
These two effects show themselves in the numerator of (5.19). The
same multiplier effect appears once again. The initial decline in the
import spending serves to lower the community's receipts from the
tariff proceeds, which in tum leads to a further decline in the import
demand and so on, so that eventually the tariff multiplier again has
a cumulative effect on Ah. If initially there is free trade, the intro-
duction of the tariff at constant p causes no change in the national
income, as is evident from (5.16) where with p* = 0 and T, = 1, dY
also equals zero. It may be noted that this is why there will then be
116 Studies in the Pure Theory of International Trade

no income term in (5.19). t The income term (mh) does appear when
th > 0, but only via the tariff multiplier.
Enough has been said on the factors determining the demand for
imports by the home country. By following essentially the same, but
a little more involved, procedure, an analogous expression for ET1
can be obtained :t
EfJ = p*[ef+sf+mf/Tf]- TJ(ef+sf). (5.20)
(l-m1 t1 jT1 )
The next step involves the differentiation of the market adjustment
condition (5.9), so that
p*+Ei = ET. (5.21)
Substituting (5.11) and (5.12) in (5.21) then furnishes
A Tj-AhT:
p* = 1 (5.22)
a1 +ah-l
The denominator of (5.22) is positive for the foreign trade market
to be stable. Hence the relationship between p and the tariff is
determined by the signs of A1 and Ah, both of which, in the absence
of inferior goods or a positive initial tariff, are positive. It is now a
simple matter to observe that, unless one of the a1 and ah is infinity,
an increase in the tariff rate by any country results in an improve-
ment in its terms of trade. With the home country, for example, an
increase in its tariff (Tt > 0, Tj = 0) will lead to a decline in p and
hence to a shift in the terms of trade in its favour. This result is
attributable to the existence of the negative relationship between
E2 and r;, when terms of trade are kept constant. From (5.17) it is
apparent that with p* = 0, E!/Tt < 0 irrespective of whether
initially the tariff is positive or zero, provided, of course, that inferior
goods are non-existent. Since at constant terms of trade an increase
in the tariff results in a decline in the demand for imports, the inter-
national relative price of the importable good must eventually
decline to restore equilibrium in the foreign trade market. However,
if the exportables are inferior in social consumption or if the im-

t This merely confirms our result derived in the previous chapter, that with initial
free trade and given terms of trade, /aissez-faire is the optimal policy.
t It may be recalled here that ET = £!1 , and that e1 , s1 and m1 carry the same mean-
ing in the foreign country as their counterparts in the home country.
The Theory of Nominal Tariffs 117

portables are the luxury goods, so that mh > 1, it is possible that a


rise in the tariff rate may lead to a rise in the demand for imports.
Evidently, this possibility requires that
11. 1 + th
mh >- = -- > (5.23)
th th

which implies that mh > 1 is the necessary condition. It is obvious


now that if inequality (5.23) is satisfied, a rise in the rate of tariff will
result in a deterioration in the terms of trade of the tariff-raising
country.
However, (5.23) is a condition for instability. As Kemp ([5] p. 37)
has pointed out, the system's reaction to the increase in the tariff
would be explosive. For an increase in the rate of tariff would
generate a further rise in imports and hence in tariff proceeds, which
in tum would give rise to a further increase in import spending and
so on. Hence if we rule out instability, a tariff increase must cause
an improvement in the terms of trade. Finally, it may be emphasised
again that if initially there is free trade, the introduction of the tariff
always results in a shift in the terms of trade in favour of the tariff-
imposing country.

H'
\
\ H
I
I
I
JE'

8 Q

Figure 5.1
118 Studies in the Pure Theory of International Trade

Given that tariffs normally result in an improvement in the terms


of trade, the improvement will be smaller, the greater are the
elasticities of demand for imports in the two countries. In the
limiting case where the foreign offer curve is perfectly elastic so that
a1 = oo, there will be no change in the terms of trade at all.
The implications of tariffs on the terms of trade are geometrically
depicted in Fig. 5.1, where the introduction of the tariff by the home
country shifts its offer curve from OH to OH', and as a result the
terms of trade move in its favour from the slope of OE to that of OE'.
Notice that at initial terms of trade reflected by the slope of OE, the
imposition of the tariff results in a decline in the import demand
from EQ to AB.

5.3 Tariffs and the Domestic-Price Ratio


The effects of tariffs on the domestic-price ratio are not as straight-
forward as may appear at first glance. Indeed, until the pioneering
contribution by Metzler [10], the problem was not even considered,
it being taken for granted that the imposition of the tariff would
raise the domestic-relative price of the importable commodity. In the
small country case this is undoubtedly true, for then, with the terms
of trade unaffected by the action of the tariff-imposing country, the
domestic-relative price of the importable good rises in proportion to
the tariff rate. However, when the country possesses monopoly
power and hence the ability to influence its terms of trade, the
domestic-price ratio is influenced by two conflicting forces. On the
one hand, the tariff tends to raise the domestic-relative price of the
importable good; on the other hand, the resultant improvement in
the terms of trade tends to lower it. In view of these countervailing
factors, the derivation of a suitable criterion predicting categorically
the change in the domestic-price ratio consequent upon the tariff is
in order.
Fortunately, with the necessary paraphernalia for analysis already
collected, this is not a difficult task. Suppose that the home country
slightly increases its tariff. We need to obtain an expression for p:
which equals p* + TZ. Substituting for p* from (5.22), we get
The Theory of Nominal Tariffs 119
Substituting for ah and Ah from (5.18) and (5.19), we obtain

(5.24)

With no change in the foreign tariff, so that Tj = 0, and with the


denominator of (5.24) positive to ensure the market stability,
Ph* ~ o, ·f a1 + mh ~ 1
Th* < 1
Th(l-mhth/Th)
< . (5.25)

If initially there is free trade, it is clear that T,(l-mhth/T,) = 1,


so that pt /Tt < 0 only if a1 + mh < 1. In other words, if the sum of
the foreign elasticity of demand for imports and the home marginal
propensity to consume importables falls short of unity, the intro-
duction of the tariff will lead to a decline in the domestic-relative
price of the importable commodity. This is the Metzler paradox.
Alternatively, this paradox requires that a1 be less than the home
marginal propensity to consume exportables ( = 1-mh). Evidently,
the necessary condition for the Metzler paradox in the absence of
inferior goods is that the foreign import demand be inelastic. On the
other hand, if the importable good is inferior, so that mh < 0, the
Metzler paradox may occur even if the foreign import demand is
elastic.
The economic explanation of this result requires consideration of
changes in each country's local markets at constant local prices that
occur as a result of the introduction of the tariff by the home country.
The improvement in the terms of trade resulting from the home
tariff leads to a rise in the home real income, a part of which at the
unchanged Ph goes into the increased demand for the home export-
able commodity. This latter effect is reflected by 1-mh. There is no
other change in the home country, because with ph held constant no
substitution effects on the production as well as the consumption
side take place. In the foreign country, by contrast, the improvement
in the terms of trade leads to a decline in the demand for the home
exports. This is represented by a1 . If the increase in the home demand
for its exportable good, l-mh, exceeds the decrease in the foreign
demand for the same good, given by a1 , there occurs an excess
demand for the home exportable good at initial Ph· Equilibrium is
120 Studies in the Pure Theory of International Trade

Figure 5.2

eventually restored by a rise in the domestic-relative price of the


home exportable good, or, what is the same thing, by a decline in
the domestic-relative price of the importable good.
This much has been known since the appearance of Metzler's
path-breaking article in 1949. What has not been realised is that there
are alternative and perhaps weaker necessary conditions available.
All we need to do is to substitute for a1 from (5.20) in (5.25). By
analogy to the breakdown of E!, E!1 can be split into the
coefficients of p* and Tj to obtain
et+st+mtllf
a!= .
1-mfttllf
The substitution of this in (5.25) yields the following conditions
under the initial situation of free trade:

~= ~ 0, if e1 +s1 +m1 +mh ~ 1.

An interesting condition now suggests itself. Evidently, the necessary


condition for the validity of the Metzler paradox is that the sum of the
marginal propensities to consume importables in the two countries falls
short of unity. The condition can alternatively be stated in terms of
one good only. The foreign marginal propensity to consume the first
The Theory of Nominal Tariffs 121

Figure 5.3

commodity (its importable good) should be less than the home marginal
propensity to consume the first commodity (its exportable good).
A further comprehension of these results may be gained from an
examination of Figs. 5.2 and 5.3. Fig. 5.2 depicts the case of an
elastic foreign offer curve, OF, whereas in Fig. 5.3 OF is inelastic in
the neighbourhood of initial equilibrium pointE and the post-tariff
equilibrium pointE', which is generated by the intersection of OF
and the tariff-distorted home offer curve OH', with the pre-tariff
home offer curve given by OH. The post-tariff terms of trade are
given by the slope of OE', and at the new international prices the
foreign country offers 0 R of the second good in exchange for RE'
of the first commodity; the home producers offer RA, so that the
difference AE' is collected by the government as tariff revenue in
terms of the first commodity. The rate of tariff in terms of the first
commodity therefore is AE'/E' R. The domestic-price ratio in the
home country can be computed simply from the formula that

or
122 Studies in the Pure Theory of International Trade

which is nothing but the inverse of the slope of the dotted ray OA. In
Fig. 5.2, where the foreign offer curve is shown to be elastic, OA is
less steep than OE, which indicates the initial terms of trade. This im-
plies that the domestic-relative price of the exportable commodity
has declined as a result of the introduction of the tariff. By contrast,
in Fig. 5.3, where the foreign offer curve is drawn inelastic in the
neighbourhood of equilibrium points, OA is steeper than OE, show-
ing thereby that the domestic-relative price of the exportable good
(importable good) has risen (declined) in the post-tariff equilibrium.
Thus the diagrams demonstrate the necessary condition for the
validity of the Metzler paradox, namely, that the foreign import
demand should be inelastic.

5.4 Tariffs and the Distribution of Income


In Chapter 2 it was established that a rise in the relative price of a
commodity resulted in a rise in the real reward of its intensive
factor, and we termed this result the Stolper-Samuelson theorem.
Although this is essentially the quintessence and more recent inter-
pretation of their results, Stolper and Samuelson formulated their
proposition in terms of the effects of tariffs on the internal distribu-
tion of income. With the help of the Heckscher-Ohlin framework,
where the importable good is intensive in the use of the relatively
scarce factor, they were able to demonstrate that protection raises
the real and the relative reward of the scarce factor. Prior to the
Stolper-Samuelson contribution, the solution to the incidence of
tariffs on factor prices had been thwarted by the so-called 'index
number' problem. The classical economists, who perhaps because
of the intractability of the problem paid little attention to the effects
of the tariffs on income distribution, were able to show only that
protection benefited a factor relatively, though not necessarily
absolutely. By a careful choice of assumptions, Stolper and Samuel-
son [13] solved this problem by showing that protection necessarily
alters the distribution of income in favour of the scarce factor. The
tenor of their argument must have been palpable by now. They
assumed a small country where a tariff necessarily raises the
domestic-relative price of the importable good. Under perfect com-
petition, full employment, constant returns to scale, perfect internal
factor mobility and incomplete specialisation, the rise in this relative
price would result in a rise in the real reward of the factor intensively
employed in the local production of the importable good, and since
The Theory of Nominal Tariffs 123

in the Heckscher-Ohlin model this is the relatively scarce factor,


protection necessarily raises the absolute and hence the relative
reward of the scarce factor. If the Heckscher-Ohlin theory is not
valid, the essential logic of the Stolper-Samuelson argument is not
affected. It is still true that the introduction of the tariff would
necessarily alter the real reward of any factor and hence the com-
plexion of income distribution.
Expressed in this manner, the Stolper-Samuelson theorem is valid
even when the tariff-imposing country possesses monopoly power,
except in the singular case where, as we have seen above, the foreign
elasticity of demand for imports equals the home marginal pro-
pensity to consume importables, so that the domestic-price ratio
and hence the factor rewards remain unaltered.
Over the years since the appearance of the Stolper-Samuelson
contribution, the theorem has been extended to the case of three
commodities and three factors by Kemp and Wegge [6], but beyond
the three-by-three case the difficulty in defining the factor intensities
of the commodities stands in the way of deriving any simple and
meaningful conclusions.

5.5 The Optimum Tariff


We have already seen above that the terms of trade move in favour
of the tariff-imposing country if the latter possesses monopoly
power in trade. This has interesting ramifications for social welfare,
which now gets subjected to two conflicting pulls. On the one hand,
the improvement in the terms of trade consequent upon the tariff
tends to raise social welfare, and this welfare gain increases with the
tariff; on the other hand, the introduction of the tariff impairs pro-
ductive efficiency and tends to lower welfare, and this welfare loss
increases with the rise in the tariff. It follows then that there is a
certain rate of tariff at which social welfare is maximised. This tariff
is the optimum tariff. Evidently, then, free trade is not the optimal
policy in the presence of variable terms of trade. The reason can be
intuitively appreciated. When the terms of trade are constant, free
trade equalises the foreign- and the domestic-price ratios. Under
these conditions, the foreign-price ratio (FP) reflects both the foreign
marginal rate of transformation (FMRT) and the foreign average
rate of transformation (FART). On the other hand, the domestic-
price ratio (DP) under competitive conditions reflects the domestic
marginal rate of transformation (DMRT) and domestic marginal
124 Studies in the Pure Theory of International Trade

rate of substitution (D MRS). Hence under free trade with constant


foreign prices

FMRT = FP = DP = DMRT = DMRS


which constitutes a necessary condition for welfare maximisation.
However, when terms oftradearevariable, FP = FART =F FMRT.
Hence under free trade DMRTand DMRS are equalised to FART
but not to F M RT. Only a tariff, which creates a divergence between
FP and DP, can effect such equalisation. This solution derives from
the general principle that the cure should be directed to where the
malady is. Since the violation of the optimality conditions stems
from the variability of the terms of trade, the cure should aim at
eliminating the distortion occurring in the foreign trade market.
Hence any policy which concentrates on the local markets like
production or consumption subsidies (taxes) will be inferior to an
appropriate tariff in the presence of variable terms of trade.
The optimum tariff formula can be derived simply by equating dY
to zero. From the balance-of-payments equilibrium condition,
pd£2 = dElf- E 2 dp. Substituting this in (5.16), we get
dY = (T,-l)d£ 11 - T,E 2 dp.

Using (5.12), we then obtain

dY = Elf(Th-1)(a1p* -A1 Tj)- ThpE2 p*


= E11 [p*{th(a1 -1)-1}-thA1 Tj]. (5.26)

For the given foreign tariff ( Tj = 0), the home welfare is maximised
It the coefficient of p* equals zero. The optimum tariff (t 0 ) formula
is then given by

(5.27)

It is evident that the existence of the pos1t1ve optimum tariff


(t 0 > 0) requires the foreign offer curve to have some range where
a,> l.
A simple geometric derivation of this formula is depicted in
Fig. 5.4, where U 1 , U2 , U3 , etc., curves represent the levels of social
The Theory of Nominal Tariffs 125

Figure 5.4

welfare in increasing order ;t E is the point of equilibrium under


free trade, and the level of welfare is indicated by U 1 which is
tangential atE to the free trade terms of trade line OE. The intro-
duction of the tariff by the home country moves the trade equilibrium
point along the foreign offer curve, OF, and as a result trade in-
difference curves also move upwards. For example, if the level of
tariff is such as to make the new home offer curve intersect OF at R,
the trade indifference curve moves to U2 , which, however, does not
represent the highest attainable level of welfare, simply because at
R the trade indifference curve intersects the foreign offer curve, so
that the DMRS still does not equal FMRT, which in tum is reflected
by the slope of OF. Maximum welfare is achieved when the trade
indifference curve, U3 , touches OF at G, so that FMRT is equated
tHere U1 , U2 and U3 are trade indifference curves; for further details on their
derivation, see Johnson [3].
126 Studies in the Pure Theory of International Trade

to DMRS. The domestic-price ratio is given by the slope of OF, AG,


so that Ph = BG/ AB, whereas the foreign-price ratio is furnished by
p = BGjOB, and since Ph = p(l + t 0 ),
BG BG
AB = OB (l +to)
OB OA OA
t0 = --1
AB AB OB-OA
l

Note that the optimum tariff is not the one that is associated with the
most favourable terms of trade. For example, if OH' intersected OF
at Q, the terms of trade would be more favourable than those
furnished by G; but the concomitant level of welfare would be
suboptimal.
So far we have assumed that the foreign country takes no action in
response to the tariff imposed by the home country. The more
realistic situation is where both countries attempt to reap gains from
their monopolistic positions in trade. When each country is ready to
retaliate against the imposition of the tariff by the other, the optimum
tariff argument loses much of its appeal. Both countries obviously
cannot gain by imposing tariffs, and if retaliatory measures con-
tinue, both may lose in the final equilibrium. In all likelihood both
countries would then be better off with the free trade policy,
although, as Johnson [3] has shown, one country under certain
conditions may be able to benefit by restricting its trade at the
expense of the other even if the latter takes retaliatory measures.

5.6 The Symmetry between the Import and Export Taxes


It has been shown by Lerner [7] that the effects of import and
export taxes are symmetrical, provided the government disposal of
the tax proceeds is not affected by the type of the tax. This follows
directly from the barter character of our model and is readily
explained with the help of Fig. 5.2, where, it may be recalled, the
imposition of the tariff rate equal to AE' IRE' shifted the home offer
curve from OH to OH'. The foreign producers in the case of the
home import tax offer AB of the second commodity and receive RE'
of the first, whereas the home producers offer RA of the first com-
The Theory of Nominal Tariffs 127

modity, the difference AE' being the tariff revenue collected by the
government. Let us now consider an export tax imposed at a rate
equal to AE'/RE'. The foreign importers now receive RE', the home
exporters offer RA in exchange for AB and the customs revenue in
terms of the first commodity again equals AE'. Hence the extent of
the shift of OH is the same in both types of taxes. Evidently, then,
the terms of trade, imports, exports, domestic production and con-
sumption are the same in both cases; the only difference is that in
one case the tax is paid by the importers, whereas in the other it is
paid by exporters.
A simple algebraic proof of Lerner's symmetry theorem is given
as follows. Let te be the export tax imposed by the home country, so
that

which is clearly the same as the relative price structure associated


with the import tax, namely,

given, of course, that th = te. This implies that the home outputs
are also the same in both cases. However, the demand for each
commodity depends not only on relative prices, but also on real
income. Hence for the symmetry theorem it is necessary to show that
real incomes are also the same under both taxes. Note that, in the
case of the tariff, real income expressed in terms of local prices is
given by
Y = X1 +PhXz +thpEz
whereas that in the case of the export tax is given by

It is interesting to note that the tariff revenue in the two cases looks
different, but is in reality the same. The apparent difference is attri-
butable to the choice of the numeraire. Therefore, after adjusting for
the different absolute prices, thp£ 2 = teEd(!+ te). Hence Y = Y,,
which, with similar relative price structure, implies that production,
consumption and imports are the same under both types of taxes.
128 Studies in the Pure Theory of International Trade
REFERENCES
[I] Graham, F. D., 'Some Aspects of Protection Further Considered', Quarterly
Journal of Economics, XXXIX (Feb 1925) 324-30.
[2] - - , The Theory of International Values (Princeton U.P., 1948).
[3] Johnson, H. G., International Trade and Economic Growth (London: Allen &
Unwin, 1958) chap. ii.
[4] Jones, R. W., 'Tariffs and Trade in General Equilibrium: Comment', American
Economic Review, LIX (June 1969) 418-24.
[ 5] Kemp, M. C., The Pure Theory of International Trade (Englewood Cliffs, N.J.:
Prentice-Hall, 1964).
[6] --, The Pure Theory of International Trade and Investment (Englewood Cliffs,
N.J.: Prentice-Hall, 1969).
[7] Lerner, A. P., 'The Symmetry between Import and Export Taxes', Economica,
III (Aug 1936) 306-13.
[8] Marshall, A., The Pure Theory of Foreign Trade (London: London School of
Economics and Political Science, 1949).
[9] - - , Money, Credit, and Commerce (London: Macmillan, 1923).
[!OJ Metzler, L. A., 'Tariffs, the Terms of Trade, and the Distribution of National
Income', Journal of Political Economy, LVII (Feb 1949) 1-29.
[II] Mill, J. S., Principles of Political Economy (London: Longmans, Green, 1909).
[12] Siidersten, B., and Vind, K., 'Tariffs and Trade in General Equilibrium',
American Economic Review, LVIII (June 1968) 394-408.
[13] Stolper, W. F., and Samuelson, P. A., 'Protection and Real Wages', Review of
Economic Studies, IX (Nov 1941) 58-73.
6 Economic Expansion and
the Terms of Trade

6.1 Introduction
The analysis of interrelations between economic expansion and the
terms of trade cannot boast of a long, uninterrupted history. Here
and there one may come across a few scattered references to terms
of trade in studies of the incidence of technological innovations on
the growing country's net gain from growth in the works of Mill
[18], Edgeworth [9], Bastable [2] and Ohlin [19]; of the desirability
of the policy of protection in the wake of technical improvements
in what is known as the German tariff controversy ;t of the celebrated
transfer problem sparked primarily by Thornton [22] and Hume
[14] and subsequently resurrected by Taussig [21] and Keynes [17]
(in the famous debate over the German reparations problem);
finally, of the impact of international capital movements featuring
in the works of Fanno [10], Iverson [15] and Ohlin [19]. But the
avowed application of the tools of general equilibrium analysis to
the diagnosis of the behaviour of the terms of trade consequent upon
growth is only a post-war phenomenon. Whatever ingenuity the
problem of shifts in inter-country commodity prices elicited from
the economist in the pre-war era was, with a few exceptions,t inci-
dental; it arose not because of an interest in the application of
economic theory to the problem of economic development, which
itself never figured prominently in neo-classical writings, but was
due to the economist's preoccupation with some· other issues of
current importance wherein considerations of the terms of trade
were unavoidable.
Like many other fields of economics, trade theory has welcomed
the dent made in it by the theory of economic growth pioneered by

t See, for example, Wagner [23) and Dietzel [8).


t Mill's work [18) constitutes an important exception.
130 Studies in the Pure Theory of International Trade

Sir Roy Harrod. t The evolution of this branch of theory has been
further accentuated by two other burning issues: (1) the problem of
international liquidity, of which the apprehensions concerning the
long-run shortage of dollars in the world markets constituted but
one phase, for since 1958 the dollar paucity has turned into the
dollar glut; and (2) the question of the secular deterioration in the
terms of trade of the countries producing primary products.
Until the middle of the 1950s it was quite fashionable among
trade theorists to argue that the dollar difficulties of the post-war
period constituted a legacy of the pre-war years; that there was in
fact a long-run tendency for the U.S. balance of payments to run into
surplus, only to create cumulative difficulties for the non-dollar
world. A variety of reasons for the tendency were advanced. Balogh
[1] and others took the faster growth of the U.S. productivity than
that in other countries as granted, with the result that the increasingly
cheaper U.S. exports were displacing the exports of other countries.
Hicks [13] suggested, in addition, that innovations in the U.S.
tended to be 'import-biased', that is, they were concentrated in those
goods which competed with American imports.
So overwhelming was the economist's obsession with the secular
shortage of dollars until even the late 1950s that Keynes's sole dis-
sension that 'in the long run more fundamental forces may be at
work, ... tending towards equilibrium', so that 'the chances of the
dollar becoming dangerously scarce in the course of the next five to
ten years [wereJ not very high' ,:j: proved only a distant cry in the
wilderness. From the strength of hindsight Keynes once again proved
right, although the affirmation of his prophecy came almost a decade
after his demise. As of 1950, the U.S. balance of payments has dis-
played chronic deficits; the entire situation has been so transformed
that strong equivocations are now cast in some quarters as to the
very stability of the dollar. The post-war international monetary
crises, climaxed by the recent Smithsonian accord leading to the

t To be sure, Harrod [12] did not address himself to the trade problems arising
from growth; but he did succeed in opening new vistas for the trade theorist to
explore. His references to growth stemming from technical progress and capital
accumulation may be deemed partly responsible for the assiduity with which the
economist has engaged himself to unravel the interrelation between growth and the
terms of trade.
t Quoted from Clement et a/. ( [ 5] p. 352). The bracketed expression is theirs.
Economic Expansion and the Terms of Trade 131

realignment of world currencies in 1971, demonstrate how the


strength of the dollar has been steadily eroded over the last twenty-
five years.
Similarly, in explaining the ubiquity of abject poverty in the
underdeveloped countries, the emphasis placed by Prebisch [20]
and others on the deterioration in the terms of trade of countries
exporting primary products due to the presence of strong mono-
polistic elements in advanced Western economies has also turned
out to be misplaced. The underdeveloped world has been deriving
large intangible gains from trade with the advanced nations, although
these benefits, direct or indirect, cannot fully, on account of their
complexity, be incorporated in the theorist's abstract model. All
that can be conceded in fairness to the adherents of the Prebisch
hypothesis is that the distribution of the gains from trade between
the underdeveloped and the developed nations may not have been
equitable.
These two hypotheses - the secular shortage of the dollar and the
secular deterioration of the terms of trade of the underdeveloped
areas- have thus transpired to be based on specious arguments. The
dramatic investigations and research aroused by these issues have
turned out to be their own nemesis. A large section of economists
have become disenchanted with these views and doubtful of their
empirical validity.
'But', as Corden rightly contends, 'out of the discussion and
especially out of Hicks's argument that technical progress in the
U.S. is import-biased and that this has an adverse effect on the
terms of trade of other countries, has come an advance in theory of
lasting value.' ([6], p. 36).
The purpose of this chapter is to survey some of the major issues
that have been raised in this field. Our task in this respect is simpli-
fied by recalling that some of the equations and results developed in
Chapter 2 can be used directly in our analysis here.

6.2 Economic Expansion


For the sake of simplicity and without loss of generality, we assume
that only the home country experiences growth, whereas its trading
partner, the foreign country, remains stationary. This implies that
the foreign offer curve is given, whereas the home offer curve may
shift its position in response to economic expansion. For the time
132 Studies in the Pure Theory of International Trade

being it is immaterial to our analysis just how this economic growth


may come about; later we shall have occasion to identify major
factors that may account for growth of one type or another.
Let G denote the expansionary (growth) agent. In addition to the
terms of trade, each country's import demand should now become
a function of the growth agent, but since the foreign country is
assumed to be stationary, only the home demand for imports will
be related to G. That is, the balance of payments equilibrium
equation will now look as follows:
(6.1)
Differentiating (6.1) totally with respect to G and utilising the fact
that in equilibrium £ 2 = Etfp, we obtain

where a1 and ah are as defined in the previous chapter. Now com-


modity units can be so chosen that pis initially equal to unity. The
change in the terms of trade consequent upon growth can then be
written as
dp o£2/oG
(6.2)
dG (a1 +ah-l)E2.
Stability in the foreign trade market demands that a1 + ah- 1 > 0.
Since £ 2 > 0, it follows from (6.2) that dpjdG ~ 0 if oE2joG ~ 0.
In other words, the sign of the change in the terms of trade depends
solely on the sign of the change in the home demand for imports.
Since o£2/oG represents the partial derivative of £ 2 with respect to
G, and since £ 2 is a function only of p and G, the evaluation of the
change in the home demand for imports arising from growth must
be done at constant terms of trade. The economic rationale behind
such evaluation is that in order to determine the impact of growth on
the growing country's terms of trade, one first needs to ascertain the
impact of growth on the growing country's demand for imports at
unchanged terms of trade. This determines the manner in which
the home offer curve will shift in response to growth. The terms of
trade of the growing country will eventually deteriorate, improve or
Economic Expansion and the Terms of Trade 133

H
R -----------

Figure 6.1

remain unchanged depending on whether its demand for imports,


at constant terms of trade, has increased, decreased or remained
unaltered as a result of growth. For example, in Fig. 6.1 the home
offer curve, 0 H, shifts outwards to 0 H', the home import demand,
at old terms of trade reflected by OE, rises from OQ to OR and, as
a consequence, its terms of trade deteriorate to those indicated by
the slope of OE' in the new equilibrium. Now
(6.3)
or
oE 2 = oD 2 • oY _ oX2 = oY(mL-zL)
oG oY oG oG oG " " ( 6 .4)

where mh = poD 2 /o Y = oD 2 /o Y is the marginal propensity to con-


sume importables in the home country and zh = (oX2 /oG)j(oYjoG)
measures the rate of change in the output of importable goods as a
proportion of the rate of change in income. Let
134 Studies in the Pure Theory of International Trade

Then m's and z's have the following properties:

(6.5)

1. (6.6)

A number of definitions have sprung up in the contemporary


literature which specify the nature of growth depending upon the
various possible values taken by z 1 and zh. If zh < 0, z 1 > 1: this
case is traditionally defined in Corden's [7] terminology as the case
of an 'ultra-export-biased' growth; if zh > 1, z 1 < 0: this case is
termed as the case of an 'ultra-import-biased' growth. In other
words, by an ultra-export-biased growth is meant one that results
in an absolute decline in the domestic production of the importable
commodity when terms of trade are kept constant, so that growth
must perforce lead to a rise in exports if inferior goods are absent
and hence to an eventual deterioration in the terms of trade.
Similarly, by an ultra-import-biased growth is meant one that results
in an absolute decline in the production of the exportable com-
modity at constant terms of trade, so that exports must decline, and
hence in the new equilibrium the terms of trade must improve. It
may be noted here that only ultra-biased types of growth admit of
determinate results; other types where zh and z 1 may be positive
fractions do not yield any a priori conclusions. Substituting (6.4) in
(6.2), we obtain
dp (oYjoG)(mh-zh)
(6.7)
dG Eia1 +ah-l) ·
This last equation furnishes the standard formula which will come
to our assistance throughout in determining the effect of growth on
the terms of trade. One glance at (6. 7) is sufficient to reveal that
dpjdG ~ 0, if mh ~ zh.
Matters may be further simplified by observing that, from our
assumption of the absence of inferior goods, mh is only a positive
fraction. Thus if zh happens to assume either a negative value or a
value greater than unity, the implications of growth on the terms of
trade can be immediately predicted. For example, if zh comes out to
be negative, mh must exceed zh, so that the home terms of trade must
deteriorate (i.e. its price of imports relative to the price of its exports,
p, must rise); if, on the other hand, zh exceeds or equals unity, it
Economic Expansion and the Terms of Trade !35

must also exceed mh, so that p must fall, or to say the same thing in
different words, the home terms of trade must improve.
From (6.4) three possibilities emerge, to wit, mh ~ zh. These are
depicted by Fig. 6.2. Growth is represented by an outward shift of
the transformation curve from H H' to GG'; changes in the outputs
and the demands for X 1 and X2 are evaluated at unaltered terms of
trade, as may be evident from the fact that the pre-growth terms of
trade given by FP are the same as those in the post-growth situation
given by FP' (FP is parallel to FP'). As a consequence of growth, the
production point shifts from P toP'. Now if the consumption point
shifts from C to C', that is, if PP' is parallel to CC', zh equals mh. The
home offer curve does not shift in this case, so that the terms of
trade will remain unaltered. If the consumption point moves to any-
where towards the left of C', say C, zh falls short of mh, so that the
terms of trade will move against the home country; if to anywhere
towards the right of C', say C", zh exceeds mh and the terms of trade
will switch in favour of the home country.

FP 1

x,
Figure 6.2
136 Studies in the Pure Theory of International Trade

6.3 Immiserising Growth


The case in which the terms of trade deteriorate while the foreign
country benefits from the home expansion is of special interest. For,
as first noted by Edgeworth [9], it suggests the possibility that the
home country might find itself in a worse position after growth than
before if the worsening in its terms of trade is large enough to wipe
out the gain in its output as a result of growth. What needs to be
comprehended here is the distinction between growth in the form of
expanded output against an increase in its real income. If the terms
of trade remain constant, the growth in the home country output
will be the same as the growth in its real income; if the terms of
trade improve, the rise in the home real income will exceed the rise
in its output; if they deteriorate, the rise in the home real income will
fall short of the rise in its output, the changes in the home output in
all three cases being assessed at constant commodity prices. Is it
then possible that the home real income may actually decline as a
result of growth? The following criterion for the change in real
income will determine the answer.
The utility function in the two-good case is given by
U = U(D 1 ,D2)
so that
l dU dD 1 dD2
U1 dG = dG + p dG . (6.8)

From the budget constraint


Dt +PD2 = X1 +PX2
and

so that (6.8) can be written as


dY dX 1 dX2 dp
dG = dG +p dG -E2 dG (6.9)

where, as before, the change in welfare is assumed to measure the


change in real income. In the presence of growth
X;= X;(p,G) where i = 1,2
Economic Expansion and the Terms of Trade 137

so that

Remembering that (oXtfop+poX2 jop) = 0, (6.9) can be written as

(6.10)

The first two terms in the right-hand side of (6.10) represent the
change in the value of production resulting from growth at constant
terms of trade, and hence equal o YjoG. Therefore

dY _ oY E dp
dG- oG- 2 dG. (6.11)

Substituting for dp/dG from (6.7) we obtain

(6.12)

Since fJ Y/oG > 0, it is clear from (6.12) that for the home real
income to decline as a result of growth

or, from (6.6),


(6.13)

In other words, if the sum of the home country's marginal propensity


to consume importables and the rate of change in the output of its
exportable goods as a proportion of its change in its total output
exceeds the sum of the two countries' price elasticities of demand
for imports, the home real income will actually decline as a result of
economic expansion. The criterion can be further concretised by
utilising Slutsky's decomposition of price effects into income effects
and substitution effects. A price elasticity of demand can always be
written as the sum of a 'compensated' (i.e. pure substitution) elas-
ticity and an income propensity. Let a~ be the income-compensated
138 Studies in the Pure Theory of International Trade

elasticity of demand for imports (i.e. an elasticity compensated by


changes in income) in the home country; then

(6.14)t

Utilising (6.14), inequality (6.13) may be written as

(6.15)

The presentation of the criterion for what Professor Bhagwati [3]


calls 'immiserising growth' in the form (6.15) helps in singling out
the key factor that may be accountable for the home country's im-
miserisation due to growth. Evidently, the key factor, given the
foreign total elasticity and the home compensated elasticity of
demand for imports, lies in the pattern of growth. For if the latter is
concentrated heavily in the exportable industry, the value of z 1 may
be large enough to satisfy inequality (6.15), in which case the home
real income would decline.
This somewhat paradoxical result is illustrated diagrammatically
in Fig. 6.3. The level of home welfare in the pre-growth situation is
given by the social indifference curve U1 which lies on line (1); if
the terms of trade remain constant after the outward shift of the
transformation curve, welfare improves to U2 which lies on line (2),
parallel to line (1), the rise in real income being commensurate with
the rise in output; if the terms of trade improve and are indicated
by the slope ofline (3), home welfare improves further to u3; finally,
if the terms of trade deteriorate, the home welfare declines below
U2 , and if the deterioration in the terms of trade is as high as the one
reflected by the switch to line (4), welfare declines to U0 which is
actually below the original, pre-growth level of welfare. This is
immiserising growth.

t Differentiating £2 = E 2 (p, Y) totally with respect top, we obtain


d£2 o£2 o£2 dY
-=-+--.
dp op ar dp
But (iJE 2/o Y) = oD 2/o Y and dYjdp = - E 2. Therefore
!!_ d£2 = !!_ c£2 _ poD2
E 2 dp E 2 cp iJY
or
Economic Expansion and the Terms of Trade 139

x,
Figure 6.3

6.4 Factor Accumulation


It is high time that we introduced the specific sources of economic
expansion. We begin with factor accumulation, whose implications
for the closed-economy prices were first explored in the second
chapter. The pioneering contribution in this field is credited to
Rybczynski, whose theorems have already been expounded. In order
to determine the impact of factor accumulation on the terms of
trade, we first need to derive the effects of factor growth on the two
outputs at constant terms of trade. Such expressions are given by
(2.37) and (2.38). Thus

(2.37)
140 Studies in the Pure Theory of International Trade

and
n _ ).,L1K* -).,K1L* (2.38)
2 - IA.I
where, from (2.17) and (2.17*),

I)., I = L1L2
A.L1A.K2-).,K1A.L2 = A.Ll-A.Kl = A.K2-).,L2 = LK (k2-kl).

To begin with, let us assume that only the labour force grows, while
the capital stock remains constant, so that L* > 0, K* = 0 and
hence the growth agent (G) is identified by L. With this in mind,
(2.37) and (2.38) can be simplified to

ax 1 XtK2
(6.16)
aL L 1L 2(k 2-k 1)
and
ax2 -X2Kt
aL (6.17)
= L 1L 2(k 2 -k 1)"
The change in national income at constant terms of trade, with
p = l initially, is then given by
ay axl ax2 x1K2 -X2K1 w
-=-+-= =- (6.18)t
aL 8L 8L L1L2(k2 -kt) Pt
so that

(6.19)

where z L measures the change in X 2 as a proportion of the change in


national income at constant terms of trade. Evidently, the place of
zh in (6.7) is now taken by (zL). It can be readily seen that zL is either
negative or greater than unity, given of course that factor intensities
in the two industries differ to ensure that the denominator of (6.19)
is non-zero. For example, if X 2K 1 > X 1K 2, or (KtfX 1) > (K2/X2),

t Solving the two price equations in Chapter 2 and remembering that p = I ini-
tially, we obtain
Economic Expansion and the Terms of Trade 141

then, with p initially unity, it means that the relative share of capital
in the first commodity exceeds that in the second commodity, that is,
the first commodity is capital-intensive relative to the second. With
the denominator of (6.19) positive, zL > 1; on the other hand, if
this denominator is negative, that is, if the first commodity is labour-
intensive relative to the second, zL < 0. Thus it is clear that the
economic expansion resulting from labour growth alone is ultra-
biased. It is ultra-import-biased if the importable good (the second
good) is labour-intensive, and ultra-export biased if it is capital-
intensive relative to the other commodity. The terms of trade facing
the growing home country will eventually improve in the former
case but deteriorate in the latter case. The reason follows directly
from the Rybczynski theorem, according to which an increase in the
supply of labour alone results at constant terms of trade in a rise in
the output of the labour-intensive commodity at the expense of the
output of the other commodity. Hence if the labour-intensive com-
modity happens to be the importable good, labour expansion is
ultra-import-biased; ifthe exportable good, it is ultra-export-biased.
Similar results can be obtained if it is capital which is the expand-
ing factor while the labour force remains unchanged. In other words,
economic expansion is necessarily ultra-biased if only one factor
experiences an increase in supply.
The geometric exposition of the effects of an increase in the supply
of one factor on the two outputs at constant commodity prices was
presented in Chapter 2 in terms of the box diagram. Here we utilise
an alternative diagrammatic technique which turns out to be of
great help in the ensuing discussion of technological change.
Consider Fig. 6.4, where AB gives the wage/rental ratio, x 1 and
x 2 are the unit isoquants of the two commodities with their points
of equilibrium at v and s respectively, and point E indicates the
factor-endowment ratio (K/L). For the sake of simplicity, the dia-
gram shows that the aggregate output in the economy equals one
unit of the first commodity (Ov) plus one unit of the second com-
modity (Os), and that since the unit isoquants lie on the same isocost
line (AB), the commodity-price ratio equals unity. It is worth point-
ing out that a good grasp of the way this diagram is drawn is essential
for comprehending the present and subsequent analysis. For
example, note that the level of the two outputs is ascertained by
drawing from E lines parallel to Ok 1 and Ok 2 , the slopes of which
indicate the capital/labour ratios in the two industries. Point E,
142 Studies in the Pure Theory of International Trade
K

L
Figure 6.4

which may be construed to have been determined by the summation


of the two vectors (Os and Ov), reflect that the economy's factor
endowment is fully absorbed by the unit level of output of each
commodity.t
Now suppose there occurs an increase in the supply of capital
alone, such that the factor-endowment point moves vertically from
E to D. We seek to explore the direction of change in the two outputs
in the new situation, keeping the terms of trade and hence k 1 and
k 2 constant. If we draw from D lines parallel to k 1 and k 2 (so that
Dn is parallel to k 2 and De to k 1 ), it can be observed that the output
of X 1 rises from Ov to On and that of X 2 declines from Os to Oc.
This is then an alternative proof of the Rybczynski theorem.
So far we have confined our analysis to those cases where the
supply of only one factor increases while the supply of the other
factor remains unchanged. The more realistic case is where both
factor supplies expand and contribute to a country's economic
growth. The effects on the two outputs in the latter situation depend

t This is assumed purely for diagrammatic simplicity.


Economic Expansion and the Terms of Trade 143

not only on the inter-industry ranking on the basis of factor inten-


sities, but also on the relationship between each industry's capital/
labour ratio and the incremental capital/labour ratio, dKjdL. For
example, suppose that the first industry is capital-intensive relative
to the second, as is the case in Fig. 6.4. In this case 1;.1 < 0, so that
from (2.37)

only if ).K 2 /AL2 ~ K*/L*, or k 2 ~ dKjdL. In other words, if the


capital/labour ratio of the labour-intensive commodity (X2 ) exceeds
the incremental capital/labour ratio, the output of the capital-
intensive commodity (X1 ) will decline. On the other hand, from
(2.38),

only if K*/L* ~ ).Ktf).Lt, or dKjdL ~ k 1. In other words, if the


incremental capital/labour ratio exceeds the capital intensity of the
capital-intensive commodity, the output of the labour-intensive
commodity will decline. It follows that, if the incremental capital/
labour ratio lies between the capital/labour ratios in the two indus-
tries, the output of both commodities will rise. Indeed, if both
factors grow at the same rate, both outputs will rise at this rate. This
is clear from the fact that, with K* = L *,

Xi= X!= L*
because it may be recalled that

These results will gain further clarification from Fig. 6.5, which
is constructed on the same principle as Fig. 6.4. It can be observed
that if factor growth follows the growth path sE, parallel to k 1 ,
there will be no change in the output of X 2 ; this depicts the case
where dK/dL = k 1 . For example, if the factor-endowment point
shifts from E to G, the output of X 1 rises from Ov to On, without
any change in the output of X 2 (equal to Os). It follows, therefore,
that if factor growth were to follow a path steeper than sE, sayER
144 Studies in the Pure Theory of International Trade
/(

Figure 6.5

(here dK/dL > k 1 ), the output of X 1 would again rise to On, but
that of X 2 would decline from Os to Oc. Similarly, it can be easily
demonstrated that if factor growth followed the growth path EP
(i.e. dKjdL = k 2 ), the output of X 2 would rise without any change
in the output of X 1 ; if it were to follow a path flatter than
vE(dK/dL < k 2 ), the rise in the output of X 2 would be accompanied
by a decline in the output of X 1 • Finally, if the growth path followed
by the factors is OE, that is, if the factor-endowment point shifts
from E to Q, the output of both commodities will rise in the same
proportion; for the output of X1 rises to On and that of X 2 toOk
and it can be easily seen that sk/Os = vn/Ov.
The impact of growth in both factors on the home terms of trade
can be determined as before. It may be observed that only those
cases admit of a priori conclusions where dK/dL ~ k 1 , or k 2 ~
dK/dL; in other cases results will depend on many additional factors
such as the magnitude of the home marginal propensity to consume
importables, etc.
Economic Expansion and the Terms of Trade 145

6.5 Technological Change


Another major source of economic expansion is the occurrence of
technical improvements in one industry or both. Here again the
basic groundwork has already been laid in Chapter 2. As usual, we
wish to determine the ramifications of technical advance on the two
outputs at constant terms of trade, assuming that the other major
source of expansion, namely factor supply, is unchanged. This last
assumption is made merely to simplify the analysis.
Various types of technical improvements in the Hicksian sense
were introduced in Chapter l. These definitions will now be slightly
modified to suit the two-good character of our model. We continue
to define neutral technical progress in any commodity as one which
leaves the capital/labour ratio of the commodity unchanged at the
old wage/rental ratio. However, instead of the labour-using or
labour-saving nomenclature that we used earlier, we introduce two
different definitions of non-neutral improvements, namely, the
'intensive-factor-using' and the 'intensive-factor-saving' types of
technical advance. t In the former case, technical progress in a com-
modity promotes a relatively larger use of its intensive factor,
whereas with the latter the relative use of the intensive factor declines
at the original wage/rental ratio. For example, suppose the
capital/labour ratio in X 1 declines at the old wage/rental ratio owing
to the technical advance. If X 1 is capital-intensive, the technical
advance is intensive-factor-saving; if it is labour-intensive, the
technical advance is intensive-factor-using.
To be specific, let us assume that technical progress occurs only
in one industry, say X 2 • Let us now go back to Chapter 2 and rewrite
in the following some of the equations presented there:
A.L1X!+A.L2X! = 1tL+fh(w*-r*) (2.23)
A.K1XT+A.K2X! = 1tK-PK(w*-r*) (2.24)
OL1w*+0K 1r* = pf (2.25)
OL2w*+OK2r* = P!+1t2 (2.26)
where PL and PK are positive, and
1t; = A.il B;~ + A.; 2 B;~ (i = L,K)
1ti = ()LiBti+OKiBki (J = 1,2)

t This nomenclature is borrowed from Johnson [16].


146 Studies in the Pure Theory of International Trade

and where BQ indicates the effect of technical progress on the use of


the ith input in the jth output. In rewriting these equations, we
have made use of the assumption that L* = K* = 1t 1 = 0. If the
terms of trade are constant, pf = P! = 0. Remembering this, and
solving (2.25) and (2.26), we obtain

w* = (6.20)

(6.21)

and

w*-r* = -j;,. (6.22)

Substituting (6.22) in (2.23) and (2.24), and solving, we obtain


..ft _ (A.K21tL -AL21tK)!0!-1t2(AK2fh +AL2{JK) (6 .23 )
1 - IA.IIOI
and

(6.24)

In the absence of technical progress in the first commodity,


1tL = A.L 2Bl2 and 1tK = A.K 2B; 2. Making these substitutions in (6.23)
and (6.24), we get
_n _ AK2AL2(Bt2- B:2)!0!-1t2(AK2{JL + AL2{JK)
(6.25)
1 - IA.IIOI

(6.26)

Neutral Technical Progress


A number of results can now be derived from (6.25) and (6.26). Let
us first consider the effects of neutral technical progress, which
implies that Bt 2 = R: 2 .t Since IA.I and 101 have the same sign,
IA.IIOI > 0. It is immediately clear then that XT < 0. The sign of
t The reader is advised to go back to Chapter 2 for verification.
Economic Expansion and the Terms of Trade 147

X! is not evident at first glance, but if we substitute for


!AI = AL 1 AK 2 -AKIAL 2 , (6.26) reduces to

Thus we conclude that neutral technical progress is ultra-biased; it


raises the output of the industry in which it occurs and lowers the
output of the other commodity when terms of trade are kept con-
stant. This result has been derived by many authors, including
Bhagwati [4], Corden [7] and Findlay and Grubert [II] among
others.
The economic rationale behind this result is not so obvious, but
becomes clear when the effects of technical progress in the second
industry alone on factor prices are analysed. The neutral improve-
ment in x2 alone lowers the unit cost of production in x2, thereby
lowering the original commodity-price ratio (p). In order to main-
tain the previous commodity prices, the price of the factor employed
intensively by X2 should rise and that of the factor employed un-
intensively by it should fall. Given that labour enters more intensively
and capital less intensively in the production of X2 , i.e. k 2 < k 1 and
IO! < 0, the wage rate of labour needs to rise whereas the rental of
capital needs to fall so as to restore the pre-improvement com-
modity prices. That this is what happens is clear from (6.20) and
(6.21) where, with 101 < 0, w• > 0 and r• < 0. The rise in the
wage/rental ratio induces the substitution of capital for labour in
both industries, so that the capital/labour ratio increases in both
commodities, which in tum creates a situation of excess demand for
capital and an excess supply of labour. To maintain full employ-
ment, the output of X2 , the industry which can absorb labour more
rapidly, will rise and that of X 1 , the industry which is in a better
position to release capital, will decline. This is how a Hicks-neutral
improvement in any commodity raises the output of the 'progressive'
commodity at the expense of the output of the other when terms of
trade are kept constant. Clearly, then, neutral technical advance in
one commodity alone is ultra-biased, and for this reason it must
have definite implications for the terms of trade, a task which, in
the light of earlier developments in this chapter, we leave to the
reader.
148 Studies in the Pure Theory of International Trade
K

L
Figure 6.6

The analysis presented above can be clarified further by a resort


to simple geometry first developed by Findlay and Grubert (II].
Consider Fig. 6.6, where x 1 and x 2 are our familiar unit isoquants
with national income equalling one unit of X 1 (Ov) plus one unit of
X2 (Os), and the factor-endowment ratio is given by E. Recall that
the level of output in each industry is determined by drawing Ev
parallel to k 2 and Es parallel to k 1 • Since the unit isoquants lie on
the same isocost line AB, the terms of trade (p) equal unity. Now
suppose X2 comes to enjoy a neutral improvement so that its unit
isoquant shifts towards the origin to x2, signifying a reduction in
its unit cost; k 2 is unaltered at the unchanged wage/rental ratio (AB
is parallel to CD) to demonstrate the neutrality of the technical
advance. Now since x2 lies on an isocost line (CD) below the one
(AB) on which lies the unit isoquant of X 1 , the commodity-price
ratio falls below unity. In order to maintain the previous one-to-one
price parity between the two commodities, x2 and x 1 should be
made to lie on the same isocost line. This is accomplished by drawing
C'D' tangential to both x 1 and x2 at q and m respectively. It is note-
worthy that unit outputs in X 1 and X 2 now become Oq and Om
respectively, and that the capital/labour ratio rises in both X 1 and
Economic Expansion and the Terms of Trade 149

X 2 to k'1 and k2. The next step, of course, is to draw from E lines
parallel to the new factor-intensity rays, so that En is parallel to k2
and Ec to k'1 • It may be observed that the output of X 2 rises from
its pre-expansion unit level of output (now equal to Om) to Oc,
whereas the output of X 1 declines from Oq to On. In other words,
neutral technical progress in X 2 is ultra-import-biased. A similar
diagram can be constructed to show that a neutral improvement in
xl alone will be ultra-export-biased.
Intensive-Factor-Saving Technical Progress
Suppose now that the improvement in X 2 , the labour-intensive in-
dustry, is labour-saving. Evidently, this is a case of intensive-factor-
saving technical progress. With technical progress in the second
industry tending to bring about a greater saving in the use of labour
than capital, B!. 2 > B't 2 • The implications for X 1 are evident from
a glance at (6.25); with 101 < 0, the output of X 1 declines even more
than was the case with the neutral improvement in X 2 alone, for
now an additional negative term iOIA.K 2 A.L 2 (Bt 2 - B't 2 ) is added to

L
Figure 6.7
150 Studies in the Pure Theory of International Trade

Xi in (6.25). By the same token, the rise in the output of X 2 exceeds


that which resulted from the neutral improvement. For with
Jl. l < 0, A.L,A.K 2 < A.L 2 A.K 1 , so that with Bh < B! 2 and IOI < 0, the
first term on the right-hand side of (6.26) is positive. If n 2 is the same
in the case of the neutral or the intensive-factor-saving improvement,
that is, the unit cost reduction is the same with both types of improve-
ment, the rise in the output of X 2 is greater now than before. Thus
we conclude that the intensive-factor-saving technical progress in
X 2 alone is more ultra-import-biased. Consequently, it will exert a
more pronounced effect on the terms of trade. Similarly, a capital-
saving improvement in X1 alone will be more ultra-export-biased.
The diagrammatic exposition of this result proceeds in the same
manner as before. Consider Fig. 6.7, where the unit X 2 isoquant
shifts towards the origin to x2 in such a manner that its capital/labour
ratio rises from k 2 to k'2 at the old wage/rental ratio. The immediate
result is a rise in the output of X 2 from its unit level toOk and a fall
in the output of X 1 from its unit level to Ou, as can be seen by
drawing Eu (not drawn) parallel to k2. Once again, in order to restore
the original price parity between X 1 and X 2 , we draw C'D' tan-
gential to both x 1 and x'2 at q and m respectively, with the result that
the capital/labour ratio rises to k'~ in X 2 and to k'1 in X 1 . The next
step is to complete the parallelogram EnOc and to observe that the
output of X 2 rises further to Oc and that of X, declines further to On.
Intensive-Factor-Using Technical Progress
When technical progress in X 2 is intensive-factor-using, the results
are not as categorical as those derived in the two cases given above.
Here B1 2 > B! 2 , and with \til < 0 it is not self-evident from (6.25)
whether Xi is positive or negative. Similarly, from (6.26), the sign
of X! is ambiguous, unless the first term is positive, which implies
that
(6.27)

Since k 2 < k 1 and B! 2 < B1 2 , both k 2 /k 1 and B! 2 /B1 2 are less than
one. The sign of X! is likely to be positive if (i) the capital/labour
ratios in the two industries are markedly different, and/or (ii) the
saving in the use of capital due to technical progress in X 2 is not
much greater than the corresponding saving in the use of labour,
that is, B! 2 / B1 2 is close to unity. On the other hand, if the
capital/labour ratios in the two industries are sufficiently close so
Economic Expansion and the Terms of Trade 151
K

L
Figure 6.8

that k 2 /k 1 is close to unity, and B! 2 /B: 2 falls considerably short of


unity so that inequality (6.27) is violated, X! may be .negative. The
crux of the whole matter is that the intensive-factor-using technical
progress may or may not be ultra-biased and that it may raise the
output of both commodities or only the output of the commodity
which is not subject to technical improvements. These possibilities
are diagrammatically depicted in Fig. 6.8, where at the same factor-
price ratio the capital/labour ratio declines as a result of the shift
of x 2 down to x~. The immediate result in this case is a rise in the
output of X 1 from Ov to Ou, as can be seen by drawing Eu (not
shown) parallel to k~- a case exactly opposite to the case of intensive-
factor-saving improvements;t C'D' is again the tangent drawn

t There appears to be some confusion on this point in contemporary literature.


Findlay and Grubert ([II] p. 119}, for example, assert that an intensive-factor-using
technical progress must in the immediate period lower the output of the commodity
in which it has occurred. This is what is pictured in Fig. 6.8, but it is by no means
certain. It depends on the extent of the technical advance and its factor-using bias.
For example, if technical progress in X 2 were only slightly labour-using, the
immediate effect would be a rise in the output of both commodities. This possibility
is shown in Fig. 6.9.
152 Studies in the Pure Theory of International Trade

common to restore the pre-improvement terms of trade, only to


raise the capital/labour ratio in X 1 to k'1 . Now in the case of X 2 there
are many possibilities. For the new capital/labour ratio in X 2 may
eventually remain the same as the old (i.e. that given by k 2 ), or be
greater or smaller than it; the changes in the output of X 1 and X 2
are then indeterminate. In Fig. 6.8 the final capitaljlabour ratio in
X 2 , k~. though rising above k~. is still below k 2 ; and the output of
X 2 remains the same at its old unit level (it is actually equal to Om,
which anyway represents its unit level now), whereas the output of
X 1 falls from Ou to On, although it is still higher than its unit level.

Figure 6.9

However, we could have obtained an entirely different result just


as well. Consider Fig. 6.9, where the final capital/labour ratio X 2 ,
given by k~. lies above k 2 • As Ec is drawn parallel to k'1 and En to
k~. the output of X 1 can be seen to have declined to On and that of
X 2 to have risen to Oc. Thus an intensive-factor-using technical
progress in any commodity may have any impact on the two
Economic Expansion and the Terms of Trade 153

outputs; its effect on the terms of trade, accordingly, cannot be


prognosticated.

REFERENCES
[I] Balogh, T., The Dollar Crisis: Causes and Cure (Oxford: Blackwell, 1949).
[2] Bastable, C. F., The Theory of International Trade, 4th ed. (London: Macmillan,
1903).
[3] Bhagwati, J., 'Immiserizing Growth: A Geometric Note', Review of Economic
Studies, xxv (June 1958) 201-5.
[4] - - , 'Growth, Terms of Trade and Comparative Advantage', Economia
Internazionale, XII (Aug 1959) 393-418.
[5] Clement, M.D., Pfister, R. L., and Rothwell, K. J., Theoretical Issues in Inter-
national Economics (Boston: Houghton Mifflin, 1967).
[6] Corden, W. M., Recent Developments in the Theory of International Trade
(Princeton: International Finance Section, Princeton University, 1965).
[7] --,'Economic Expansion and International Trade: A Geometric Approach',
Oxford Economic Papers,vm (Sep 1956) 223-8.
[8] Dietzel, H., 'The German Tariff Controversy', Quarterly Journal of Economics,
XVII (May 1903) 365-416.
[9] Edgeworth, F. Y., 'On a Point in the Pure Theory of International Trade',
Economic Journal, IX (March 1899) 125-8.
[10] Fanno, M., Normal and Abnormal International Capital Transfers (Minneapolis:
Univ. of Minnesota Press, 1939).
[II] Findlay, R., and Grubert, H., 'Factor Intensities, Technological Progress, and
the Terms of Trade', Oxford Economic Papers, XI (Feb 1959) 111-21.
[12] Harrod, R. F., Towards a Dynamic Economics (London: Macmillan, 1948).
[13] Hicks, J. R., 'An Inaugural Lecture: The Long-Run Dollar Problem', Oxford
Economic Papers, v (June 1953) 117-35.
(14) Hume, D., Writings on Economics, ed. E. Rotwein (Madison: Univ. of
Wisconsin Press, 1955).
(15] Iverson, C., Some Aspects of the Theory of International Capital Movements
(Copenhagen: Levin & Munksgaard, 1936).
[16) Johnson, H. G., 'Effects of Changes in Comparative Costs as Influenced by
Technical Change', in R. Harrod and D. Hague (eds), International Trade in a
Developing World (London: Macmillan, 1963) chap. 4.
(17) Keynes, J. M., 'The German Transfer Problem', in Readings in the Theory of
International Trade (Philadelphia: Blakiston, 1950).
[18] Mill, J. S., Principles of Political Economy, ed. Sir W. J. Ashley (London:
Longmans, Green, 1909) bk 111.
[19] Ohlin, B., Interregional and International Trade (Cambridge, Mass.: Harvard
U.P., 1933).
[20] Prebisch, R., 'Commercial Policy in the Underdeveloped Countries', American
Economic Review, XLIX (May 1959) 251-73.
[21] Taussig, F. W., International Trade (New York: Macmillan, 1927).
[22] Thornton, H., An Enquiry into the Nature and Effects of the Paper Credit of
Great Britain (London: Allen & Unwin, 1939).
[23] Wagner, A., 'Agrarian State versus Manufacturing State', in F. W. Taussig
(ed.), Selected Readings in International Trade and Tariff Problems (Boston:
Ginn, 1921).
7 Intermediate Products:
The Inter-Industry Flows

Until now we have assumed that goods are produced with the help
of primary factors only, but since in practice much of the production
activity would come to a halt if there were no intermediate
products - goods which are produced to be used as inputs in other
goods - this assumption, although a convenient one, is very
unrealistic. There is hardly any justification for this assumption,
even though the bulk of trade theory has ignored the presence of
material inputs which constitute a very large proportion of the
total volume of world trade. t It is only recently that trade theorists
have come to recognise the importance of intermediate products in
the production process, but even here the general tendency has been
to defend the neglect of the treatment of material inputs in the earlier
literature. In his book on trade theory, for example, Kemp
([7], p. 148), citing Vanek [10], defends the neglect of the
incorporation of intermediate goods in earlier trade theory. In
some respect, this neglect may be justified. Many properties of the
general equilibrium model presented in the previous chapters carry
over to the model with intermediate goods. However, there are
some crucial differences which have not been recognised before.
The analysis of such similarities and differences is the subject-
matter of this chapter.
Within the discussion on intermediate products, the earlier
analyses have been disproportionately concentrated on the implica-
tions of inter-industry flows, goods that serve the dual role of
intermediate as well as final products. However, it cannot be
denied that there exist several types of goods which are used not at
all for consumption, but are produced solely to be used as inputs
in other final goods. These may be called 'pure' intermediate
products. In this chapter we are concerned only with the inter-

t See, for example, Yates [II].


Intermediate Products: The Inter-Industry Flows 155

industry flows. The latter type of intermediate goods is analysed in


the next chapter.

7.1 The Model with Inter-Industry Flows


It is assumed that there are two commodities (X1 and X2 ), each of
which requires in the production process two primary factors, L
and K, as well as the use of the other good. Other assumptions
concerning production functions, perfect competition, etc., made
at the beginning of Chapter 2, remain the same.
Let Xii denote the amount of ith good used as an input in the
production of the jth good, and let xi be the net output of the jth
good (i,j = 1, 2). The production functions for the two commodi-
ties may then be written as
(7.1)
and
(7.2)
where .\} now becomes the gross output of the jth good. The
expressions for unit outputs then become

and

where cij• as before, denotes the direct requirement of the ith input
per unit of the jth good. Full employment of the primary factors
implies that
(7.3)
and
(7.4)
It is convenient, however, to express the full-employment conditions
in terms of net outputs only. Solving for X12 and X21 from the
following two equations:
C12 = X12/(x2+X21)
156 Studies in the Pure Theory of International Trade

and

we obtain

X 12 -_ C12(x2+C21x1) an
d X _ C21(x1+C12x2)
21 - ·
1-c12 • C21 1-c12 · C21
Substituting these in the full-employment equations, (7.3) and (7.4),
we get

(7.5)

and

[ cK1 + cK2 • C21] x1 + [cK2 +CKl · c12] x2-


_ K. (7.6)
l-C12. c21 I-c12. c21
The coefficients of x 1 and x 2 in the last two equations reflect the
total (or true) requirements of primary factors per unit of net
outputs. For example, (CLl + CL2 . c21)/(l- c12. c21) represents
the amount of labour which directly enters the production of a
unit of X1 plus the amount of labour embodied in that quantity of
X 2 which X1 uses as an input. To complete the description of the
model, we need add only two equations for product prices, which
under perfect competition reflect unit costs. Thus
(7.7)
and
(7.8)

Substituting for p 2 in (7. 7) and for p 1 in (7.8), we obtain

( CLl +CL2 · C21) w+ (CKI +CK2 • C21) r = P1 (7.9)


I-C12. C21 I-C12 . C21
and

( cL2+CL1 · C12) w+ (CK2+CK1 · C12) r = p 2 . (7.10)


1-c12 · C21 1-c12 · C21
The basic difference between the full-employment and price
equations presented here and the corresponding equations presented
Intermediate Products: The Inter-Industry Flows 157

in Chapter 2 where inter-industry flows were ignored must now be


apparent. If each of C12 and C21 equals zero, we obtain the simple
two-good model without intermediate goods. In the presence of
material inputs, however, the model not only becomes more
complicated, but it is also no longer certain whether the system has
a solution. For the production of any positive net amount of the goods
is possible if and only if (I- c12 • c21) > 0, a requirement which is
similar to the so called Simon-Hawkins condition well familiar to
students of linear programming; and there is no guarantee that this
condition will be satisfied with arbitrarily chosen production
functions. Thus, in this respect at least, the existence of inter-
industry flows modifies the model without intermediate products.
In what follows, we assume that C 12 • C21 < I.
Let Rij denote the total requirement of the ith primary factor per
unit of the jth good. Then the full-employment and price equations
can be written as
RLlxl +RL2x2 =L (7 .II)
RK1X1 +RK2X2 =K (7.12)
RLlw+RKlr = P1 (7.13)
and
RL2w+RK2r = P2· (7.14)

Let R denote the matrix of production coefficients contained in


(7.11) and (7.12). Then

The determinant of [ R] is given by

IRI = RL1RK2-RK1RL2 = RL1RL2[(RK2/RL2)-(RKdRL1)]

whose sign, it may be surprising to note, is the same as the sign of


ICI obtained in Chapter 2, for
RK2 RKl CK2+CK1c12 cK1+CK2C21
---
RL2 RLl cL 2 +CL1C12 eLl +CL2C21
o- c12c21)(cL1 cK2- cKl cL2)
(CL2+CL1C12)(CL1 +CL2C21).
158 Studies in the Pure Theory of International Trade

Since cl2c21 < 1 by assumption, and since ICI = eLl CK2- CK1CL2•
the sign of IR I is the same as the sign of Iq. t The sign of Iq reflects
the factor-intensity ranking of the two commodities in terms of
'apparent' or net production coefficients, whereas the sign of IRI
expresses the factor-intensity ranking in terms of 'true' or gross
production coefficients. The conclusion is unmistakable. In the
inter-industry flow model, the gross and the net factor-intensity
rankings of commodities are identical. In other words, if a
commodity appears to be, say, capital-intensive relative to the
other, then it will remain capital-intensive even when account is
taken of the amounts of the primary factors embodied in the
production of those quantities of a goo.d which are used in the pro-
duction of the other. This identity suggests that at least those
theorems, usually derived in the absence of intermediate products,
that depend exclusively on the inter-industry factor-intensity
ranking will hold without any qualitative modification even in the
presence of inter-industry flows. As will be shown later, this indeed
turns out to be the case. But first we attend to the properties
associated with the transformation curve in the present model.

7.2 The Transformation Curve


The important questions associated with any discussion of the
transformation curve concern its slope as well as its shape. It can be
readily shown that, just like the model without intermediate
products, the marginal rate of transformation between net outputs
equals the commodity-price ratio. Similarly, the transformation
curve concerning net outputs is concave to the origin, although
that concerning gross outputs may have any shape. The discussion
in this section is devoted to an examination of these two questions.
Since the total value of net output equals the sum of factor
payments,
plxl +p2x2 = wL+rK.
Differentiating this totally and keeping in mind that L and K are in
inelastic supply,
(7 .15)
Taking the factor prices as given in competitive factor markets, the

t See the expression for ICl in Chapter 2.


Intermediate Products: The Inter-Industry Flows 159

entrepreneur minimises its unit cost by setting the first derivative


of (wRLi+rRKi)U = 1, 2) to zero; that is,
(7.16)
From (7.11) and (7.12),
Ldw+Kdr = (RLix 1 +RL 2 x 2 )dw+(RK 1 x 1 +RK 2 x 2 )dr
= x 1 (RL 1 dw+ RK 1 dr)+x 2 (RL2dw+ RK 2 dr).
Using (7.13), (7.14) and (7.16),
Ldw+Kdr = x 1 dp 1 +x 2 dp 2 •
Substituting this in (7.15) yields
p 1 dx 1 +p 2 dx 2 = 0
or

(7.17)

which accords with the slope of the transformation curve obtained


in the model without intermediate products.
Coming now to the shape of the transformation curve, we
begin with the observation that in models where (7.17) holds, the
shape of the transformation curve can be directly inferred from the
response of the outputs to changes in their relative prices.
Specifically, if the supply curves are upward-sloping, the underlying
transformation curve is concave to the origin; if they are downward-
sloping, the transformation curve will be locally or globally convex
to the origin. Therefore, what we need now are expressions for
output response to changes in prices. For this purpose, the full-
employment and price equations should be transformed into
equations of relative change. Totally differentiating (7 .11 )-(7 .14),
we get
(7.18)

(7.19)

(7.20)
and
(7.21)
160 Studies in the Pure Theory of International Trade

where A.ii denotes the proportion of the total endowment of the ith
primary factor used directly and indirectly in the jth sector (for
example, A.Ll = RL 1x 1/L), and ()ii stands for the total distributive
share of the ith primary factor in the jth sector (for example,
()Ll = RL1wfp 1 ). As usual, in obtaining (7.20) and (7.21) use has
been made of the cost-minimising condition (7.16). There is yet
another way of writing (7.20) and (7.21), one which for some
problems turns out to be more useful. Differentiating (7. 7) and
(7.8), we get
(7.20*)
and
(7.21*)

where Pii is the net relative share of the ith factor (i = L, K, 1, 2) used
in the production ofthejth product; for example, PL1 = CL1wfp 1 •
Evidently, then, PL1 +pK 1 +p 21 = 1, and so on. Solving (7.20) and
(7.21) simultaneously, we obtain
w• = (OK2PT-OKlP!>!IOI (7.22)

r• = (- OL 2 PT + OLlp!>fiOI (7.23)
and
(w*-r*) = -(p! -pVJIOI (7.24)

where

It can be readily seen that the sign of IOj is the same as the sign of
jRj, so that if k 2 ~ k 1 , jOj ~ 0.
Each direct input-output coefficient CiJ• and thus each total
input-output coefficient Rii, is a function of the prices of productive
inputs. Therefore, we may write

Ci1 = Ci1(w,r,pJ (i = L,K,I,2;j = 1,2; i of= J).

However, since commodity prices from (7.24) are related to factor


prices, the total input-output coefficients are uniquely determined
solely by w and r.
Intermediate Products: The Inter-Industry Flows 161

It is shown in the appendix (section 7.6) that

Rl. 1 = - (a.+pz 1fJ)(w*-r*)


QL

Rlz = - (/3 + p 12 a.) (w*- r*)


YIL
• -_ (a.+pzl/3)( w• -r*)
RKl
QK
• -_ (/3 + P121X) ( w• - r *)
R K2
YIK
where
IX = PL1P210f:za£2 + PK1P210f.zalz + PL1PK1a£K
f3 = PL2P1zOf:1ai1 +PK2P120f.1ai1 +PLzPKzaiK
nL = PL1 + PLZP21 > o, nK = PKl + PK2P21 > o,
YIL = PLz + PL1P12 > 0, YIK = PKz + PK1P12 > 0,

and where a{k denotes the partial elasticity of substitution between


factors i and k in the jth sector. t
We normally expect a rise in the wage/rental ratio to lower the
labour coefficients and raise the capital coefficients. A sufficient con-
dition for these results to hold is that both a. and f3 be unambiguously
positive. Where only two primary factors are used in the production
of the two goods, or where intermediate products are used in fixed
proportions, so that alz = ai 2 = af_ 1 = aL = 0, the elasticity of
substitution between labour and capital in each sector is necessarily
positive, i.e. a{K > 0. It then follows that (a., /3) > 0. With three
factors of production, however, the sign of the various elasticities
of substitution is indeterminate, although it can be established
that at most one partial elasticity may be negative.:j: If all factors
are gross substitutes, all elasticities of substitution are positive, in
which case a simple glance reveals that both a. and f3 are positive.
However, if any two factors are complementary to each other,
some elasticities of substitution are negative; the signs of a. and f3
appear to be ambiguous in this case. However, it turns out that if
production functions are linearly homogeneous, something we have

t For further remarks on u{t> see the appendix.


t See the appendix for further details.
162 Studies in the Pure Theory of International Trade

assumed all along, there are certain restrictions on u{k which are
sufficient to ensure the positive signs of tx and {3. This complicated
and lengthy task is taken up in the appendix. Returning to (7 .18)
and (7 .19), let

and

Then YL and I'K are necessarily positive because tx and p are


positive. Substituting these expressions in (7 .18) and (7 .19), and
setting L * and K* to zero, we obtain
(7.25)
and
(7.26)

We have now gathered all the ingredients necessary to investigate


the response of net outputs to changes in the commodity-price
ratio as well as the shape of the locus of competitive outputs in the
inter-industry flows model. Solving (7.25) and (7.26) simultaneously
for xT and x~, we get

* (A.K2/'L +AL21'K)( * *)
x1 = IA.I w -r (7.27)

and

(7.28)

where

The sign of IA.I clearly depends on the ranking of the two


commodities in terms of their capital/labour ratios. It is readily
shown that IA.I always possesses the same sign as 101 or IRI. Since
we are concerned with the response of outputs to changes in
commodity prices, we may substitute for (w*- r*) from (7.24) in
(7 .27) and (7 .28) to obtain

* _ (AK2YL + AL21'K) ( * _ *) (7.29)


x1 - I.A.I·IOI Pl P2
Intermediate Products: The Inter-Industry Flows 163

and
* _ -(A.KlYL +A.LlYK)( * _ *)
Xz - IA.I·IOI Pl Pz . (7.30)

SinceyLand YKarepositiveand since IA.I·IOI > 0, xT/(pf -p!) > 0


and x1/(pf- p!) < 0. In other words, a rise irt the relative price of
the first commodity leads to a rise in its net output and a decline in
the net output of the second commodity, and conversely. The result is
similar to the one derived in the model without inter-industry flows.
As suggested earlier, if the tangency condition expressed in (7.17)
holds, the shape of the transformation curve can be inferred from
the direction of the response of outputs to changes in the commodity-
price ratio. This can be seen by differentiating (7 .17) with respect
to x 2 to obtain
d 2 X1 d(pz/Pl)
dx~ = dx 2
Now (7.30) implies that d(p 2 /p 1 )/dx 2 > 0, so that d 2 xtfdx~ < 0.
In other words, the transformation curve between net outputs is
strictly concave to the origin. t

The Geometrical Exposition


One special case where the material inputs are utilised in fixed
proportions is of further interest, for it permits the geometrical
derivation of the net transformation curve from the gross trans-
formation curve. This derivation was first accomplished by Vanek
[10], while Guisinger [6] and Batra and Pattanaik [2] later on
provided certain qualifications to Vanek's technique.
Consider Fig. 7.l, where TT' is the transformation between
gross outputs of X1 and X 2 drawn under the assumption that
C12 = C21 = 0; TT' is thus derived in the absence of inter-
industry flows. If any of C12 and C21 is not zero, the new trans-
formation curve will lie within the area enclosed by TT'. Consider
any point such as Eon TT'; draw EX and EY parallel respectively
to the X2 axis and the X1 axis; next, draw two straight lines from
E, the slopes of which with reference to the new axes (EX and EY)
t It may be noted here that it is the net output of each good which is positively
related to its price. The gross output of the commodity may behave in any way.
For this reason, there is nothing to guarantee that the gross transformation curve
would also be concave to the origin. For further details, see Ethier [5].
164 Studies in the Pure Theory of International Trade

G'
Figure 7.1

equal C12 and C21 , that is, the slope of EC12 intersecting 0 X1 at
A, and that of EC21 intersecting OX2 at B, represent, respectively,
the requirement of X1 per unit of X 2 and of X 2 per unit of X1 ; OA
is then the net output of X1 and OB the net output of X 2 , and these
two furnish a point E' that corresponds to E on the gross
transformation curve TT'. In other words, AX of X1 and BY of
X2 serve the role of material inputs in producing net outputs equal
to OA and OB. By repeating such construction for other points on
TT', we derive all the points on the net transformation tt' except
its corner points, t and t'. For example, t' is obtained by drawing
from the origin a ray OC', parallel to EC21 and intersecting TT'
at C'; drawing C't' parallel to EC 12 then furnishes one of the
Intermediate Products: The Inter-Industry Flows 165

corner points of the new transformation curve. In other words,


et' amount of X1 is utilised by eC' amount of X 2 , which in turn is
fully utilised by X 1 , so that the net output of X 2 is zero and that of
X1 is Ot'. It may be noted that t' does not correspond to T' but to
C' and is derived from the latter in a manner different from the
derivation of E' from E, in order to ensure that the net supply of X2
is zero. Similarly, we can derive the other corner point of tt' by
drawing first OC parallel to EC12 and then Ct parallel to EC21 .
Thus the net transformation curve tt' corresponds only to the CC'
segment of the gross transformation curve TT'. As rigorously
established above, tt' must be strictly concave to the origin.
An interesting possibility emerges if the economy is exposed to
international trade, for then a part or all of the imported good can
be used as an input in the production of the exportable good. When
this possibility is admitted, the net transformation curve extends
its frontier to the negative quadrants and is actually given by GG';
here the corner points G and G' are aligned respectively to T and
T'. For example, point G implies a maximum net output of X 2
equal to OT which is produced with the help of the inelastically
supplied primary factors, plus GT amount of X1 which is obtained
from abroad. A similar explication applies toG'. Thus if material
inputs can be imported, the net transformation curve extends to
the negative quadrants. Now if material inputs are used in fixed
proportions, the net transformation curve has corner points such as
G and G '. But if all factors are used in variable proportions, the
net transformation curve has virtually no limits, for if any amount
of the material input can be imported from abroad, the output of
the exportable good can be increased without limit by continued
addition of the imported input to the fixed domestic primary
factors. The limits on the output of the exportable good are then
provided only by the availability of the world export surplus.

7.3 Some Standard Trade Theorems


Let us now see whether some of the standard trade theorems
continue to be valid in the presence of intermediate products.
To begin with, let us re-examine the Stolper-Samuelson theorem
which is concerned with the impact of commodity prices on real
factor rewards. From (7.22) and (7.23),

• p* (}Kj(pf-pt) (7.31)
w - j = IOI
166 Studies in the Pure Theory of International Trade

and
• • -(hj(p!-p!)
r -pi = IOI (7.32)

If the first commodity is capital-intensive, 101 < 0. Clearly, then,


a rise in its relative price (p! > p!) causes a rise in the reward of
its intensive factor, capital, and a decline in the reward of the
other factor, labour, in terms of both commodity prices, showing
that the Stolper-Samuelson theorem remains unscathed by the
introduction of the intermediate products.
Next we turn to the Rybczynski theorem, which deals with the
effects of changes in factor supplies on the outputs at constant
commodity prices. The latter stipulation implies that factor
rewards and hence the input-output coefficients are constant.
In other words, the expressions in the parentheses of the right-hand
side of (7 .18) and (7 .19) reduce to zero. Solving them simultaneously
then yields

(7.33)

and
*_ A.L1K*-A.K1L*
(7.34)
x2 - IA.I
With xl capital-intensive relative to x2, IA.I < 0. A rise in the
supply of, say, capital alone then raises the net output of the
capital-intensive commodity, x 1 , and lowers the net output of the
other commodity, x 2 , to show that the Rybczynski theorem con-
tinues to hold in the presence of material inputs. However, the
theorem holds only qualitatively, because the magnitude of the
changes in the net outputs is different in the presence of inter-
mediate goods. In the case where only the labour force is growing,
for example, we may write from (7.33) and (7.34) that
dx 1 cK2+CK1c12
(7.35)
dL CL1CK2- CL2CK1
and
dx 2 -(CKl + cK2c21)
(7.36)
dL eLl cK2- cL2cK1
Given that the denominator of (7.35) and (7.36) is negative, the
Intermediate Products: The Inter-Industry Flows 167

decline in the net output of X 1 and the rise in the net output of X 2
resulting from labour growth alone will be greater in the presence
of intermediate goods (so that cl2 and c21 are positive) than
would otherwise be the case. In other words, the economic
expansion resulting from labour growth becomes more ultra-
biased in the presence of inter-industry flows. This is because,
with the increase in the output of one commodity, additional units
of output of the other commodity must be withdrawn from final
consumption, a factor that vanishes in the absence of intermediate
products. It follows then that the magnitude of the changes in the
terms of trade consequent upon the increase in labour supply will
also be more pronounced in the presence of inter-industry flows.
Thus we see that the Stolper-Samuelson and the Rybczynski
theorems are unaffected qualitatively by the incorporation of inter-
industry flows. It is now a simple matter to show that the magnifi-
cation effect and the fundamental duality that were found to exist
between these two theorems also remain unaltered. A simple proof
consists in the observation that equations (7.31)-(7.34) are exactly
the same as the corresponding equations (2.35)-(2.38) in Chapter 2,
except that j),l and 101 in each chapter are different in magnitude
but not in sign.

7.4 Technical Progress


We have shown in the previous chapter that neutral technical
progress in any commodity is ultra-biased, because at constant
commodity prices it promotes the expansion of the 'progressive'
commodity at the expense of the output of the other commodity.
We now wish to examine whether this theorem holds in the
presence of inter-industry flows without any additional qualifica-
tion. The extension of this theorem to the case of inter-industry
flows has recently been provided by Casas [3], and our analysis
here runs parallel to his discussion. In the presence of technical
change
(7.37)
where t denotes the technical improvement. To be more specific,
we assume that only the first industry experiences a Hicks-neutral
improvement at a rate T*, where

T* = - -1- aci 1 dt > 0 (i = L, K, 2). (7.38)


acil at
168 Studies in the Pure Theory of International Trade

Assuming that the factor supplies are unchanged, so that


L* = K* = 0, equations (7.13) and (7.14) then change to
A.L1x! + A.L2x! = - (A.L1RL1 + A.L2Ri2)
Au (1 + C12 x 2 /x 1 )T*
+ (7.39)
(1- c12C21)
Ax1x!+A.x2x! = -(A.K1R:1 +A.x2R:2)
+ Ax1(1 +C12x2/x1)T* (7.40)
l-C12C21
where Rt again denotes the change in Rii resulting from a change
in the wage/rental ratio only. Similarly, (7.15) and (7.16) are now
replaced by

(7.41)

and

*
* () * --P2+(1-C12C21)
() L2w+x2r p 12 T* (7 42)
.

where Pii and ()ii are defined as before. If commodity prices are
kept constant, (p!- p!) = 0 so that, by subtracting (7 .42) from
(7.41), we obtain

(7.43)

As before, the first expression on the right-hand side of (7.39) and


(7.40) can be expressed in terms of YL and Yx, although, since
(w*- r*) is now different, their values will be different. Substituting
(7.43) in the expressions for YL and Yx, the resultant YL and Yx in
(7.39) and (7.40), and then solving the latter two equations
simultaneously, we obtain
T*
x!=----
(l-C12C21)
x [(l + C / ) + (1- Pd(A.x2YL + A.L2Yx)J (7.44)
12x2 xl IA.IIel

*_ -T* [(1-pd(A.xlYL+A.uYx)J (7.45)


x2 - (l-C12c21) IA.IIOI 0
Intermediate Products: The Inter-Industry Flows 169

Our conclusions in this section, just like the ones derived in the
section on the transformation curve, depend chiefly on the signs of
YL and YK· Recalling that both YL and YK are positive, and given our
assumption that C12 C21 < l, it can be seen that xt > 0 and
x~ < 0. In other words, the output of the commodity enjoying
the technical improvement rises and that of the other commodity
declines when terms of trade are kept constant. Thus neutral
technical progress in any commodity is ultra-biased even when
inter-industry flows are present.

7.5 The Gains From Trade


The existence of inter-industry flows has very interesting and
important implications for gains from trade. It has been argued
by Chipman ([4] pp. 509-ll) and McKinnon ([8] pp. 601-10),
among others, that the existence of trade in intermediate products
tends to raise a country's gains from trade and hence its social
welfare. A simple demonstration of this result involves the use of
Fig. 7.2, where HH' is the home net transformation curve in the
absence of trade in intermediate products. That is to say, if the
imported good is used only for final consumption and not as an
intermediate product, the maximum possible net output of the
exportable good, x 1 , equals OH'. However, if the importable good
is also used as a material input, the transformation curve, as argued
before, extends to the negative quadrants. Let us first consider the
case where the importable good is used for final consumption only.
For any country, large or small, the maximum level of social
welfare under free trade is the one that is associated with complete
specialisation in the exportable commodity. Let us assume that
world prices are such that the home country does in fact produce
only its exportable commodity in the free trade equilibrium. Such
a world-price ratio is given by FP, production is at H', con-
sumption at C, and social welfare is given by U1 • Now let us
suppose that the imported good is used not only for final
consumption but also for the production of the exportable good.
If FP is such that it is just sufficient to take the production point
to H', that is to say, if FP just equals the marginal rate of
transformation in the neighbourhood of H', there will be no change
in the output of x 1 even though the transformation frontier now
extends to the negative quadrants. At the same time there will be
no change in welfare. However, if FP is greater than the marginal
170 Studies in the Pure Theory of International Trade

FP 1

Figure 7.2

rate of transformation in the neighbourhood of H', the production


point will change when the imported good can be used in the
production of the exportable good and welfare will improve.
Such a possibility is depicted in Fig. 7.2, where the foreign-price
line FP', parallel to FP, takes the production point to P, the net
output of X1 equals OA, consumption is at C', and welfare is at
U 2 , which is higher than the welfare level attained when X 2 cannot
be used as an intermediate product. The total amount of X2
imported equals C' B which is used for final consumption, plus AP
which is used as an input in the production of X1 . This is how the
Intermediate Products: The Inter-Industry Flow~ 171

existence of trade in intermediate products leads to an increase in


gains from trade and hence to an increase in social welfare.
However, this is not the end of the story. Melvin [9] has raised
an interesting point which, in the present context, serves as a note
of warning against hasty generalisations. Introducing inter-industry
flows in a Ricardian, single-primar-y-factor framework, he shows
that even though world welfare as a whole goes up when inter-
mediate products are traded, it is no longer certain that each
individual country will gain from trade. In order to comprehend
the subtleties of Melvin's argument, it is necessary first to derive
the transformation curve when inter-industry flows are introduced
in the Ricardian model. The derivation is much simpler than that
in the case of the two-primary-factor world, and requires the use
of total product curves for each good. Consider Fig. 7.3, where
0/2 and OJ;. are, respectively, the total product curves for X 2 and
X1 , and where the production of each good requires the use of

,,
Figure 7.3
172 Studies in the Pure Theory of International Trade

labour and the other good. Each of Of2 and 0/1 are drawn under
the constraint of full employment of labour, which in Melvin's
words implies that 'all possible production points must be convex
combinations of points on/1 and};' ([9] p. 142). In Fig. 7.3, inputs
are measured along the negative axes and outputs along the
positive axes; M 1 and M 2 are the points where/1 and/2 respectively
approach their maxima. 1,-et AB be the common tangent drawn to
/ 1 and};. Then M 2 AHH'BM 1 is the net transformation curve. In
autarky, or in the case where the imported good is used only for
consumption, the transformation curve is given by the segment
HH'. If the home country is a small country and faces a given
world-price ratio different from that given by the slope of HH',
then, of course, the home country will benefit by the introduction
of trade. It also follows, from the argument presented above in the
two-primary-factor case, that the gains from trade will be higher
if the world-price ratio is such as to take the production point to
the negative quadrant. However, if the equilibrium world-price
ratio is obtained through the interaction of the offer curves of the
trading countries, it is not necessary that the free trade price ratio
will differ from the autarky price ratios of both trading partners.
This follows directly from an examination of Fig. 7.4, where OKH

H'

K'

Figure 7.4
Intermediate Products: The Inter-Industry Flows 173

and OGF are the home and foreign offer curves of the Ricardian
model if the imported goods are not used as inputs. Thus K and G,
as in Chapter 3, indicate the points of complete specialisation in
the home and the foreign countries respectively. Point K, for
example, conforms to point H' in Fig. 7.3. However, if the
imported goods are not only used for consumption but also for
production, the points of production are no longer limited to either
H or H'; as suggested, they extend to the negative quadrants. This
means that the kinks in the offer curves in Fig. 7.4 also occur at
higher volumes of trade. Now if the foreign offer curve extends
only to OG 'F' and the home offer curve undergoes a large shift to
OK'H', the free trade equilibrium will then occur at A instead of
at £, which is the point of intersection between the two offer curves
in the absence of trade in intermediate products. The reader may
notice a clear difference in the two situations. If intermediate goods
are not traded, the free trade price ratio (0£) differs from the
home autarky price ratio (0 K), whereas in the presence of trade in
intermediate products the free trade price ratio and the home
autarky price ratio do not differ, as is the case in Fig. 7.4. In the
former situation the home country definitely benefits from trade,
as is described in Fig. 7.3, where FP is the world-price ratio
corresponding to OE, production moves from the self-sufficiency
equilibrium point S to H', consumption to C', and social welfare
improves from U0 to U 1 • In the latter situation, where the autarky
price ratio reflected by the slope of AB remains unchanged,
production moves from S to any point on H' B to the south-east
of H' (but it cannot go beyond B because the price ratio is
unchanged), consumption remains at S, again because of the
absence of any change in autarky prices, and welfare is unchanged
at U0 , which lies clearly below U1 , the welfare level that could be
attained if the imported good was not an intermediate good. In
Fig. 7.3 the actual production point in free trade is given by B, the
net output of X 1 equals OP, the import of X 2 equals GS, part of
which, SC, is used for consumption and the other part, CG, in the
production of X 1 • This is how Melvin shows that the introduction
of trade in intermediate goods may not be gainful to both countries.
The converse of this result is also true. In Melvin's words, 'a situa-
tion which results in a gain for only one country when there is trade
in consumption goods only, may very well result in gains for both
countries when trade in intermediate goods is allowed' ([9] p. 151).
174 Studies in the Pure Theory of International Trade

H'

Figure 7.5

This argument is illustrated in Fig. 7 .5. In the initial situation of


trade in final consumption goods only, the home offer and the
foreign offer curves are such that the home autarky price ratio
remains unchanged at OA, so that the home country derives no
benefit from trade. If the imported goods can also be used for
production, the two offer curves shift to OK'H' and OG'F' in such
a way that the free trade equilibrium price ratio is given by the slope
of OE, which is different from OA. Obviously, here the home
welfare will benefit from trade.

7.6 Appendix
Since each production coefficient is a function of input prices, we
can write Cij = Cij(w,r,pJ(i = L,K,I,2; j = 1,2; i =I=}). Differen-
tiating this totally, and remembering that p 1 does not affect the
input choice in the first commodity, we have
C!1 = PLluhw* + PK1u£Kr* + Pz1ubp~
c:l = PLl(JiKw*+PKl(Jhr*+PzlUkzp~ (A7.1)
Ci1 = pLlubw*+PKlu}czr*+pzlU~zP~·
Intermediate Products: The Inter-Industry Flows 175

Similarly,
C!2 = PL2aiLw* + PK2uiKr• + P12uf.,p!
c;2 = PL2uiKw• + PK2uicKr• + p12ufop! (A7.2)
Ci2 = PL2ui, w• + PK2afor* + P12uLp!

where u{k denotes the partial elasticity of substitution between the


ith and the kth factor in the jth industry. For instance, ulK is the
partial elasticity of substitution in the first industry between the
pair of factors labour and capital against the intermediate product.
If there are only two factors of production, each elasticity of factor
substitution (as defined, for example, in Chapter 1) must be
positive. This requirement follows from the assumption of linearly
homogeneous and quasi-concave production functions.t However,
when there are more than two productive inputs, this linear
homogeneity is compatible with the negative sign of some
elasticities of substitution. In the presence of inter-industry flows,
each industry utilises three factors and possesses six elasticities of
substitution out of which three may be called 'own' elasticities and
three 'cross' elasticities. For example, in the first industry, alL,
ukK, u~ 2 are the 'own' elasticities which are all negative, whereas
ulK, ub and uk 2 are the 'cross' elasticities. The negative sign of
the own elasticities arises from the fact that the demand for a factor
varies inversely with the change in its price when other factor prices
are kept constant. Only one of the three cross-elasticities may be
negative if the production functions are quasi-concave.t If the two
inputs are gross substitutes, then the corresponding cross-elasticity
of substitution is positive; if they are complements, it is negative.
The linear homogeneity in our three-factor case thus requires that
a factor may be complementary with only one other input, or that
all may be substitutes.
With these remarks in mind, equation (A7.1) and (A7.2) can be
simplified by noting that
(A7.3)

(A7.4)

t This is also proved in Chapter I.


t See Allen([!) pp. 504-8) for further details.
176 Studies in the Pure Theory of International Trade

where the lJ's are, as before, the gross share of the relevant primary
factor. If all factor prices change in the same proportion, the
production coefficients remain unchanged. Thus if w• = r* =
pf = p~, C;j = 0. This implies that

(A7.5)

-pL2uf_L = (pK2uf_K+pl2uf_ 1) > 0 (A7.6)

-PKtUkK = (pLluiK+P21ub) > 0 (A7.7)

- PK2uiK = (PL2uf.K+ P12ui1) > 0 (A7.8)

-p21U~2 = (pL1ul2+PK1ub) > 0 (A7.9)

- P12ui 1 = (PL2uf.t + PK2ui1) > 0. (A7.10)

The positive sign of all the expressions in (A 7 .5)-(A7 .10) follows


from the fact that the own elasticities of substitution are all
negative. Substituting (A7.3)-(A7.10) in (A7.1) and (A7.2) yields

Cft = -(P2t(}K2ub+PKtuiK)(w*-r*)
c:1 = (P21(}L2ub+pL1uiK)(w*-r*) (A7.11)
q't = (pL1lJK2ub-PKteL2ub)(w*-r*)
and
Cl2 = -(pl2(}Ktuf.t +PK2uf.K)(w*-r*)
c:2 = (Pt2(}L1uh +PL2UEK)(w*-r*) (A7.12)
Ct2 = (PL2(}Ktuf.t-PK2(}L1u'it)(w*-r*).

It will be recalled that RL 1 was defined as

Hence
• _ dRLI _ CL1Clt +C21RL2C!1 +C21(CL2Cl2+C12RL1Ct2)
RL l- -
RLl eLl+ cL2C21
PL1Ctt +p21lJL2Ct1 +P2t(PL2Cl2+P12lJL1Ct2)
=
Intermediate Products: The Inter-Industry Flows 177

Substituting from (A7.11) and (A7.12), and noting that


(}Ll + (}K1 = (}L2 + (}K2 = 1
• _
R L1-- [<PLl P21 Oi:2ub + PK1 P21 Of.2u:C2 + PK1PL1u£K)
(PL1 + PL2P21)
+ P21 (PL2Pu0i:1 uf.1 + PK2Pu0f.1 ui:1 + PK2PL2uf.K)J (w*-r*)
(pLl +PL2P21)
- (~+;: 1 p)(w*-r*).

In a similar fashion, it can be shown that

Rt2 = - (P+P12~)(w*-r*)
'1L
• _ <~+P21P>( •
R K1- *)
QK W -r

• _ <P+P12~)( w • -r *)
R K2-
'1K
where
~ = PL1P210i:2ub+PK1P21(}f_2ub+PL1PK1u£K
P = PL2PuOi.1uf.1 + PK2PuOl1ui.1 + PL2PK2u1K
QL = PL1 +PL2. P21• QK = PK1 +PK2. P21
'1L = PL2 + PL1 • P12• '1K = PK2 + PK1 • Pu·

It is clear that the signs of the Rt"s depend crucially on the


signs of ~ and P; we shall now proceed to show that, given our
assumptions with respect to the production functions, both ~ and
p are positive.
(A7.5) can be rewritten as:

0'LL -PK1)
1 -- (- -P21)
- 0'1LK + (- - 0'1L2·
PL1 PL1

Substituting for ulK from (A7.7) and for ub from (A7.9), we get

1 -
O'LL- 1 ( PK10'KK
-2-
2 1 +2 PK1P210'K2
1 + P210'22.
2 1)
PL1
178 Studies in the Pure Theory of International Trade

This, in turn, may be rewritten in matrix notation as

1
(TLL =
1 [
-2 (A7.13)
PKl
PL1
We have previously argued that under the assumption that pro-
duction functions exhibit constant returns to scale and are quasi-
concave, the 'own' partial elasticities of substitution are negative,
which implies that the quadratic form on the R.H.S. of (A7.13)
must be negative definite. This constraint is satisfied if (i) uh < 0,
(ii) u1 2 < 0, and (iii) ul::Ku1 2 - (ul:: 2 ) 2 > 0. The first two conditions
have already been assumed to be valid, while the third may be
written as

which is possible only if

Remembering that
a:= P21PL1ei_2ub+P21PK10l2ul::2+PL1PK1uh
we can now write that

or

or

or
(A7.14)
Intermediate Products: The Inter-Industry Flows 179

Since the right-hand side of (A 7.14) is positive, ex > 0. In a


similar fashion, we can establish the positive sign of fJ.

REFERENCES
[I) Allen, R. G. D., Mathematical Analysis for Economists (London: Macmillan,
1938).
[2) Batra, R.N., and Pattanaik, P. K., 'Economic Growth, Intermediate Products,
and the Terms of Trade', Canadian Journal of Economics, IV (May 1971) 225-37.
[3) Casas, F. R., 'The Theory of Intermediate Products, Technical Change and
Growth', Journal of International Economics, II (May 1972).
[4) Chipman, J. S. 'A Survey of the Theory of International Trade: Part 3, The
Modern Theory', Econometrica, XXXIV (Jan 1966) 18-76.
[5) Ethier, W., 'Input Substitution and the Concept of the Effective Rate of
Protection', Journal of Political Economy, LXXX (Jan-Feb 1972) 34-47.
[ 6) Guisinger, S. E., 'Negative Value Added and the Theory of Effective Protection',
Quarterly Journal of Economics, LXXXIII (Aug 1969) 415-33.
[7) Kemp, M. C., The Pure Theory of International Trade and Investment
(Englewood Cliffs, N.J.: Prentice-Hall 1969) chap. 7.
[8) McKinnon, R. 1., 'Intermediate Products and Differential Tariffs: A Generaliza-
tion of Lerner's Symmetry Theorem', Quarterly Journal of Economics, LXXX
(Nov 1966) 584-615.
[9) Melvin, J. R., 'Intermediate Goods, the Production Possibility Curve, and the
Gains from Trade', Quarterly Journal of Economics, LXXXIII (Feb 1969) 141-51.
[10) Vanek, J., 'Variable Factor Proportions and Inter-Industry Flows in the
Theory of International Trade', Quarterly Journal of Economics, LXXVII
(Feb 1963) 129-42.
[II) Yates, P. L., Forty Years of Foreign Trade (London: Allen & Unwin, 1959).

SUPPLEMENTARY READINGS
[12) Warne, R. D., 'Intermediate Goods in International Trade with Variable
Proportions and Two Primary Inputs', Quarterly Journal of Economics,
LXXXV (May 1971) 225-36.
8 Pure Intermediate Products

In the previous chapter our analysis focused entirely on the implica-


tions of inter-industry flows for various theorems in trade theory.
As noted there, most of the trade literature on intermediate products
concerns the commodities which serve the dual role of goods used
for final consumption and as productive inputs. However, there
exists another class of intermediate goods which are produced solely
to be used as productive inputs and not for final consumption at all.
Such goods may be called pure intermediate goods, and include
produced inputs like raw materials, spare parts, steel, etc. The
current literature in trade theory pays scant attention to the im-
plications of such products. If the ramifications of pure intermediate
products ran closely parallel to those of inter-industry flows, the
current neglect of the former could be easily vindicated. However,
as will be demonstrated in this chapter, there are important differ-
ences between the implications of the two types of material inputs,
crucial among them being the fact that, unlike with the inter-industry
flows, the gross and net factor-intensity rankings need not be iden-
tical in the presence of pure intermediate products. This implies that
those traditional theorems in trade theory that depend exclusively
on the nature of the differences between the factor intensities in the
two final products may no longer be valid in the model that allows
for the presence of pure intermediate products, even though these
theorems continued to be valid in the model with inter-industry
flows.
In some respects, of course, the two types of intermediate goods
give rise to identical consequences for certain variables in the
economy. For example, the number of productive factors rises from
two to three, so that with both classes of intermediate goods there
arises the possibility of complementarity among inputs, although,
as with inter-industry flows, this in itself introduces no qualitative
modification to the results.
Pure Intermediate Products 181

These and some other interesting points emerge vividly from the
subsequent discussion, which also highlights the need for a full-
fledged and separate analysis of pure intermediate products.

8.1 The Model with Pure Intermediate Goods


It is assumed that the economy under consideration consists of three
goods, two final goods, X1 and X2 , and one intermediate product,
M, which is produced solely to be used as an input in the production
of the final products. There are no inter-industry flows as far as the
final goods are concerned, that is, none of X1 and X2 serves the role
of an intermediate good. Relaxing this assumption adds only to
complications without altering substantially any of the results de-
rived from the simpler model. In all other respects, however, the
assumptions of this chapter are the same as those of the previous one.
Full-employment equations now become
(8.1)

(8.2)
and
(8.3)

where CMi is the requirement of the intermediate good per unit of


the jth final good (j = 1,2). These three equations can be reduced
to two equations by substituting (8.3) in (8.1) and (8.2) to obtain
(CLl +CLMCMl)Xl +(CL2+CLMCM2)X2 =L
and

where the terms in the parentheses represent total factor require-


ments per unit of the final product. Denoting Rii as the total require-
ment of the ith primary factor per unit of the jth final good (i = L,K;
j = 1,2), these equations can be written as
(8.1*)
and
(8.2*)
182 Studies in the Pure Theory of International Trade

Similarly, the price equations are given by


CL1w+CK1r+CM1PM = P1 (8.4)
CL2w+ CK2r+ CM2PM = P2 (8.5)
and
(8.6)
or by
(8.4*)
and
(8.5*)

where PM is the price of the intermediate good.


At this stage it is necessary to introduce a distinction between
gross and net capital/labour ratios in the final-good industry in our
model with pure intermediate products. The net or apparent capital/
labour ratio in the jth final good is as usual given by ki = K)Li,
whereas the gross capital/labour ratio (kj) equals the ratio between
the amounts of capital and labour utilised directly and indirectly
by the final product. For example, the gross amount of capital em-
ployed in the jth final good (Kj) equals the amount of capital
utilised directly (Ki) plus the amount of capital used by the inter-
mediate good (KM) multiplied by the proportion of the intermediate
good used in the production of the jth final good; that is,
Kj = Ki+(Mi/M)KM
so that the gross capital/output coefficient is given by
K~ K (M.jM)KM
___]_ = ___}_ + J = cKj+ cMjcKM = RKj.
xj xj xi
Similarly, the gross labour/output coefficient is given by RLi· Let
R be the matrix of gross production coefficients presented in (8.1*)
and (8.2*). Evidently, then, the sign of the determinant of R deter-
mines the gross factor-intensity ranking for each commodity, that is,
>- -
-RKl <::: - , or k'1 ~
RK2 k;, only if IR I = (RL1RK2-RK1RL2) § 0.
RLl RL2
Whether or not the factor-intensity ranking in the gross sense will
Pure Intermediate Products 183

be identical to that in the net sense is determined by the expansion of


JRJ in terms of the net input-output coefficients. Thus JRJ ~ 0 im-
plies that

or
eLl cL2(k1 -k2)+ cMl cLMcL2(kM-k2)
+ CM2CLMCL1(kt -kM) ~ 0 (8.7*)

where ki is the apparent capital/labour ratio in the jth commodity


(j = l,2,M). For the sake of illustration, suppose that the first
industry is relatively capital-intensive in the net sense, that is,
k 1 > k 2. Now it will be truly capital-intensive only if k~ > k~ or if
(8. 7*) is positive. It is clear at a glance that (k 1- k 2) may not have
the same sign as (k~ -k;), that is, the net and the gross factor-
intensity rankings may not be identical in the presence of pure inter-
mediate products. This, incidentally, is one of the basic differences
between the implications of the pure intermediate products and
those of the inter-industry flows analysed in the previous chapter.
However, we can derive the sufficient conditions that will ensure
the identity between the net and the gross factor-intensity rankings.
For example, if kM lies between k 1 and k 2, one can observe that the
sign of (k 1- k 2) is the same as the sign of (k~- k;). Thus sign
(k 1-k 2) = sign (k'1-k].) if k 1 ~ kM ~ k 2. This identity continues
to hold ifkMdoesnot lie betweenk 1 andk 2, but CM 2CL 1 = CMtCL 2,
which implies that both final products possess the same intensity of
use of the intermediate product, that is, (CM 2/CL 2) = (CMtfCLt).
The problem arises in the non-trivial case where this latter equality
does not hold. There are two other possibilities: (i) k 1 ~ k 2 ~ kM,
and (ii) k 2 ~ k 1 ~ kM. It may be observed that (k 1- k 2 ) has the
same sign as (k~ - k;) if the last two terms in (8. 7*) have the same
sign as the first term. It turns out that in case (i) this requires that
CMz/CLz > C MtfCLt, whereas in case (ii) it is sufficient if CM2/C Lz <
CMtfCLt· Take, for example, the possibility where k 1 > k 2 > kM.
184 Studies in the Pure Theory of International Trade

Here (k 1 -k 2 ) > 0, and if CM 2 ICL 2 > CMtfCLl so that CM2CL1


> CM 1CL 2 , one can see that

because (k 1 -kM) > (k 2 -kM), which in tum implies that the first
and the last two terms in (8.7*) have identical signs. From all this
discussion, the following lemma may be derived:

Lemma 1. The factor-intensity rankings of the final products in


the net and the gross sense are identical if (l) kM lies between k 1
and k 2 , or (2) the commodity whose capital/labour ratio lies
between the capital/labour ratios of the intermediate product
and the other final product is at least as intensive in the use of the
intermediate product as the other final product. Any one of these
conditions is also sufficient to generate an unambiguous sign
ofiRI.

The intensity in the use of the intermediate product is defined by


CM11CLJ = M 1/L1. If condition (2) in Lemma l is not satisfied, there
arises the possibility of a conflict between the factor-intensity rank-
ings in the net and the gross sense; that is to say, a commodity may
be relatively capital-intensive in the net sense, but relatively labour-
intensive in the gross sense, and vice versa. Stated differently, the
sign of (k 1 -k 2 ) may be opposite to the sign of (k~ -ki), in which
case the sign of IRI will be ambiguous.
To summarise, the basic difference between the model with inter-
industry flows developed in the previous chapter and the one with
pure intermediate products is that the gross and net factor-intensity
rankings of the final products are not necessarily identical if some
goods are produced solely for the sake of further production in the
final products. As stated earlier, this has significant ramifications
for those theorems which depend crucially on the inter-industry
factor-intensity rankings. Otherwise the present model can be
developed in much the same manner as the one presented in the
preceding chapter. This similarity enables us to present the equa-
tions of change without recording detailed explanations.
Pure Intermediate Products 185

8.2 The Structural Relations


Proceeding in precisely the same manner as we did in the previous
chapter, the following equations can be derived by differentiating
the system of equations presented in the foregoing section:
AL1Xi"+A.L2X:i = L*-(A.L1Rl1 +A.L2Rl2) (8.7)
A.Kl Xi"+ A.K2X:i = K*- (A.Kl Rl1 + A.K2Rl2) (8.8)
PL1w*+PK1r*+PM1P%t = P! (8.9)
PL2w*+PK2r*+PM2P%t = P! (8.10)
PLMw*+PKMr* = p%, (8.11)

where, as before, A.ij is the proportion of the gross amount of the


ith primary factor used in the production of the jth final good, and
the p's refer to the net relative share of the relevant factor. Substi-
tuting (8.11) in (8.9) and (8.10),
(8.12)
and
(8.13)

where ()ij is the gross relative share of the ith primary factor in the
jth final good. As in the previous chapter, the expressions for Rt
(i = L,K;j = 1,2) can be obtained by differentiating

cij = Cij(w,r,pM)
(i = L,K;j = 1,2)
and

It is shown in the appendix (section 8.8) that

-(a+ e) (w*-r*)
(}Ll
- (P+b) (w*-r*)
(}L2
• l_- -
RK (a+e)(.
( ) - w -r
*)
Kl

Rt2 = (P()+J) (w*-r*)


K2
186 Studies in the Pure Theory of International Trade

where
tx = PLtPKtulK + PiMPLtPMtulM+ Pf.MPKtPMtulM > 0
f3 = PL2PK2uf.K + PiMPL2PM2uf.M + PiMPK2PM2uiM > 0
e = PLMPKMPMtu'tK > 0
b = PLMPKMPM2u'tK > 0.

8.3 Some Properties of the Model


To begin with, we show that the transformation curve between the
final goods in the model with pure intermediate goods has the same
properties as the model where the intermediate goods are absent.
This result, then, is in line with the one obtained in the inter-industry
flows model developed in the previous chapter.
Since the total value of production in each sector equals the sum
of factor payments,
PtXt = wLt +rKt +pMMt
p 2X 2 = wL 2+rK2+pMM2
and

so that
p 1 X1 +P2X2 = w(L 1 +L 2)+r(K1 +K2)+pMM
= w(L 1 +L 2+LM)+r(K1 +K2+KM) = wL+rK.
Differentiating this totally,
(p 1dX1 +p 2dX2)+(X1dp 1 +X2dp 2) = Ldw+Kdr. (8.14)
However, from (8.1*) and (8.2*) and (8.12) and (8.13),
Ldw+Kdr = X 1 [RL 1 dw+RK 1 dr]+X2[RL 2dw+RK 2dr]
= X 1 dp 1 +X2dp 2.

Substituting this in (8.14), we obtain


dXt P2
dX2 Pt
which, of course, shows that the commodity-price ratio reflects
the marginal rate of transformation between the final goods.
Pure Intermediate Products 187

The shape of the transformation curve can as usual be inferred


from the price-output response. Solving (8. 7) and (8.8) simul-
taneously, utilising R1j and (8.12) and (8.13), and setting L* and K*
to zero, we derive
X* _ ().Kzh +A.L2YK) ( * *)
I - IA-1·181 Pt -P2
and
X*2

where
YL = -(A.LIR!, +A.L 2R! 2)/(w*-r*) > 0
and YK = (A.KtR1, +A.KzRlz}/(w*-r*) > 0.
It is clear that Xt /(p!- p!) > 0 and Xi(P!- p!) < 0, because both
lA. I and 181 have the same sign as IRl:
RLIX! RLzXz
A.Ll A.L2
L L
IA.I = = x,x21RI
RKIX! RK2xz LK
AK! AK2
K K
RLlw RK 1 r
eLl 8K!
Pt Pt
and 181 RL2w
=~IRI.
P1P2
RKzr
8L2 8K2
P2 P2
All this suggests that the transformation curve is strictly concave to
the origin. t

8.4 The Standard Trade Theorems


Let us now examine the Rybczynski and the Stolper-Samuelson
theorems in our model incorporating the pure intermediate product.
With R1j = 0 at unchanged commodity prices, the solution of (8. 7)
and (8.8) yields

t For further details concerning the shape and the properties of the transformation
curve in the presence of pure intermediate products, see Batra and Casas [I].
188 Studies in the Pure Theory of International Trade

• A.LlK* -A.K1L*
and X2 = jA.j

If factor intensities are defined in the gross sense, jA.j, which has the
same sign as jRj, has a definite sign; if they are specified in the net
sense, then again !A.! (or jRj) has an unambiguous sign only if the
net and the gross factor-intensity rankings are identical. It follows
then that the Rybczynski theorem is necessarily valid only if factor·
intensities are defined in the gross sense; if they are specified in the
net sense the theorem may not be valid. For example, with L* > 0
and K* = 0, Xi > 0 and X! < 0 ifjA.j > 0 or if the second industry
is capital-intensive relative to the first in the gross sense. However,
if X1 is labour-intensive relative to X 2 in the net sense, that is, if
k 1 < k 2 , but it is capital-intensive in the gross sense, so that
k~ > k;, the Rybczynski theorem does not hold, because then the
output of the apparently labour-intensive commodity will decline
as a result of a rise in the supply of labour alone when terms of trade
are kept constant.
In the similar fashion, the Stolper-Samuelson theorem may not be
valid if the gross and the net factor-intensity rankings are not
identical. This can be seen by solving (8.12) and (8.13) to obtain

and

Since again !O! (or jRj) may not have the same sign as (k 1 -k 2 ), the
Stolper-Samuelson theorem may not be valid in terms of the net or
the apparent factor-intensity rankings.

8.5 Technical Progress in the Final Good


We have seen above that the theorems by Rybczynski and Stolper
and Samuelson do not hold if there is a conflict between gross and
net factor-intensity rankings of the final products. This kind of con-
flict, however, does not interfere with the traditional results con-
cerning the implications of neutral technical progress in any final
product. A rigorous demonstration of this result has been provided
by Casas [2].
Pure Intermediate Products 189

As in the previous chapter, let

T* = -=-!. acil dt (i = L ,K,M )


cil ot
be the rate of Hicks-neutral technical progress in the first industry.
The two equations of change (8. 7) and (8.8) are now replaced by
A.LlXi+A.L2Xi = -(A.LtR!t +A.L2RL2)+A.LIT* (8.15)
and AK1Xi+A.K2Xi = -(A.KlRkl +A.K2Rk2)+A.K1T* (8.16)
where Rt now indicates the change in the input-output coefficients
resulting from the change in factor prices alone. Similarly, (8.12) is
now replaced by
(8.17)
whereas (8.13) is unaltered because no technical improvement occurs
in the second industry. At constant commodity prices, the solution
of (8.13) and (8.17) yields

w* - r * = T*w· (8.18)

Solving (8.15) and (8.16), and using (8.18) and the expressions for
Rt presented in the preceding section, we obtain

X*
I
= [l + A.K2YL1..1.1·1+..1.L2YK]
01
T* > 0

and

which shows clearly that neutral technical progress in a final-good


industry raises the output of that good and lowers the output of the
other final product when commodity prices are kept constant.

8.6 Technical Progress in the Intermediate Product


So far we have assumed that technical progress occurs in a final
product. What happens if technical progress occurs in the industry
producing the intermediate product, a possibility which does not
arise in the inter-industry flow model? This section is devoted to an
examination of the implications of this type of technical advance
for the output of the final goods, when commodity prices are
unchanged.
190 Studies in the Pure Theory of International Trade

The implications of a Hicks-neutral technical improvement in M


are more complex than a superficial glance might suggest. Such a
technical change has the effect of lowering the unit cost of produc-
tion in all three commodities, where the extent of the unit cost
reduction in the two final products depends on the rate of technical
advance in the intermediate good and the two material input-output
coefficients, CMi· This amounts to the occurrence of technical im-
provements in the jth final product in proportion to the initial level
of CMi• and one may argue that the entire question could be treated
in the same way as that presented in the previous section. However,
a technical improvement in the intermediate good also tends to raise
the supply of the material input, because the same output level of
this product can now be produced by smaller quantities of capital
and labour. The rise in the supply of the material input in turn has
its own repercussions on tht< output of the final products. The final
outcome depends on these two forces, which may be contradictory
or complementary to each other. Let
1 aciM
T~ = --C --,dt (i = L,K)
iM ut

be the rate of technical advance in M. Differentiating (8.1)-(8.3), we


have
A.L1Xt+A.L2Xt = YL(w*-r*)+cPLMT~ (8.19)
and A.K1Xt+A.KzXt = -YK(w*-r*)+cPKMT~ (8.20)
where ¢ LM and ¢ KM are, respectively, the proportion of the labour
force and the capital stock employed in M. Similarly, differentiating
(8.4)-(8.6) yields
eLl w* + eKlr* = Pt + PM1 T~ (8.21)
and eLzw*+8Kzr* = P!+PMzT~. (8.22)
From (8.21) and (8.22), withp! = p! = 0, we have
* * (PM1- PMz)T~ (8.23)
w -r = IOI
Using (8.23), (8.19) and (8.20) can be solved to obtain
Xt = (AKzYL +A.Lzi'K)(PM1 -PMz) + AKzcPLM-A.L2¢KM (8.24)
T~ IA.i·IOI IA.I
Pure Intermediate Products 191

_ (A.KlYL +ALlYK)(PMt -pMz) + A.Ll<f>KM-AKt<f>LM


IA.I·IBI IA.I
(8.25)
Let 'cost effect' denote the effect of technical progress in the inter-
mediate product on the unit costs of the two final goods, and let
'factor-supply effect' be the effect of such a technical change on the
total supply of the material input. The cost effect depends on the
relative share of the intermediate good in the total cost of each final
product, or simply on PMt and PMz· If PMt = PMz• unit costs in
both commodities decline in the same proportion. If PMt > PMz•
the cost effect is equivalent to a higher rate of technical advance in
X1 than X 2 , and vice versa. Consider the case where PM 1 > PMz•
so that the unit cost in X1 declines relative to the unit cost in X2 • If
commodity prices are to be kept constant, then the relative price of
the primary factor employed intensively by X1 must rise. That this is
the case follows clearly from (8.23) where, with IBI > 0, implying
that X 2 is capital-intensive relative to X1 in the gross sense,
(w*- r*) > 0. This in turn will lead to a rise in the gross capital/
labour ratio in both final products, and eventually to a decline in the
output of the capital-intensive industry, X 2 , and a rise in the output
of the labour-intensive industry, xl, if full employment is to be
maintained. This is the cost effect of a technical change in the inter-
mediate product on the two final outputs and is given by the first
term in (8.24) and (8.25).
The second term in these two equations arises from the factor-
supply effect. If the output of the intermediate good is kept constant,
then depending on the rate of technical advance in M, capital and
labour are released from Min the proportion kM. Therefore, if kM
lies between k 1 and k 2 , the output of both final goods will rise from
the factor-supply effect. If k 1 < kM < k 2 , then IRI and hence IA.I
are positive, which implies that (i) (A.K2/A.L 2 ) > (A.Ktf.A.L 1 ) and (ii)
(A.K 2 /AL 2 ) > (<f>KMI<I>LM) > (A.K 1/A.Ll).t Clearly, then, both the
numerator and the denominator of the second term in (8.24) and
(8.25) are positive. In the opposite case where k 1 > kM > k 2 , both
of these are negative. Hence the second term in both of the equations
is positive when kM lies between k 1 and k 2 • However, if the latter
condition is not fulfilled, IA.I may not have a definite sign and the

t This is because 1/!~eMII/!LM = kM.


192 Studies in the Pure Theory of International Trade

signs of the numerator and the denominator of the second term in


one of the two equations (8.24) and (8.25) may be opposite. For
example, suppose that IA.I > 0 so that (A.K 2 /A.L 2 ) > (A.KdA.Ll). How-
ever, (fi1KMifi1LM) may be greater than (A.K 2 /A.L 2 ) and less than
(A.KI/A.Ll) if kM does not lie between k 1 and k 2 • In this case the second
term is negative in (8.24) but positive in (8.25).
The overall effect of the technical progress in the intermediate
good on the two final outputs is the sum of the cost effect and the
factor-supply effect, which is determinate only if both effects have
the same sign. On the basis of the discussion in this section, the fol-
lowing theorem can be derived:

Theorem 8.1. If the capital/labour ratio of the intermediate


product lies between the capital/labour ratio of the final products,
then technical progress in the intermediate good raises the out-
put of the commodity which has the higher material cost
component, given that the terms of trade are constant; the effect
on the output of the other commodity is indeterminate.

If kM does not lie between k 1 and k 2 , then the output of X1 may


actually decline and that of X2 rise even if PM 1 > PM 2 , or the output
of both X1 and X2 may rise.

8. 7 Trade in Intermediate Products


In the previous chapter we showed that the introduction of trade in
intermediate products resulted in a rise in world welfare. This result,
derived from an inter-industry flow model, cannot be extended to
the present model where intermediate products are produced solely
for the sake of further production in the final products. Until now
we have tacitly assumed that the intermediate product is not traded,
so that its local supply always equals its local demand. If we intro-
duce trade in the intermediate product in addition to the trade in
final products, the model becomes indeterminate, for if all prices,
p 1 , p 2 and PM• are determined exogenously, we would have three
independent price equations (8.4), (8.5) and (8.6) to determine two
variables wand r. With the number of equations exceeding the num-
ber of unknowns there will be inconsistency in the model. On the
other hand, if the intermediate product is non-traded, we can
visualise the attainment of equilibrium in a situation where any
arbitrary level of world prices p 1 and p 2 determines w and r from
Pure Intermediate Products 193

(8.4) and (8.5), and wand r in turn determine PM in (8.6) in such a


way that the output of all three industries is positive. Another way
out of the indeterminacy would be to assume that one of the final
products is non-traded. Under no alternative, however, is it possible
to compare the level of gains from trade which a country derives in
two situations, one where the intermediate good is traded and the
other where it is not. All we can show is that, for a small country,
free trade is the optimal policy irrespective of the type of goods that
are traded.t

The Pattern of Trade


It is fashionable nowadays to explain the pattern of trade between
two countries in terms of the Heckscher-Ohlin theorem according
to which a country exports the commodity which uses intensively
its relatively abundant factor, and imports the commodity which is
intensive in the use of its relatively scarce factor. Previously, the
validity of this theorem has been shown to depend crucially upon
the validity of the Stolper-Samuelson and the Rybczynski theorems.
Unfortunately, in our model where three goods are produced, the
pattern of trade turns out to be indeterminate if all three goods are
to be traded, even if the Rybczynski and the Stolper-Samuelson
theorems hold unambiguously.
Suppose initially no trade in M is allowed. If we assume that inter-
national production functions are identical, then the Heckscher-
Ohlin theorem follows directly from the validity of the Stolper-
Samuelson theorem. If, in addition, we assume that consumption
patterns are also identical internationally, then the Heckscher-
Ohlin theorem derives from the validity of the Rybczynski theorem.
If all three goods continue to be produced in free trade equilibrium,
factor prices are equated between countries; M also commands the
same price between countries, and there is no incentive to trade M
once the prohibition to such trade is lifted.
None the less, trade could take place in either direction. World
outputs and world demand for final products are unaffected, be-
cause the free trade equilibrium gives rise to a given set of world
prices. But each country's transformation curve shifts-outwards
for the country importing M and inwards for the other. Suppose
there are two countries, the home country and the foreign country,

t This can be shown easily by following the procedure developed in Chapter 4.


194 Studies in the Pure Theory of International Trade

and suppose further that the former is relatively capital-abundant


and that k 1 > kM > k 2 • If M is exported by the home country, at
free trade prices the home production of both xl and x2 declines
(and rises abroad). Thus the home country might eventually import
both X1 and X 2 in exchange for M, although the Heckscher-Ohlin
theory would require the home country to export X1 . Alternatively,
if the home country imports M, it might (i) be an exporter of X1 and
an importer of X 2 , or (ii) export both X1 and X 2 and import M. Thus
without further restrictions it is impossible to pick out any com-
modity and state whether it will be imported or exported by the home
country.
The solution that we wish to propose must have become obvious.
by now. If one of the three goods is non-traded, not only do we
solve the indeterminacy prpblem in the model, but we can also sal-
vage the Heckscher-Ohlin theory. Note that our solution does not
require that the non-traded good be the intermediate good only.
Let us define the home country to be capital-abundant relative to
the foreign country if

where w = wjr is the wage/rental ratio and the subscripts h and f


refer, respectively, to the home and foreign countries. If the inter-
mediate good is not traded, all that we need to demonstrate in
order to prove the Heckscher-Ohlin theorem is that (dw/dp) has
an unambiguous sign. From (8.12) and (8.13),
w* -1
-p* -- 181 -
(p* - p*2 - p*)
1 . (8.26)

From (8.26) it is clear that the relationship between wand pis unique
only if jej has an unambiguous sign. If jOj < 0, then with wh > wf
it is clear that

Ph> PJ
where Pi is the autarky-relative price of the second good in terms of
the first in the jth country (j = h,f). The home country will then
export X1 and import X 2 . On the other hand, if jej > 0, then with
(J)h > (J)f
Pure Intermediate Products 195

so that the home country will export X 2 and import X1 • In general,


with the intermediate good non-traded, the Heckscher-Ohlin
theorem will hold if sign IBI =sign (k 1 -k 2 ), and we already know
that a sufficient condition for this is that kM lie between k 1 and k 2.
Let us now consider the other case where one of the final goods is
the non-traded good. Let X1 be such a good. For the demonstration
of the Heckscher-Ohlin theorem, we now need an expression re-
lating PMIP 2 with w. From (8.12) and (8.13),

(8.27)

and
*_ BLip! -8L2p!
(8.28)
r - IBI
Similarly, from (8.13) and (8.11) and (8.27) and (8.28),
* * (8L2- PLM)p*
PM-P2 = IBI (8.29)

so that by using (8.26), we obtain


P!-Pt
w* = eLl- PLM = PL2PKM- PLMPK2 (8.30)

= rpL2PLM (kM-kl)
w

where, in obtaining (8.29) and (8.30), we have made use of the fact
that PLM+PKM = ()Ki+()Li = I (j = 1,2) and that IBI = ()Kz-{}KI
= ()LI -{}L 2 • If wh > w1 and k 2 > kM, then from the Heckscher-
Ohlin theorem we should expect the home country to export X 2 and
import M. For this pattern of trade to take place, we require that

* *
pz-PM < O
w*

so that (p 2/pM)h < (p 2/PM)J when wh > w1 . Evidently, this condi-


tion is satisfied if k 2 > kM. From this we conclude that if one of the
final goods is treated as a non-traded good, the Heckscher-Ohlin
theorem continues to hold.
196 Studies in the Pure Theory of International Trade

Figure 8.1

The Nature of the Free Trade Equilibrium


Given that one commodity is not traded, what is the nature of equi-
librium in free trade? If the non-traded good is the intermediate
product, the answer is very simple and conforms to the traditional
graphical explanations, where the world prices in conjunction with
the transformation curve determine the free trade production point,
and where the consumption point is determined by the demand con-
ditions. The problem is not so simple when the non-traded good is
one of the final commodities. The free trade consumption point
relating to final goods is, as before, determined by the demand con-
ditions, but the production point is determined not only by world
prices but also by the volume of trade.
As an illustration, consider Fig. 8.1, which depicts the case of the
home country exporting X 2 and importing M while treating X1 as
the non-traded good. As previously seen, such a pattern would arise
if wh > w1 under autarky and k 2 > kM. In Fig. 8.1, HH' is the home
Pure Intermediate Products 197

country's net transformation curve and S is the point of self-


sufficiency equilibrium which is determined by the tangency of the
social indifference curve U1 with HH'. Since the home country
exports X 2 and imports M, the switch from autarky to free trade
causes a rise inp and a decline inpM/p 1 , with all prices expressed in
terms of the non-traded good. For the time being, assume that noM
is imported. Then the rise in pas a result of the introduction of trade
shifts production from StoP where GD, the price line prevailing in
the free trade equilibrium, is tangential to the net transformation
curve, and consumption moves from S to C which lies on a higher
indifference curve U2 • P, however, is not the final equilibrium pro-
duction point, for the imports of M will shift the transformation
curve outwards, the extent of this shift depending upon the volume
of M imported. Such a transformation curve is given by TT' and the
final production point is given by P*, where the price line EF,
parallel to GD, touches the new transformation curve.
Several points deserve further exposition. First, the free trade
equilibrium production point P* is horizontally aligned with the
consumption point C, showing that the supply of and the demand
for X1 are equal. Second, the output of X 2 corresponding to P* may
be smaller or larger than that corresponding to P. This follows from
the fact that the import of M is equivalent to the injection of capital
and labour into the home country in the proportion CKMICLM
= kM. However, if k 2 > k 1 > kM, the import of M will result in a
decline in the output of x2 and a rise in the output of xl at the free
trade prices if X 2 is capital-intensive in gross terms, t and vice versa.

t If we rewrite equation (8.3) as


M = C.., 1 XI +C.w2X2-£.w
where£.., denotes the imports of the intermediate good, then at constant world prices
and with fixed supplies of primary factors, it is readily shown that
axl xlxlcLlcL..,(kl-k.w>
oE.., LKI).I
and
axl xlxlcLlcL.w(kl-k..,)
-=-
oE.., 1;,1
If k 2 > k 1 > k.., and if X2 is capital-intensive in gross terms relative to X1 , then
1).1 > 0 and oXdoE.., > 0 and oX2 /oE.., < 0. Under these conditions the import of
M causes a rise in the output of X 1 and a decline in the output of X2 at given world
prices. On the other hand, if k 2 > k .w > k 1, then both outputs rise as a result of the
importation of M when 1).1 > 0. Hence the new production point P* may or may not
reflect the decline in the output of X2 .
198 Studies in the Pure Theory of International Trade

For this reason, if we assume that sign lA. I = sign (k 1 -k 2 ), it follows


that the shift in the transformation curve from H H' to TT' will be
biased towards X1 . Third, the length CP* measures the value of the
imports of Min terms of the second commodity and, in equilibrium,
this equals the value of the exports of X2 • This follows from the fact
that the change in the value of home production, equal to CP* in
Fig. 8.1, must be equal to the amount of the increase in the avail-
ability of the intermediate input multiplied by its price. Expressed
in terms of the second commodity, this equals PMEM/p 2 , where EM
is the amount of M imported; that is,t

ButpMEM/P 2 is nothing but the value ofthe imports of M expressed


in terms of the second commodity. With balanced trade, this must
equal the export of X 2 which, in Fig. 8.1, is also given by CP*. In
other words, the location .of the final production point P* satisfies
the balance-of-trade equilibrium condition, namely, PMEM = p 2 E,
where E is the amount of X 2 exported. In toto, Fig. 8.1 shows that as
a result of the switch from autarky to free trade, production moves
from S to P*, consumption moves from S to C, and welfare im-
proves from U1 to U2 •
An analogous but opposite procedure can be followed for the
case where the home country is an exporter of M and an importer of
X2 . Here, k 2 < kM with wh > w 1 under autarky. This case is pic-
tured in Fig. 8.2, where the switch from autarky to free trade takes
production from S to P*, consumption from S to C, and welfare
from U1 to U2 • CP* here equals the imports of X 2 as well as the
value of the exports of M expressed in terms of the second com-
modity, with the demand for X 1 equal to the domestic production
of X1 .

t At constant prices, the increase in the total value of domestic output resulting
from the importation of EM units of the intermediate good is given by

ilX1
[- ax2]
+ p - EM.
oEM oEM

This can be easily seen to be equal to PMEMIP 1 in terms of the first commodity, and
hence to PMEM/p 2 in terms of the second.
Pure Intermediate Products 199

Figure 8.2

8.8 Appendix
With u{k denoting the partial elasticity of substitution between fac-
tors i and k in the jth sector, and pij the direct share of the ith factor
per dollar of the jth good, the differentiation of cij = Cij(w,r,pM)
yields

(A8.1)

wherej = 1,2. Similarly, from ciM = ciM(w,r) (i = L,K),


ClM = PLMufLw*+uKMufKr*
(A8.2)
c:M = PLMu~w*+PKMulfKr*.
Equations (A8.l) and (A8.2) may be simplified by noting that

PAt = PLMw* + PKMr* (A8.3)


pLjuiL = -(pKjuiK+PMju{M) < 0 (j = 1,2) (A8.4)
PKiukK = -(PLPiK+ PMiUkM) < 0 (j = 1,2) (A8.5)
200 Studies in the Pure Theory of International Trade

PMiu{,M = -(PLicr{M+ PKP~M) < 0 (j = 1,2) (A8.6)


PLMcr'fL = - PKMcr'fK < 0 (A8.7)
and
(A8.8)
Substituting into (A8.1) and (A8.2) gives
Cti = -(PKicr{K+PMiPKMcr{M)(w*-r*)
c:j = (PLicr{K+ PMiPLMCT~M)(w*- r*) (A8.9)
Ct,i = (PLiPKMcr{M-PKiPLMCT~M)(w*-r*)
and
CiM = - PKMcr'fK(w*- r*)
(A8.10)
c~M = PLMcr'fK(w*- r*).
From the definition of Rii• we may differentiate logarithmically
RLI to obtain
PL! Ctl + PLMPMl(CiM+ Ct,l)
(A8.11)
(}Ll

Substituting for Ct 1 , CtM and CZ, 1 from (A8.9) and (A8.10) then
gives
(A8.12)

where IX = PLtPKtcriK+ PiMPLIPMtcriM+ PiMPKtPMtcriM


and ~ = PLMPKMPMtcr'fK > 0.
Similarly, it can be shown that
R~t (IX+~)
(A8.13)
(w*-r*) (}Kl
R!2 -(f3+t5)
(A8.14)
(w* -r*) (}L2
and
Rl2 ({3 + {j)
(A8.l5)
(w*- r*) (}K2
where P= PL2PK2cr£K+PiMPL2PM2cr£M+PiMPK2PM2criM
and (j = PLMPKMPM2cr'fK > 0.
Pure Intermediate Products 201

From (A8.12)-(A8.15) it is readily seen that YL and Yx are positive


if ex and f3 are positive. We shall now show that ex > 0. Now, from
(A8.4)-(A8.6), it can be ascertained that

1
aLL=
1
-2-
(2 1 2 1
Px10"xx+ Px1PM10"xM+PM10"MM ·
2 1) (A8.16)
PLl
Since ah < 0, the quadratic form on the R.H.S. of (A8.16) must
be negative definite, implying that
akxa1M-(akM) 2

or
alx > -pM1alMakM
----~------~~
(pLlalM + Px1 akM)"
Substituting this in the expression for ex, we find that

or
PM1 (
1 +
1 1 )2
!X> ( 1 ) PKMPL10"LM-PLMPK1(1KM .
PLlO"LM Px10"KM
Since (pL 1alM+Px 1aicM) = -pM 1a1M > 0, it is readily seen that
ex > 0. Similarly, it can be shown that f3 > 0.

REFERENCES
[I] Batra, R.N., and Casas, F. R., 'Intermediate Products and the Pure Theory of
International Trade: A Neo-Heckscher-Ohlin Framework', American Economic
Review, LXIII (June 1973).
[2] Casas, F. R., 'Pure Intermediate Products, Factor Intensities and Technical
Progress in the Theory of International Trade', Southern Economic Journal,
XXXIX (July 1972).

SUPPLEMENTARY READINGS
[3] Batra, R. N., and Singh, R., 'The Intermediate Products and the Two-Sector
Growth Model', paper presented at the Econometric Society Meetings, Detroit
(Dec 1970).
[4] Khang, C., 'A Dynamic Model of Trade between the Final and the Intermediate
Products', Journal of Economic Theory, I (Dec 1969) 416-37.
[5] Kuo, C., 'A Two-Sector Growth Model with an Intermediate Product in an Open
Economy', Ph.D. dissertation, Department of Economics, Univ. of Western
Ontario (Apr 1972).
9 The Theory of Effective
Protection

Why do countries impose tariffs? The main objective of the tariff-


imposing country, apart from curbing imports and possibly captur-
ing the advantage of its natural monopoly power in trade, is to
encourage the domestic production of the import-competing indus-
tries whose survival is threatened by foreign competition. Until
recently, the protective effects of a tariff on the imports of a final
good were taken for granted. It was generally believed that a higher
rate of tariff or a higher nominal rate of protection would lead to a
higher output level for the protected commodity. With the recogni-
tion by trade theorists of the importance of trade in intermediate
goods, a new measure of protection, widely known as the effective
rate of protection (E.R.P.), has been evolved to take into account
the implications of the entire tariff structure, rather than tariffs on
individual imported commodities, for the pattern of output in the
economy. In contrast to the theory of nominal protection, which
analyses the impact of tariffs on the total value of final goods, the
E.R.P. measures the percentage change in the contribution of the
domestic primary factors or, what is the same thing, in the value-
added of an industry as the economy moves from a free trade situa-
tion to one of protected trade. The ramifications of the new measure
of protection are far-reaching. A number of empirical studies have
been made concerning the tariff structure of several developed as
well as underdeveloped countries,t and the results vary from
interesting to unbelievable. Balassa [ l ], for instance, has discovered
that the effective tariff on ingots and steel forms in the U.S. is almost
ten times as large as the nominal rate, whereas the corresponding
rate on agricultural machinery is actually negative.

t See, for example, Barber [2], Balassa [!], Basevi [3], Grube! and Lloyd [8] and
Travis [ 17] among countless others.
The Theory of Effective Protection 203

A simple example will illustrate the difference between nominal


and effective protection. Suppose a locally produced, import-
competing commodity is sold at the fixed world price of $100, and
of this the contribution of domestic primary factors is $40 and that
of intermediate goods, including the imported inputs, is $60.
Suppose further that the prices of all non-primary inputs equal the
world prices, and that all factors are used in fixed proportion. If a
nominal tariff of 10 per cent is imposed on the importable good
alone, its domestic price rises to $110, but since the prices of inter-
mediate goods are unchanged, the contribution of the primary
factors or the value-added increases by the full amount of $10; the
percentage change in the value-added, or the E.R.P., under these
conditions is 25 per cent. The E.R.P. in this case exceeds the nominal
tariff of I 0 per cent. Indeed, the effective rate varies inversely with
the proportion of value-added in the total value of the output. For
example, if in the free trade equilibrium the value-added equalled
$20, the E.R.P. would be 50 per cent. Although this example is based
on very simplified conditions, it does suggest that small shifts in
nominal tariffs can lead to relatively large changes in effective rates.
Next, suppose that a nominal tariff is also imposed on the im-
ported inputs. If this rate also equals 10 per cent, the E.R.P. equals
the two nominal rates, because in this case the value-added would
increase by $4. Thus if the nominal tariff rates on the imports of final
and intermediate goods are the same, the E.R.P. equals the nominal
rates. On the other hand, suppose the tariff rate on the imported
input equals 5 per cent; then the value-added increases by $7 and the
E.R.P. is 171 per cent. Thus if the nominal tariff on the final products
exceeds that on the intermediate product, the E.R.P. is greater than
both the nominal rates. By contrast, if the nominal tariff on the final
good falls short of the one on the imported inputs, the E.R.P. is less
than both nominal rates. For example, suppose the nominal tariff
on the material input is 15 per cent. Then the value-added increases
by only $1, which means that the E.R.P. equals 21 per cent. Of
greater interest is the case where the E.R.P. may be negative. This
would happen if the nominal rate on the imported input was, say,
20 per cent, for this would imply a decline in the value-added of $2
and an E.R.P. of -5 per cent.
In general, the concept of effective protection incorporates two
elements that generate conflicting pulls on the output level of the
204 Studies in the Pure Theory of International Trade

protected commodity: (i) the element of subsidy implicit in the grant


of the tariff on the final good tends to raise its output, and (ii) the tax
element involved in the tariff on the imported input, by raising the
production cost, tends to lower its output. The final result, which is
the outcome of these two opposing forces, is that the output of the
final good may or may not rise. This suggests that there is a close link
between the sign of the E. R.P. and the direction of the change in the
output of the import-competing final product. This indeed turns out
to be the case in the partial equilibrium analyses which bulked large
in the development of the earlier E.R.P. literature. In this sense at
least, the prognosis of the 'new' theory of protection constitutes a
surprising departure from most other recent developments in trade
theory, where the analytical ingredients have been provided by and
large by the two-factor, two-good, general equilibrium model. The
earlier contributions by Barber [2], Johnson [10], Balassa [1] and
Corden [5], among many others, were all cast in the partial equili-
brium mould. More recently, however, trade theorists have begun
to take stock of the deficiencies in the earlier E.R.P. literature by
using general equilibrium models, which, of course, is the only
correct approach. For, as suggested above, the interest in the E.R.P.
concept was sparked primarily by economists' concern with the
resource-allocational implications of a country's tariff structure, and
the partial equilibrium models are not adequate for this type of
analysis. Yet the partial approach is not without its merits. Apart
from being simple, it is perhaps the only way through which the
effective tariff rates for various industries and countries can be
empirically computed.
In this chapter we begin with a brief account of the partial
approach and then tum to the implications of effective protection for
resource allocation in a general equilibrium model whose basic
framework is one already developed in the previous chapter.

9.1 The E.R.P. Concept and the Partial Equilibrium Approach


The partial analysis of effective protection uses the familiar concepts
of demand and supply for an importable commodity; however, since
this particular approach ignores the income effects arising from the
introduction of tariffs, there is no need to use the demand curve. t
t See Leith ((13] p. 74) for this observation. What it really means is that the demand
curve does not shift as a result of the imposition of the tariff, implicitly assuming that
national income is constant. Hence nothing is lost if the demand curve is ignored.
The Theory of Effective Protection 205

S'

lz
I
I
I
I
I
I
----,T
N I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
0 E G w F

Figure 9.1

Fig. 9 .I accordingly depicts only the supply curve, which is positively


inclined like the S curve. In order to ensure that the supply curve has
the usual positive slope, it is necessary to assume under the partial
equilibrium approach that the supply of some inputs is less than
perfectly elastic. Under the small country assumption, the imported
inputs are available at constant world prices. Therefore, if the supply
curve is to be less than infinitely elastic, it is necessary to assume that
the supply elasticity of at least one of the primary factors falls short
of infinity. For the sake of simplicity of diagrammatic construction,
206 Studies in the Pure Theory of International Trade

we assume that there is only one primary factor, labour, whose


supply curve is given by S L in Fig. 9 .I, where w and p 2 are measured
along the vertical axis and X 2 and labour supply are measured along
the horizontal axis. If we assume that all inputs are used in fixed
proportions, the supply curve for X 2 can be drawn parallel to SL, as
is the case in the diagram. t Let the labour-output coefficient, L/ X 2 ,
equal unity. Then any point in the diagram represents the supply of
X 2 as well as the amount of labour utilisation corresponding to any
price and the wage rate. At the given world price OA, for example,
the output of X 2 and the amount of labour used equal OE, and the
unit value-added under free trade divided by the price is given by
Ell DE = wL/p 2 X 2 , which is nothing but the relative share oflabour
in the total value of output. If a nominal tariff t 2 is imposed on the
imports of the final good, the local price of X 2 rises to OB where
OB = OA(l + t 2 ). In addition, if a tariff (tQ) is imposed on the
imports of the intermediate good, Q, the supply curve shifts upwards
to the solid line S' to reflect the consequential rise in the unit cost.
In the absence of tQ, the output of X 2 would have risen to OF; how-
ever, in its presence the output rises only to OG. The actual or net
stimulus to the output of X 2 is therefore given not by AB but by
AC = ac, which is the difference between the increase in the local
price of X 2 and the increase in the unit cost of production at the new
level of output, i.e. AC = AB- BC. Now the increase in the unit cost
equals (!l.pQQ/X 2 ), where PQ is the given world price of the imported
input and !l.pQ is the tariff-induced change in this price to the local
producers. Hence the net stimulus to the output of X 2 is given by

AC = !l.f'2 -!l.pQCQ 2 , where CQ 2 = Q/X2 •

The effective protection rate (e 2 ) is then obtained by dividing AC by


the contribution of the primary factor or the value-added (v 2 ) in free
trade, so that

t If all factors were available at constant factor prices, the supply curves for the
goods would be infinitely elastic. If one factor has an elasticity of supply less than
infinity, the supply curves for goods will have the identical elasticity only if that factor
is used in fixed proportions. If it is not, the rise in the unit cost would be less than the
rise in the factor price, and the supply curves for goods and the factor would possess
different elasticities.
The Theory of Effective Protection 207

where ()Q 2 = pQCQ 2 /p 2 is the share of the imported input in the total
cost of the imported final good, and
_ pz(I+tz)-pz _ Apz d _ PQ(I+tQ)-PQ _ ApQ
t2 - - - an tQ - - .
Pz Pz PQ PQ
If many imported inputs are used in the production of an importable
commodity, then the E.R.P. formula for thejth industry is given by
n
tj-
j; I
I eijti
ej = ---n--

1- I
j; I
eij

where ()ii is the share of the ith imported input in the total cost of the
jth good. It can easily be seen that this formula measures the pro-
portionate change in the value-added that occurs as the economy
switches from a free trade situation to one involving tariffs. Let v~
be the value-added per unit of output in the second industry when
tariffs are introduced. Then

and

so that
v~- v 2 t 2 - ()Q 2 tQ
e2 = -- = .
v2 I-{}Q2

It is clear from this formula that


(i) if t 2 = tQ, e2 =t 2 =tQ;

(ii) if t 2 > tQ, e 2 > t 2 > tQ; and


(iii) if t 2 < tQ, e 2 < t 2 < tQ.
In terms of Fig. 9.1, the E.R.P. formula implies that
AC ac MN
e2 = JE = JE = GN.

The derivation of the E.R.P. formula from Fig. 9.1 serves to show
that the sign of the formula is linked with the direction of the change
208 Studies in the Pure Theory of International Trade

in the output. For example, if the input tariff, ta, is sufficiently large
and raises the supply curve beyond S' to an extent that increased the
unit production cost above the increase in p 2 , the E.R.P. would be
negative. But then the output of X 2 would also decline. Hence, under
the partial approach, there is a one-to-one relationship between the sign
of the E.R.P. and the direction of the shift in the output of the protected
commodity. A positive change in the output implies a positive E.R.P.,
and vice versa.
This conclusion continues to hold even when we allow substitution
among inputs, which tends to lower the increase in the unit produc-
tion cost because the possibility of variablity in production coeffi-
cients permits savings in costs by giving the producers the oppor-
tunity to economise on the use of inputs with increased prices.
Indeed, the higher are the substitution elasticities among inputs, the
smaller would be the upward shift in the unit cost. In Fig. 9.1, the
supply curve shifts only to the dotted line S" when substitution
among inputs is permitted. The output of X 2 now equals OW and
the net stimulus to output is given by AR > AC. Hence, other things
remaining equal, the substitution among inputs gives rise to a higher
E.R.P. Under these conditions

v2 = P2(1+t2)-[Ca2PaO+ta)+Pa~Ca2J
so that the E.R.P. formula is given by
, t 2 -8a 2(ta+q2)
e2 = (9.2)
l-8a 2

where C~ 2 = ~Ca 2 /Ca 2 is negative, which simply implies that a rise


inpa, with all other input prices unchanged, leads to a decline in Ca 2 •
With q 2 < 0, it is clear that e2 > e2 • In Fig. 9.1, e2 is given by
ARjJE = TZ/WT. However, in spite of all these changes, the rela-
tionship between the E.R.P. and the output of the protected com-
modity is still positive, as can be seen from Fig. 9.1.

9.2 Negative Value-Added


One result, which has been the cause of some controversy, concerns
the appearance of negative value-added in some empirical studies.
Several authors have noted that in some cases the free trade value
of inputs exceeds the corresponding value of outputs, so that the
value-added at world prices for some industries is negative. Con-
The Theory of Effective Protection 209

siderable debate has been sparked by this discovery, which at first


glance appears implausible. Several misgivings have been raised
concerning the meaning and the legitimacy of the concept. The
issue, however, has recently been resolved by Guisinger [9] who,
apart from clarifying the concept, has also shown that the negative
value-added is neither the result of the constancy of production
coefficients, as contended by some economists, t nor that of the gross
inefficiency in the production system, as asserted by others.t The
reason why negative value-added has occasionally appeared in some
empircal studies is attributable to the fact that there is no way of
getting actual data on production costs under the free trade situation,
because in general few countries follow the policy of universal free
trade. Thus most empirical studies have begun with the observed
data on the tariff-distorted cost structure in various industries, and
worked their way backwards in computing production costs in the
free trade situation. The value-added in the absence of the tariffs is
then calculated by deducting the value of inputs from the value of
outputs at the observed world prices. This procedure is, of course,
open to several objections; but even if the procedure yields an
accurate approximation, the concept of the negative value-added is
not so intractable. The negative value-added at world prices simply
implies that the commodity in question cannot be produced in the
absence of the tariff and that it is being currently produced only
under the umbrella of the protective tariff. Alternatively, the exist-
ence of the negative value-added at world prices implies that the
industry concerned will have to be granted a production subsidy in
the absence of the tariff. That an industry may not exist in the
absence of the tariff is hardly surprising. As suggested by Guisinger,
it simply implies that at free trade prices the local supply curve of the
import-competing industry lies wholly above the world supply curve.
For further exposition of the concept of the negative value-added,
consider Fig. 9.2, which is drawn on the same basis as Fig. 9.1; OB
is the price of X 2 which can be observed in the protected trade situa-
tion and OG is the output of X 2 • Now if all tariffs are removed, the
local supply curve will shift from S' to Sand the local price of X 2
would decline to the free trade price OA. However, at this price
nothing will be produced locally because, for the output to be

t See, for example, Leith [12].


~ See Ellsworth [7] for this view.
210 Studies in the Pure Theory of International Trade

0 G
Figure 9.2

positive, the price must be greater than OF. Since production is


assumed to be taking place under competitive conditions, this im-
plies that the local unit cost or the total value of all inputs per unit of
output exceeds the free trade price at all levels of output. All this
means that the value-added at free trade prices is negative.

9.3 Effective Protection in General Equilibrium


The formulation of the concept of effective protection in terms of
formulae (9.1) and (9.2) derives from the percentage change in the
value-added as we move from a laissez-faire situation to one of
protected trade. It was established that a positive E.R.P. implied a
rise in the output of the protected final good above its free trade
level, irrespective of whether the production coefficients were fixed
or variable. We shall now examine these results in a general equi-
The Theory of Effective Protection 211

librium model where one of the two final goods is produced with the
help of two primary factors, K and L, and two intermediate products,
one of which is domestically produced but not traded, and the other
which is imported but not locally produced. The main difference
between this model and the production models used so far is that
in the presence of traded inputs the effective factor supplies become
variable, for the traded inputs can be imported (or exported) at
world prices which are given under our small country assumption
that we maintain throughout. This is true in spite of the constancy
of the supply of domestic primary factors which is generally assumed
in production models without traded inputs. Thus, in the present
model, the prices of some inputs are exogenously given but the
demand for them is endogenously determined, in direct contrast to
the models used up to the previous chapter, where factor supplies
were exogenously given but factor prices were dependent variables.
This difference is crucial to the existence of paradoxical results that
have appeared in some general equilibrium analyses attempting to
explore the consequences of the E. R.P. for resource allocation.
Unless otherwise specified, we assume throughout our analysis
that there are three domestically produced commodities: X 2 , the
importable good, X 1 , the exportable good, and M, the non-traded
intermediate product, which is produced solely to be used as an input
in the production of the importable good. A fourth commodity, Q,
is imported from abroad and used as an input again in X 2 but not in
X1 . The material inputs could also be utilised in X 1 , but this would
unnecessarily complicate the analysis without adding anything new
to the exposition. There is perfect competition in all markets, pro-
duction functions exhibit constant returns to scale and diminishing
marginal rates of substitution, and two primary factors, labour (L)
and capital (K), are inelastically supplied and fully employed. World
prices of all traded commodities are constant, but domestic prices
of the importables as well as the non-traded input may be altered
owing to the imposition of tariffs.
Let Cii represent the amount of the ith factor per unit of the jth
product (i = L,K,M,Q;j = 1,2,M). With full employment,
(9.3)
and
(9.4)
212 Studies in the Pure Theory of International Trade

Since M =CM X 2 2, these two equations can be written as


cLlxl +(CL2 + cLMCM2)X2 =L (9.3A)
cKlxl +(CK2 + cKMCM2)X2 = K. (9.4A)

The price equations are now given by


eLl w+ CKlr = Pl (9.5)
CL2w+CK2r+CM2PM+CQ2PQ = P2 (9.6)
and
(9.7)

Substituting (9.7) in (9.6), we obtain


(9.6A)

The production coefficients in turn are determined by the following


equations:
Ci2 = Ci2(w,r,pM,PQ) (i = L,K,M,Q) (9.8)

cil = cil (w,r) (i = L,K) (9.9)


and
(i = L,K) (9.10)
The production side of the model described by the system of equa-
tions (9.3)-(9.10) can be solved as follows. First, the three equations
(9.5)-(9.7) containing three parameters,p 1 ,p 2 and PQ• can be solved
to obtain the three unknowns, w, r and PM• in terms of the inter-
national prices of the traded goods and the input-output coeffi-
cients. We assume that this solution is unique. With w, r and PM so
determined and with PQ exogenously given, Cii's can be solved from
(9 .8)-(9 .I 0). These values in turn can be substituted in (9 .3A) and
(9.4A) to derive expressions for X1 and X 2 in terms of the given
supplies of K and L; finally, CM 2 and X 2 can be used to obtain M.
Hence the set of equations (9.3)-(9.10) specifies a completely deter-
minate production side of the model. The demand side will not be
presented, since the focus of this chapter is on resource allocation
and factor-price changes brought about by changes in the prices of
the two importables, X 2 and Q, through the imposition of tariffs.
The Theory of Effective Protection 213

Factor intensities in the final products are defined by columns of


the C matrix which is formed by incorporating the Ci/s appearing
in the full-employment equations (9.3A) and (9.4A):

As regards the price equations (9.5) and (9.6A), the factor intensities
are specified by the rows of the C matrix, even though it does not
contain CQ 2 ; the reason for this asymmetry is attributable to the
fact that, although CQ 2 enters into the determination of p 2 , it absorbs
no proportion of domestic primary factors, because Q is produced
abroad. Since it is the proportion of the domestic primary factors
used in the production of each commodity that determines its factor
intensity, the use of Q by X 2 causes no modification to the factor
intensities.
As in the previous chapter, the presence of the domestically pro-
duced intermediate good (M) makes it necessary to introduce a dis-
tinction between the net and the gross capital/labour ratio in the
second industry. For analytical convenience, we assume that X 2 is
capital-intensive relative to X 1 in both the gross and the net sense.
This implies that

ICI = CL1(CK2+CKMCM2)-CK1(CL2+CLMCM2) > 0


because

The Structural Relations


The next step in facilitating the exposition is to derive expressions
portraying the structural relations in the model. As before, let an
asterisk denote the relative rate of change in a variable. Under the
procedure that has been used repeatedly in the previous chapter, the
following equations can be obtained by differentiating the system of
equations presented above:

ALl Xi+ (A.L2 + A.LMA.M2 )X!


= -(ALl C!1 +AL2q2 +ALMAM2CtM+A.LMAM2C:12) (9.11)
214 Studies in the Pure Theory of International Trade

A.K 1 Xt + (A.K2 + A.KMA.M2)X!


- (A.Kl ql + A.K2 Cl2 + A.KMA.M2 qM + A.KMA.M2 qf2) (9.12)
(9.13)

(}L2w*+lJK2r*+lJM2P"k = P!-0a2Pa (9.14)

(}LMW* + (}KMr* = p%, (9.15)

where in this chapter A.ii equals the proportion of the ith factor used
directly in the production ofthejth good, and (}ii is the relative share
of the ith input in the jth good. Substituting (9.15) in (9.14) yields

Let A. and 0, respectively, be the matrices of A. and(} coefficients on


the left-hand side of (9.11) and (9.12) and (9.13) and (9.16). Then,
since A.M 2 = 1,

and

where the determinants of A. and (} are given by


lA. I = A.Ll (A.K2 + A.KM)- A.Kl (A.L2 + A.LM)
xlx2 J
= KL [CLl(CK2+CKMCM2)-CKl(CL2+CLMCM2

= xlx21CI
KL
and
IOI = (}Ll (OK2 + (}KM(}M2)- (}Kl (OL2 + (}LM(}M2)
= ~ [CL1(CK2+CKMCM2)-CK1(CL2+CLMCM2)]
P1P2

=~ICI·
P1P2
The Theory of Effective Protection 215

Evidently, the signs of jA.j and jOj are the same as the sign of jej,
which implies that both IA.I and 101 are positive if X 2 is capital-
intensive relative to X 1 in the gross sense. The determinant of 0 can
be written in yet another manner which highlights the role played
by the imported input. Let t/IL 2 = OL2 + OLMOM 2 be the gross share
of labour and t/JK 2 = OK 2+0KMOM 2 be the gross share of capital in
the second industry. Since factor shares in any industry add up to
unity,

and

so that
jOj = OL1t/IK2-0K1t/1Lz = OLlt/JKz-(1-0Ll)(l-OQz-t/IK z)
= t/JKz-0K 1(1-0Q2) (9.17)
or
(9.18)

If no imported input was being used in X 2 , OQ 2 would be zero and


X 2 would be capital-intensive relative to X 1 in the gross sense if
t/IK2 > OK1 from (9.17) or if eLl > t/IL2 from (9.18). The presence
of payments for the use of the imported input, however, gives rise to
a decline in the gross share of labour and capital in the second
industry. Here X 2 is capital-intensive in the gross sense only if either
t/IK2 > OK1(1-0Q2), or eLl > t/1Lzl(l-OQ2). Since (l-OQ2) < I,
jOf > 0 only if OL 1 > t/IL 2, but t/IK2 need not exceed OK 1. These
remarks will be fruitful in facilitating the subsequent analysis.

9.4 Effective Protection and Real Wages


Some of the structural relations presented in the preceding section
may now be solved to derive the implications of the E.R.P. for real
rewards of the primary factors. As noted in Chapter 5, according to
the Stolper-Samuelson theorem nominal protection raises the real
reward of the factor employed intensively by the importable com-
modity and lowers the real reward of the factor employed intensively
by the exportable good. We now examine the impact of the E.R.P.
on real wages and see under what conditions the primary factor
216 Studies in the Pure Theory of International Trade

employed intensively by the importable good stands to benefit from


effective protection. Solving (9.13) and (9.16) yields

w* = -w
eKl < •
P2-
eQ2PQ*) = eKl
-we2
(l -
eQ2 ) (9.19)

and

r
* -- eLl < *
TOT e *) eLl (1 - eQ2 )·
P2- Q2PQ = TOf e2 (9.20)

In obtaining (9 .19) and (9 .20) we have equated p! to zero in view of


the fact that, with given world prices of all tradables, only p 2 and PQ
are subject to change as a result of tariffs. It is clear from these two
equations that with JeJ > 0, w is negatively and r positively related
to the E.R.P. on X 2 • Since

the proportionate rise in r exceeds e 2 • However, it is by no means


certain that the proportionate decline in w also exceeds e 2 , for

need not exceed unity. Furthermore, although the grant of effective


protection to X 2 causes a rise in r and a decline in w, does it also
result in a rise in the real reward of capital and a decline in the real
wage rate in terms of the two final commodities? Since p 1 is constant,
a decline in w and a rise in r amount to a decline in the real wage rate
and a rise in the real reward of capital in terms of the first com-
modity. Furthermore, since P! > 0, the decline in w also implies a
decline in the real wage rate in terms of the second commodity.
However, the outcome is not so clear for the real reward of capital
in terms of the second commodity. Subtracting p! from both sides
of(9.20), and using (9.18), we obtain
The Theory of Effective Protection 217

With 101 > 0, r* > P! only if

P! > eLl eQ2


Pa eLleQ2+1/JL2.
(9.21)

Since the right-hand side of (9.21) is less than unity, a sufficient


condition for the real reward of capital to rise is that p! ~ Pa· In
other words, if the nominal tariff on the final product exceeds or
equals that on the intermediate good, Q, the E.R.P. results in a rise
in the real reward of capital in terms of both final goods. Even if
p! < Pa. the real reward of capital may rise unambiguously, pro-
vided the share of the non-traded material input, M, is sufficiently
large to confer high value on 1/1 L 2 .
How does all this fit into the Stolper-Samuelson theorem? Evi-
dently, the theorem continues to hold in terms of the E.R.P. if(9.21)
is satisfied. However, if the output tariff is less than the input tariff,
the theorem may not be valid even if the E.R.P. itself is positive.
An intuitive explanation is not hard to find. According to the E.R.P.
formula furnished by (9 .1 ),

Now r* always exceeds e 2 . Hence if t 2 ~ tQ, r* will necessarily


exceedp!.
Of greater interest is the fact that the E.R.P. always raises the
relative reward of capital, as is evident from

even though it may not unambiguously raise the real reward of


capital. Thus we are back in the pre-Stolper-Samuelson world, for
the main contribution of Stolper and Samuelson was to demonstrate
that (nominal) protection moves the real and the relative rewards of
a factor in the same direction. With effective protection, however,
this may no longer be true, and we confront the same 'index number'
problem which, prior to the Stolper-Samuelson contribution, is
believed to have stood in the way of reaching definite conclusions
concerning the effects of protection on income distribution.
218 Studies in the Pure Theory of International Trade

9.5 The E.R.P. and Resource Allocation


The results derived in the previous section have direct relevance to
the question of the implications of the E.R.P. for the allocation of
resources. In the two-factor, two-commodity model without im-
ported inputs, nominal protection always results in the shift of
resources away from the unprotected commodity towards the pro-
tected commodity, and in the process raises the output of the latter
at the expense of the output of the former. Does this result continue
to hold when nominal protection gives way to effective protection?
The answer depends wholly on how factor prices change in response
to the introduction of the E.R.P.
We have shown in the preceding section that the E.R.P. conferred
on the capital-intensive commodity, X 2 , raises the reward of capital
(though not necessarily its real reward) and lowers the wage rate. If
the supply of the imported input were fixed, with given supplies of
capital and labour, this would ensure a rise in the output of X 2 , a
decline in the output of X1 and a rise in the X 2 /X 1 ratio. However,
when the world prices are given and an intermediate good is im-
ported at these prices, the supply of the imported input is not
constant, nor does its demand remain unchanged, especially when
substitution between the domestic and imported inputs is permitted.
Is it then possible that X 2 /X 1 may actually decline even though
effective protection was provided to X 2 ? In this section we first
show that such a 'perverse' result is within the realm of possibility,
and then derive the conditions that are necessary for the existence
of this outcome.
With this purpose in mind, it is natural for us to revert to (9.11)
and (9.12). However, the solution of these two equations will not
yield any meaningful results unless we obtain the lengthy but indis-
pensable expressions for Ct. Differentiating (9.8)-(9.10), we get

C!1 = OLl uh w* + OKl ulKr*


Cl1 = OLl ulKw* + OKl u}cKr*
,.,. -- 0L20'LLW
2 2 *+0 M20'LMPM
*+0 K20'LKr 2 • +0Q20'LQPQ
2 •
'--i.2
,.,.
'--ic2 -0
- L20'LKW
2 *+0 K20'KKr
2 *+0 M20'KMPM
2 *+0 Q20'KQPQ
2.

,.,.
'--M2 -0
- L20'LMW
2 *+0 K20'KMr
2 *+0 M20'MMPM
2 *+0Q20'MQPQ
2.

C!M = 0LMui"Lw*+0KMui"Kr*
qM = OLMui"Kw* + OKMO'~Kr*
The Theory of Effective Protection 219

where the (}'s are as defined and u{k as usual is the partial elasticity of
substitution between factors i and kin thejth industry. Substituting
these in (9.11), we obtaint
A.LlXT +(A.L2 +A.LM).f!
= /h(w*-r*)+/Ja 1 (w*-p~)+/Ja 2 (r*-p~) (9.22)
where
PL = A.Ll(}Ktuh+A.L2(}K2ulK+A.LM(}KMufK
+ AL2 (}M2 lJi:MulM + ALM(}K2 (}LMO'i:M
Pat = A.L2(}Q2ula+A.LM(}LM(}Q2u'ita
and
Pa2 = A.LM(}KM(}Q2 u'ita·
Performing analogous substitutions in (9.12) yields
AKtXT + (A.K2 + A.KM).fl
= -PK(w*-r*)+/Ja 3 (w*-p~)+/Ja 4 (r*-p~) (9.23)
where
PK = A.Kl(}Lluh+A.K2(}L2ulM+A.KM(}LMufK
+ AKM(}KM(}L2 ulM + AK2 (}M2 lJfMui:M
Pa3 = A.KM(}LM(}Q2u'ita
and
Pa4 = A.K2 lJa2 ui:a + A.KM(}KM(}Q2 u'ita·
Subtracting (9.23) from (9.22) and performing some manipulations,
we derive
IA.I(X!-X!) = (PL +PK)(w*-r*)+A.L20Q2uia<w*-po)
+ A.K2(}Q2ui:a(p~- r*) + lJa2 (A.LM- A.KM )ui,.a(ptt- P~). (9.24)
From (9.19) and (9.20) we get

(9.25)

t As in the last two chapters, the expressions for c,7 can be simplified by noting
that, if all input-price ratios were constant, Cij = 0. This is what has been done in
obtaining the following equations.
220 Studies in the Pure Theory of International Trade

- eKtP! + <eKl -1/1 K2)Pa


w•-pa = IOI (9.26)

-8LtP!+(8Ll-l/JL2)pa
pa-r• = IOI (9.27)

Subtracting pa from both sides of (9.15) and using (9.26) and (9.27),
we obtain
* * ((JKM(]Ll -{JLM(]Kl)P!
PM-PQ = IOI
( ((JLM(]Kl- (JKM(]Ll)- ((JLM(JK2- (JKM(JL2)]pa
(9.28)
+ IOI
With this last equation we have collected all the ingredients necessary
to examine the implications of effective protection for the relative
outputs of the two final commodities. One of the basic differences
between the effective protection and the nominal protection model
is that, unlike the latter, there are several ways of conferring protec-
tion on a final commodity when tariffs on imported inputs are per-
mitted; for example, X 2 can be protected either by imposing an
output tariff, or by lowering the input tariff (or by subsidising the
imports of the input), or by any combination of the two such that
(p!- 8Q 2Pa) > 0. To begin with, we consider the two polar cases
which involve the imposition of the output or the input tariff. In a
model where only imported intermediate goods are allowed to exist,
Jones [11] has demonstrated that when protection is conferred
purely through output tariffs, the relative output of the protected
commodity must rise even when substitution is allowed between
primary factors and imported intermediate inputs; if protection is
provided through the other extreme way of lowering the input
tariff, the relative output of the protected good may or may not
rise, depending on whether or not the imported input is a better
substitute of one primary factor than of the other. We shall now
examine these results in the context of our model which allows for
the existence of two intermediate inputs, one non-traded and the
other imported from abroad.
Consider first the case where p~ is zero but p! > 0, so that pro-
tection is provided by the simplest way of the output tariff. Actually,
this case is no different from the conventional model of nominal
protection where intermediate inputs are ignored. First, we have to
The Theory of Effective Protection 221

determine the signs of fh and PK appearing in (9.24). Following


Jones, we assume that all factors are substitutes so that the cross-
elasticities are positive, that is, u{k > 0 (i =F k). tIt follows then that
PL and PK are both positive. Let us now examine (9.24) and the signs
of the four terms that make up the equation to determine the sign of
(X!- .\1). For analytical convenience, we continue to assume that
X 1 is labour-intensive relative to X 2 in the gross sense, so that IA.I
and 101 are positive. With p~ = 0, we know that (w*- r*) < 0,
w* < 0 and r* > 0. All this follows from the discussion in the
previous section. This' means that the first three terms on the right-
hand side of (9.24) are all negative, whereas the fourth term may be
positive or negative, depending on the signs of(A.LM-A.KM) and pt.
From (9.28), pt ~ 0 if kM :S k 1 ; on the other hand,

ALM-AKM =; (k-kM) ~ 0 if k ~ kM

where k = K/L is the overall capital/labour ratio in the economy.


It can easily be seen that if both k 1 and k are either greater or less
than kM, Pt and (A.LM-A.KM) have opposite signs and the fourth
term is negative, in which case (X!- X!) < 0, so that the output
tariff alone leads unambiguously to a rise in the relative output of
x2. However, if kM lies between kl and k, the fourth term in (9.24)
is positive, in which case it is no longer clear that (X!- X!) < 0.
The imposition of the output tariff alone may then lead to the paradox
of a decline in the relative output of the protected commodity. Note
that this possibility cannot arise if (i) there is no non-traded inter-
mediate input, (ii) there is no substitution possibility between the
non-traded and the traded inputs, so that u~Q = 0, or (iii) the
capital/labour ratio in the domestically produced input equals the
overall capital/labour ratio. These are sufficient conditions, and
obviously none of them is necessary to avoid the perverse possibility
of a decline in the relative output of the protected commodity as a
result of the introduction of the nominal (output) tariff. The neces-
sary and sufficient conditions for the paradoxical result can be
derived by substituting (9.25)-(9.28) in the general equation (9.24)

t There are two reasons why we make this assumption. First, it contributes to sim-
plicity, and second, we want to make sure that some paradoxical results derived below
do not depend on the special case where some factors are complementary to each other.
222 Studies in the Pure Theory of International Trade

to obtain
).A.)(X!-Xi)
2 2 e2(1-0a2)
= -[(fJL +fJK)+(AL20KtO'LQ+AK20LlO'KQ)OQ2] )0)
2 e 2(1-0a 2)
- [0Q2(ALM-AKM)(0Kl -OKM)aMQ] )0)
- OQ2 [(.A.L2ara- .A.K2aia) + (.A.LM- .A.KM)a~a]P~· (9.24A)
It can be seen that introducing t 2 alone can lead to the lowering of the
relative output of x2 if

({JLO+fJK) +.A.L20Ktara+AK20LlaiQ < (AKM-ALM)(OKl-OKM)a~Q·


Q2 (9.29)
Since the left-hand side of inequality (9 .29) is positive, it is clear that
the necessary condition for this inequality to be satisfied and the
paradox to occur is that the two terms within the parentheses of the
right-hand side of (9.29) have the same sign, which in turn implies
that kM lies between k 1 and k.t The likelihood of this paradox
increases if, ceteris paribus, afa and aiQ are close to zero and al,Q is
sufficiently large, that is to say, if the imported input is more substi-
tutive with the domestically produced input than it is with the
primary factors. All this discussion gives rise to the following
general theorem:
Theorem 9.1. When protection is provided by means of an output
tariff alone, the necessary condition for the relative output of the
protected commodity to decline is that the capital/labour ratio
of the non-traded intermediate product lie between the capital/
labour of the unprotected commodity and the overall capital/
labour ratio in the economy.

t If A.KM > A.LM• then k > kM. If IlK, > IIKM• the share of capital in the first industry
exceeds that in the second industry, which implies that k 1 > kM, as can be seen from
the fact that

and

where w = wfr.
Thus, if k > k M > k 1 , both expressions in the parentheses on the right-hand side of
(9.29) are positive. The same result holds if k < k 1 < kM.
The Theory of Effective Protection 223

Next, consider the other polar case where protection is granted to


the second commodity by subsidising imports of the intermediate
input Q, so that p! = 0 and p~ < 0. The first and third terms in
(9.24) are again negative, but the second and fourth terms may not
be. t With p~ < 0, the second term is negative if (w•- p~) < 0,
which from (9.26) requires that (}K 1 > t/IK 2 ; however, from (9.17)
we know that (}K 1 may be less than or greater than t/IK2 for jej > 0,
which introduces the possibility that (w*- p~) may be positive, that
is, the proportionate decline in w may fall short of the corresponding
decline in PQ as a result of the provision of subsidy to the imports of
the input Q. Thus even if the fourth term is zero or even negative,
(Xi- X!) may be positive. This is the result derived by Jones and,
as we have shown, the existence of the non-traded intermediate
good may attenuate or reinforce this result. The necessary condition
for Jones's result can be obtained as a special case by equating p!
and a~Q to zero in (9.24A) and using the definition of jej, so that
(Xi- X!) > 0 only if

(}Q2PH(PL +PK)+AL2afQ((}Kl -t/JK2)+AK2aiQ((}Ll -t/JL2)] > 0

which, with p~ < 0 and (}L 1 > t/IL2 for 1e1 > 0, requires that
(}K 1 < t/JK 2 . If the non-traded good does not exist, t/IK 2 reduces to
(}K 2 . Hence the necessary condition for the 'perverse' possibility to
arise is that the relative share of capital in the first industry be less
than that in the second industry, which is, of course, within the
realm of possibility, because it does not, as we have shown above,
conflict with the positive sign of 1e1.t When a non-traded material
input is introduced, but a~Q = 0, then the necessary condition
involves the comparison between the gross share of capital t/IK2 in
the second industry using the intermediate goods with the share of
capital in the first industry which does not utilise any intermediate
good. This discussion leads to the following generalised theorem
which applies to Jones's result as well as ours:
Theorem 9.2. Given that a~Q = 0, the necessary condition for the
relative output of the protected final good to decline, when
effective protection is granted by subsidising imports of the

tIt is perhaps necessary to remind ourselves here that with P! = 0, e2 (1-8Q 2 )


reduces to - 8Q 2 PZ·
t See the remarks made at the end of section 9.3.
224 Studies in the Pure Theory of International Trade

intermediate input, is that the (gross) relative share of the primary


factor employed intensively by the protected commodity exceed
the relative share of the same factor in the other commodity.
When a~Q > 0, this theorem continues to be valid if k = kM so
that the fourth term in (9.24) falls to zero; if the fourth term there is
non-zero, evidently 8K 1 < t/IK 2 may no longer be a necessary con-
dition for (X 1 /X 2 ) to rise.
The results derived in this section can be illustrated in terms of
what is well known as activity-analysis diagrams. Purely for the sake
of analytical convenience, we analyse the case where the non-traded
input does not exist, so that the output of M falls to zero. The full-
employment equations (9.3) and (9.4) are diagrammatically depicted
in Fig. 9.3, where X 2 is measured along the vertical axis and X 1 along
the horizontal axis. The given supplies of capital and labour together

E"

(L/CL2)
...........
E
fL/Cf2J

(K/ChJ

---
G
(KjCK2)

8 F' F x1
I !
(L/C[1J (LjCL1J

Figure 9.3
The Theory of Effective Protection 225

with the input coefficients furnish the lines EF and GH which are
~rawn by joining the points that determine the maximum output of
one commodity, with zero output of the other. For example, OE =
LjeL2 is the maximum output of X 2(with zero output of X 1 )t and is
obtained by employing the entire supply of labour in X 2 • Along EF
labour would be fully employed, along GH capital would be, and at
their intersection at J there would be full employment of both factors.
The diagram is drawn in such a way as to reflect our assumption that
X 2 is capital-intensive relative to X 1 . This can be verified by com-
paring the slopes of EF and GH, which are, respectively, given by
eLI;eL2 and eK 1 jeK 2 , so that eK 2 IeL 2 > eK 1 jeLI. At the produc-
tion point given by J, the output of X 1 equals OB and that of X 2
equals JB, and the ratio between the two outputs, X 2 /X1 , is given
by the slope of OJ.
If there were no imported inputs, then eii would be related only
to w and r and a rise in r and a decline in w consequent upon the
grant of protection to X 2 would lead to a rise in eLl and eL2 and a
decline in eK 1 and eKz· This follows from the assumption of linearly
homogeneous production functions. Let the prime on an input
coefficient denote its magnitude in the new situation involving
tariffs. Then
e~i > eLi and eii < eKi U = 1,2).
With given supplies of K and L, all the four points E, F, G and H in
Fig. 9.3 would move along the axes either up or down. For example,
with ef. 2 > eL 2 , pointE would move down to, say, E'. Similarly,
G would shift toG', Fto F' and H to H', and the new full-employment
equilibrium would be established at J' where the dotted lines E' F'
and G' H', representing, respectively, the labour- and capital-
constraint lines in the new situation, intersect. It may be observed
that X 2 and the (X2 /X 1 ) ratio have risen and X 1 declined in the new
situation. This is how nominal protection in the usual two-good,
two-factor model raises the output of the protected commodity at the
expense of the output of the unprotected commodity.
Let us now suppose that an intermediate good (Q) is also imported
in addition to X 2 • If the effective protection is granted only by intro-
ducing a nominal tariff on X 2 , the conditions of the model do not

t Assuming for the time being that each good could be produced with the help of
only one factor.
226 Studies in the Pure Theory of International Trade

undergo a substantial change; PQ remains constant and the only


factor prices to change are w and r, so that the results are qualita-
tiv~ly similar to those derived above from Fig. 9.3. Consider the
other extreme case where effective protection is granted to X 2 by
subsidising the import of the intermediate good, so that PQ declines
and (p!- (}Q 2 PQ) > 0 because P! = 0 and PQ < 0. This type of
effective protection also results in a rise in rand a decline in w. Since
the first industry does not use the intermediate product, CLI would
rise and CK 1 would decline as before, so that Fin Fig. 9.3 would
shift backwards, say again to F', and H would shift outwards, say to
H'. The outcome is not so clear in the second industry, where input
coefficients now depend on three factor prices, w, rand PQ· Here the
change in C Ll and C Kl is determined by the changes in w and r rela-
tivetothechangeinpQ.Sincer* > O,w* < OandpQ < O,both(r/w)
and (r/pQ) rise, so that substitution must occur against the amount
of capital employed in the second commodity and in favour of labour
and Q. In other words, CKl must decline. In addition, if PQ > w*,
that is, if the proportionate decline in the wage rate exceeds the
corresponding decline in PQ so that both (PQ/w) and (r/w) rise, sub-
stitution takes place against the imported input and in favour of
labour, so that CL 2 must rise. Thus if

r* > PQ > w* (9.30)

CK 2 declines and CL 2 rises, which in tum ensures the outward shift


of G to, say, G' and a backward shift of E to, say,£'. In other words,
the direction of shifts in the input -output coefficients is the same as
was the case when effective protection involved the imposition of a
tariff on the imports of X 2 alone. Thus if condition (9.30) is satisfied,
the E.R.P. granted to X 2 by subsidising the imports of Q leads to a
rise in X 2 relative to X 1 • However, if PQ < w* < r*, that is, if the
decline in w falls short of the decline in PQ so that (w/pQ) rises,
substitution in the second industry may take place against labour
and in favour of the imported input, and if this effect is sufficiently
large, C Lz may actually decline. Thus if

r* > w* > PQ (9.31)

CK 2 declines as before, but CL 2 may also decline which means that


point E shifts outwards along the vertical axis, and if this shift is
The Theory of Effective Protection 227

sufficiently pronounced, (X2 /X 1 ) may actually decline.t This possi-


bility has been depicted in Fig. 9.3, where E shifts toE" and the solid
line E" F', which now represents the labour-constraint line, inter-
sects the capital-constraint line G' H' at J" to show that the (X2 /X1 )
ratio has declined from the slope of OJ to that of OJ".
There are two conditions, supplementing Theorem 9.2, which are
necessary for the existence of the perverse output response to
effective protection. First, condition (9 .31 ), which provides the basis
for CL 2 to move in the 'wrong' direction, must be satisfied. Second,
the intermediate product must be a better substitute of labour than
it is of capital. This is because the change in CL 2 is governed by the
sign not only of(w*-p~!) which is required to be positive, but also
of (w*- r*) which is negative. The latter tends to raise CL 2 whereas
the former tends to lower it. Therefore, if CL 2 is to decline on balance,
the imported intermediate product must be a better substitute of labour
than it is of capital. In the diagram, point J", when compared to J,
shows that not only the relative output of X 2 but also its absolute
output has declined. However, this is by no means necessary. An
interesting property of this model is that both x2 and xl may rise
as a result of the effective protection and yet the (X2 /X 1 ) ratio may
decline. This would, for example, be the case if E" Fwere to intersect
G' H' slightly to the left of J", but to the right of J. This result can be
easily verified by solving (9.22) and (9.23) to obtain the expressions
for individual outputs.
Until now we have considered two polar cases, one where effective
protection involves only the imposition of the tariff on the imports
of the final product, and the other where effective protection is pro-
vided by subsidising the imports of the intermediate input. The
output response is normal in the former case simply because no
change in PQ occurs, whereas with the latter case the output response
may be perverse precisely because of the change in PQ· For the same
reason, if the imported input is used in fixed proportions, the
perverse output response is ruled out.
In the general case, where effective protection is conferred by
means of both the input and the output tariff such that p~, p~ and
(p~- QQ 2 p~) are all positive, the output response may again be
perverse because of the change in PQ, provided (i) the imported

tAn examination of(9.24) confirms the view that (9.31) is the necessary condition
for the validity of this paradoxical result.
228 Studies in the Pure Theory of International Trade

input is used in variable proportions, (ii) p 0, though less than


p!f£JQ 2 , is sufficiently large, and (iii) Po does not lie between r* and
w*, that is condition (9.30) is not satisfied. The latter alone is of
course sufficient to rule out the perverse result. Moreover, the neces-
sary condition for the perverse result when both p! and Po are
positive is not given by condition (9.31), for with Po > 0 and
w* < 0 this condition cannot be fulfilled. It is now the following
condition, namely,
PO > r* > w* (9.32)
that is crucial for the existence of the perverse result. As before,
CK 1 declines and CLl rises and as a result H rises to H' and Fdeclines
to Fin Fig. 9.4, which is drawn on the same principles as Fig. 9.3.
Since bothp0and r* are greater than w*, both (r/w) and (pQfw) and
hence C Ll rise, so that E shifts down to E' in Fig. 9 .4. However, if
Po> r*, (pQ/r) also rises and so may CK 2 ; thus it is now CK 2
(instead of CL 2 ) that may shift in the 'wrong' direction, and if this

H' x,
Figure 9.4
The Theory of Effective Protection 229

shift is sufficiently large, it is possible that the (X 2 /X 1 ) ratio may


actually decline. This is depicted in Fig. 9.4, where G moves down
to G', G' H' intersects E' F' at J', and, as a consequence, the output
ratio (X2 /X 1 ) declines to the slope of OJ'. It is interesting to note
that with this particular type of effective protection the absolute
output of both goods may decline.
When a non-traded intermediate product is introduced, the model
and the explanatory logic become more complicated. If we allow
substitution among all factors of production, we can no longer show
that effective protection provided by means of the output tariff
alone will necessarily lead to a rise in the output of the protected
commodity, as has been demonstrated in the model where the non-
traded intermediate product did not exist. The logic behind this
result is more or less the same as that given above, where it is the
change in PQ• effected by the input tariff, that is responsible for the
perverse output response. If PQ is unchanged, the latter result cannot
occur. In the presence of the domestically produced intermediate
product (M), p't, from (9.28) may be non-zero as a result of a rise in
p 2 irrespective of whether PQ changed or remained constant. Thus
in the model allowing for the existence of traded and non-traded
inputs, p't, plays the same role as Pa with one difference, namely,
the change in PM• unlike PQ• is endogenous, so that even when Pais
zero, as is the case in the presence of the output tariff alone, a change
in PM resulting from the rise in p 2 could give rise to the same possi-
bilities as the change in PQ did in the model without the non-traded
input. In the second industry we now have to deal with gross produc-
tion coefficients in X 2 (R; 2 ). For normal results we now require a rise
in (eL2 + eLMeM 2) and a decline in (eK 2 + eKMeM 2 ) as the (w/r)
ratio declines consequent upon the imposition of the output tariff.
As usual, eKI declines and eLI rises, because PM does not influence
the input choice in the first industry. For the same reason, eLM rises
and e KM declines. Since r* > 0 and w* < 0, and since changes in
both w and r affect the change in PM, it is obvious that

r* > p't, > w*

which implies that substitution in the second industry takes place in


favour of labour and away from capital, so that eL 2 rises and eK2
declines. These changes serve to raise (eL 2+ eLMeM 2 ) and lower
(eK 2 + eKMeM 2 ), as is required for the normal results. However,
230 Studies in the Pure Theory of International Trade

CM 2 may decline or rise because it is influenced by changes in three


factor prices, r, PM and w. Suppose CM 2 actually declines. This par-
ticular change serves to reinforce the decline in the gross capital
coefficient in X 2 , but tends to raise its gross labour coefficient, and if
CM 2 falls sufficiently, the latter may eventually decline. Such shifts in
the production coefficients can be introduced in Fig. 9.3, if we only
replace the net production coefficients in X 2 by the gross production
coefficients. The interested reader will find that for some level of
the decline inCMl the output ratio (X2 / X 1 ) will also decline, thereby
creating the paradox.
Even if CM 2 was to rise the perverse result could still occur,
because this factor tends to raise the gross capital coefficient in X 2 ,
and if the impact is sufficiently pronounced this could actually rise,
which would again run into conflict with the conditions sufficient to
rule out the paradox.
When effective protection is provided by means of output as well
as input tariffs the picture becomes fuzzier, but the explanation
advanced above still applies. Specifically, the presence of non-traded
material inputs may impair or strengthen the results derived in their
absence. However, the source of the paradox always lies in the
presence of the imported input which makes the effective factor
supply variable, irrespective of the existence or the non-existence of
the non-traded material input.t For if there is no imported input in
the model, the output of a final good always responds positively to
the rise in its relative price, as was established in the previous chapter.

t The analysis of this chapter, like most recent general equilibrium studies that have
attempted to evaluate the E.R.P. concept in terms of its predictable effect on resource
allocation, is far from encouraging to the use of effective tariffs. The villain, as
suggested above, is not the non-traded input, not even the concept of the effective
tariff, but the imported input whose availability at constant world prices makes the
effective factor supply variable when it is substitutable with the domestic inputs.
Jones [11], Ramaswami and Srinivasan [14] and Corden [6] have provided some
basis for salvaging the E.R.P. concept by demonstrating the necessity of biased
substitution effects for the existence of the paradoxical resource-allocational implica-
tions. Our analysis, however, destroys even that little hope, for here, even if substitu-
tion effects are unbiased, the relative output of the effectively protected commodity
may still decline. Witness here the fact that in (9.29) even if

so that there are no biased substitution effects, inequality (9.29) could be satisfied to
produce the paradox.
The Theory of Effective Protection 231

This can also be verified by setting Oa 2 to zero, so that (9.24A)


reduces to
lA. I ·IOI<X!- X!) = - (fh + PK)P! < 0
because both PL and PK• which include elements of the non-traded
material input, are positive.
Up to now in our general equilibrium analysis of effective protec-
tion we have used the E.R.P. formula given by (9.1) wherein q 2 is
taken to be zero. If we use the 'unbiased' E.R.P. formula given by
(9.2), the results derived above are unmodified at least qualitatively,
if not quantitatively. For this reason alone, the additional complica-
tions introduced by the use of the unbiased formula are not worthy
of detailed exploration.

9.6 Effective Protection and the Gains from Trade


The perverse resource-allocation implications of effective protection
established in the preceding section have interesting and important
ramifications for gains from trade. As established above, it is not in
general possible to rank the E.R.P. in terms of the relative output of
the protected commodity. In this section we shall show that, precisely
because of its perverse resource-allocational effects, the E.R.P.
cannot be conferred a unique ranking in terms of social welfare as
well, in sharp contrast to the nominal tariff rate which, as we demon-
strated in Chapter 4, is negatively related to community welfare,
barring, of course, the presence of inferior goods.
But first we demonstrate that, for a small country, free trade con-
tinues to be the optimal policy even in a world where intermediate
products are traded internationally. In order to be able to do this we
need an expression for the M RT in our model with imported inputs.
This can be obtained by differentiating
P1X1 +p2X2-paQ = wL+rK
so that, by using (9.3A) and (9.4A),
(PI dX1 +P2dX2-PadQ) + (X1 dp1 + X2dp2)
= Ldw+Kdr+Qdpa
= dw[CL 1X 1+RL 2X 2] +dr[CK 1 X 1 +RK 2X 2] +X2Ca 2dpa
= X 1[CL1dw+ CK 1dr] +X2[RL 2dw+RK 2dr+Ca 2dpa]
= X 1dp 1+X2dp2
232 Studies in the Pure Theory of International Trade

whence
(9.33)
or
dX 1 (p 2 -pQdQ/dX2 )
(9.34)
dX 2 = Pt
which shows that theM RT in the present model equals the ratio between
the marginal value-added in each final commodity. t Coming now to
the demonstration of the optimality of free trade, let us consider the
following system of equations:
U= U(D 1 ,D 2)
E 1 = (P2/P 1 )E2 + (PQIP 1 )Q (under balanced trade)
D 1 = X 1 -£ 1 = X 1 - [(P2/P 1 )E2 +(pQ/pdQ]
and
D2 = X2+E2.
Differentiating the utility function totally with respect to the E.R.P.
(e 2 ), we obtain

dU U1 [dDt+ U2. dD2] = U 1 [dDt +P2(1 +t 2). dD 2]


de 2 de 2 U 1 de 2 de 2 p1 de 2
U1 [dXt _P2. d£2 _PQ. dQ +P2(1 + t2) {dX2 + d£2 } ]
de 2 p 1 de 2 p 1 de 2 p1 de 2 de 2

U1 [{dXt +P2(1 +t2). dX2 _PQ. dQ}+ t 2P2. d£2 ] ·


de2 Pt de2 Pt de2 Pt de2

Now in the presence of the effective tariff, (9.33) can be written as

dX 1 +p 2(1 + t 2). dX2 _PQ0 + tQ). dQ = O.


de2 Pt de2 Pt de2
Using this

(9.35)

t Here RL 2 and RK 2 denote the gross production coefficients. Furthermore,


CQ 2 = Q/X2 , so that Q can be substituted by X 2 CQ 2 •
Since X 1 uses no imported input, the marginal value-added there is the same as its
price.
The Theory of Effective Protection 233

It is clear from (9.35) that if initially there is no tariff, so that t 2 =


tQ = 0, dU/de 2 = 0. In other words, the initial situation offree trade
is the optimal policy.
In order to demonstrate this result geometrically, we follow the
geometrical technique developed recently by Ruffin [15], who
assumes that Ca 2 is constant so that the effective tariff necessarily
raises the relative output of the import-competing final good. If
Ca 2 = Q/X2 is constant, then dQ/dX2 = CQ 2 , so that (9.34) reduces
to
dX1 (P2- PaCa2) = V2
dX2 P1 P1

This means that the slope of the transformation curve HH' at any
point in Fig. 9.5 reflects not (P2/P 1 ) but v 2 /p 1 . Suppose Pis the free
trade production point; then AB, whose slope equals v2 /p 1 , may be
conceived as the budget line facing the producers. The equilibrium
on the consumption side, however, is reached when the MRS =
(p 2 /p 1) > (v 2 /p 1). Since X 1 does not use the imported input, the

Figure 9.5
234 Studies in the Pure Theory of International Trade

value of production and consumption is the same in terms of


XI ( = OA), but not in terms of x2.t Hence the budget line facing
the consumers, though steeper than AB, must start from point A.
In Fig. 9.5, AE is such a budget line, consumption is at C, and the
welfare level under free trade is given by V 1 , which is the maximum
level of welfare; PG of X 1 is exported and GC of X 2 is imported.
The imports of the intermediate good (Q) equal FP in terms of the
second good. This can be established by noting that RP = AR(p 1/v 2 )
and RF = AR(ptfp 2), so that FP = RP-RF = ARp 1[(p 2 -v 2)/
p 2 v2 ] = RP(p 2 - v2 )/p 2 . Since RP is the output of X 2 at the produc-
tion point P,
FP = PQQIP2·
To show that V 1 is the maximum level of welfare, consider any other
policy such as the imposition of an input tariff on Q1 which lowers
the output of X 2 and moves the production point, say to J. Since
p 2 and PQ are given, the international price line facing the producers
will be parallel to AB but will pass through J, and, as a result, the
budget lines facing both producers and consumers will lie below
AB and AE respectively (see the dotted lines in Fig. 9.5); social
welfare will also then lie below V 1 • Thus we see that free trade is the
optimal policy even when imported inputs exist in the modeJ.t
The implications of the E.R.P. for welfare, however, are uncertain
if CQ 2 is not given. This can be seen by expanding (9.35) to get

dV
de 2
= VI [tQ PQ dQ + t2P2 {dD2-
p 1 de 2 p 1 de 2 de 2
dX2}]
or

(9.36)

If the imported input is used in fixed proportions, dQ/dX 2 = CQ 2 ;


therefore (9.36) reduces to

I dV p2 [ dD2 dX2] (9.37)


- - d = - t2-d -e2(1-{}Q2)-d .
VI e2 P1 e2 e2

t Since X 2 uses the imported input for which the foreign country must be paid, the
value of consumption in terms of X2 must be less than the value of its production.
t This can be seen by drawing a non-intersecting social indifference curve (not
drawn) touching the dotted lineAE.
The Theory of Effective Protection 235

Since D 2 = D 2 (ph, Y) where Ph = p(1 + t 2 ), dD 2 /de 2 = (oD 2 /oph)


(dphfde 2 )+(mhfph)(dYfde 2 ). Substituting this in (9.37), we obtain

If the effective tariff is increased by a combination of a rise in t 2


and/or ta, then dPhfde 2 > 0, and with Ca 2 constant, dX2 /de 2 > 0.
Since oD 2 foph < 0, all this implies that dY/de 2 < 0, given that
0 < mh < 1. In other words, if the effective tariff is raised by in-
creasing t 2 , then social welfare declines provided Ca 2 is constant, no
matter what happens to ta. However, if e 2 is raised by lowering both
t 2 and ta, then dYfde 2 possesses an ambiguous sign, because then
dphfde 2 < 0, although dX 2 fde 2 is still positive. Similarly, if Ca 2 is
not given, the sign of dYfde 2 becomes indeterminate, because now
the sign of dX2 /de 2 becomes uncertain.
The analysis of gains from trade in the presence of imported inputs
gives rise to some interesting possibilities which do not arise when
such inputs are ignored. The credit for opening new directions in
this connection goes to Ruffin [15, 16]. He introduces the concept
of the optimum effective tariff which maximises the level of welfare
in the presence of the nominal tariffs on the inputs and outputs. The
principal result derived by Ruffin is that when the imported input is
used only by the importable good, the optimum effective tariff is zero,
where the optimal effective tariff is one that maximises the second-best
level of welfare. Since free trade continues to be the optimal policy
for a small country even when imported inputs are taken into
account, any level of effective tariff, positive or zero or even nega-
tive, is bound to cause a decline in social welfare. The problem is to
devise that kind of tariff structure which results in the minimum
decline in welfare, or, what is the same thing, in the second-best
level of welfare.
We shall now derive Ruffin's theorem and then go on to show that
his result depends on his special assumption that the initial situation
is one involving the tariff on the final good. If, in the initial situation,
the tariff is on the imported input, the optimum effective tariff turns
out to be negative. This latter result has been recently established by
Casas [4].
Let us assume that the home and the foreign prices of all import-
able goods differ because of the home tariffs t 2 and ta. Under these
236 Studies in the Pure Theory of International Trade

conditions, the first differential of the utility function can be shown


to equal
dU tQPQ
- = dY = 12pdEi+- dQ.
ul P1
This also follows from (9.35). Now from Q = CQ 2 X 2 , we have
dQ = X 2dCQ 2 +CQ 2dX 2. (9.38)
Using this and the fact that £ 2 = D 2 -X2 , we obtain
1QPQ
dY = 12pdD 2 -pdX2(1 2 -OQ 21Q)+-- X 2dCQ 2 (9.39)
Pt
where eQ2• as before, is the share of the imported input in the total
cost of the second good valued at world prices. Now
D2 = D 2[p(I + 12), Y]
and (9.40)

Suppose we start with an initial situation of the tariff on the final


good and keep it fixed at that level. Then
iJD2 mh
dD 2 = - · dY = - dY
iJY Ph
and

where d[pQ(I + 1Q)/p 1 ]/d1Q = (pQ/p 1 ). Substituting these, (9.39)


becomes

or
dY(l- T2mh) = - pdX2 (1- OQ 2 )e 2 + 1QpQX2 dCQ 2 /p 1

where T2 = (1 2 /1 +1 2 ) < I. It is clear that dY = 0 only if

(where 1Q = p~). (9.41)


The Theory of Effective Protection 237

In other words, the second-best level of welfare is maximised when


the effective tariff equals the term contained in the right-hand side
of (9.41), which in view of the ambiguous sign of p~/X! may be
positive or negative. If the imported input is used in fixed propor-
tions, q 2 = 0, which means that the optimal effective tariff is zero.
This is Ruffin's result. What this really means is that, in the presence
of the initial output tariff, an input tariff should be introduced in
such a way that (9.41) is satisfied.
However, suppose the initial tariff is on the imported input and is
kept unchanged at that level. Then by varying the tariff on the final
good we can maximise the tariff-distorted level of welfare. Then,
from (9.40),
_ poD 2 oD 2 y poX2
dD 2 - op(I + tz) dt 2 + oY d and dX 2 = op(l + tz) dt 2 > 0

because dp(I +t 2 )/dt 2 = p. Substituting these in (9.39), we get


z oDz a
dY(l-T2 mh) = t 2 p 0 ( )dt 2 -pdX2 (!-uQ 2 )e 2
p l+t 2
under the assumption that q 2 = 0. Maximising welfare in the
presence of the initial input tariff then requires that
t2p iJD 2 dt 2
ez = . -
(I-OQ 2 ) op(I+t 2 )dX2

Since dX2 /dt 2 > 0 and oD 2 jop(l + t 2 ) < 0, the optimal effective
tariff is negative.
The economic explanation of these results is quite simple. As
shown in Chapter 4, the introduction of the output tariff gives rise
to a distortion on both the production side and the consumption side
and thus to the production and the consumtpion loss. However,
the production loss can be mitigated by introducing an input tariff
which tends to lower the output of the importable goods and take
it closer to the free trade output level. The production loss is com-
pletely eliminated if the tariff on the input is equated to t 2 /0Q 2 , so
that e2 = 0, provided CQ 2 is constant. The reason lies in the fact that
when the imported input is used in fixed proportions, the E.R.P. and
the output of X 2 are positively related, so that when e 2 = 0, the
output of X 2 is the same as what it would have been under free trade,
with the result that the production loss resulting from the tariffs
238 Studies in the Pure Theory of International Trade

disappears. Thus Ruffin's optimal effective tariff is one that leaves


the free trade output levels unchanged.
However, this logic does not apply to the case where initially the
tariff has been imposed on the input, which by lowering the output of
X 2 causes the production loss but not the consumption loss. It is
worth noting here that the production loss occurs whenever the
actual output pattern differs from the one prevailing under free
trade. A tariff on the final good promotes the expansion in its output
but also introduces the consumption loss. Therefore, as the produc-
tion loss is reduced, the consumption loss is introduced, and if t 2 is
equated to ()Q 2 tQ so that e 2 = 0, the production loss is eliminated,
but welfare does not rise to its maximum level because of the intro-
duction of the consumption loss. Hence the optimal effective tariff is
one which, apart from minimising the production loss, keeps the
consumption loss at its minimum. This inevitably requires that e2 be
negative.

REFERENCES
[I] Balas sa, B., 'Tariff Protection in Industrial Countries: An Evaluation', Journal
of Political Economy, LXXIII (Dec 1965) 572-94.
[2] Barber, C. L., 'Canadian Tariff Policy', Canadian Journal of Economics and
Political Science, XXI (Nov 1955) 513-30.
[3] Basevi, G., 'The U.S. Tariff Structure: Estimate of Effective Rates of Protection
of U.S. Industries and Industrial Labour', Review of Economics and Statistics,
XLVIII (May 1966) 147-60.
[4] Casas, F. R., 'Optimal Effective Protection in General Equilibrium', American
Economic Review, LXIII (Sep 1973).
[5] Corden, W. M., 'The Structure of a Tariff System and the Effective Protective
Rate', Journal of Political Economy, LXXIV (June 1966) 221-37.
[6] --,'The Substitution Problem in the Theory of Effective Protection', Journal
of International Economics, I (Feb 1971) 37-58.
[7] Ellsworth, P. T., 'Import Substitution in Pakistan: Some Comments', Pakistan
Development Review, VI (autumn 1966) 395-407.
[8] Grube!, H. G., and Lloyd, P. J. 'Factor Substitution and Effective Tariff Rates',
Review of Economic Studies, XXXVIII (Jan 1971) 95-104.
[9] Guisinger, S. E., 'Negative Value Added and the Theory of Effective Protection',
Quarterly Journal of Economics, LXXXIII (Aug 1969) 415-33.
[10] Johnson, H. G., 'The Theory of Tariff Structure with Special Reference to World
Trade and Development', in Trade and Development (Geneva: Institut Univer-
sitaire des Hautes Etudes Internationales, 1965).
[II] Jones, R. W., 'Effective Protection and Substitution', Journal of International
Economics, 1 (Feb 1971) 59-82.
[12] Leith, J. C., 'Substitution and Supply Elasticities in Calculating the Effective
Protection Rate', Quarterly Journal of Economics, LXXXII (Nov 1968) 588-601.
[13] --,'The Effects of Tariffs on Production, Consumption and Trade: A Revised
Analysis', American Economic Review, LXI (Mar 1971) 74-81.
The Theory of Effective Protection 239
(14] Ramaswami, V. K., and Srinivasan, T. N., 'Tariff Structure and Resource Allo-
cation in the Presence of Factor Substitution', in J. N. Bhagwati eta/. (eds.),
Trade, Balance of Payments, and Growth (Amsterdam: North-Holland, 1971),
[15] Ruffin, R. J., 'Tariffs, Intermediate Goods, and Domestic Protection', American
Economic Review, LIX (June 1969) 261-9.
[16] - - , 'The Welfare Implications of Effective Protection', in H. G. Grube! and
H. G. Johnson (eds.), Effective Tariff Protection (Geneva: General Agreement
on Tariffs and Trade and Graduate Institute of International Studies, 1971)
chap. 5.
[17] Travis, W. P., 'The Effective Rate of Protection and the Question of Labour
Protection in the United States', Journal of Political Economy, LXXVI (May/June
1968) 443-81.

SUPPLEMENTARY READINGS
(18] Anderson, J., and Naya, S., 'Substitution and Two Concepts of Effective Pro-
tection', American Economic Review, LIX (June 1969) 607-11.
[19] Corden, W. M., The Theory of Protection (Oxford: Clarendon Press, 1971).
[20] Ethier, W., 'Input Substitution and the Concept of the Effective Rate of Pro-
tection', Journal of Political Economy, LXXX (Jan/Feb 1972) 34-47.
10 Factor Market Imperfections

There is a general agreement among economists that perfect com-


petition is a 'myth'. Yet the bulk of the analysis in trade theory-
and economic theory in general - has been carried out under this
'mythical' assumption. Perhaps the best case that can be made for
a perfect market is that it provides an 'ideal' yardstick to evaluate
the efficiency of existing systems which generally fail to satisfy
the stringent requirements of competitive conditions. This sug-
gests that the assumption of perfect markets, which up to the
previous chapter contributed much to the simplicity of our analysis,
stands in need of replacement. Relaxing this assumption in turn
permits a wide variety of production systems such as monopoly,
oligopoly, imperfect factor markets, etc., all of which, individually
or collectively, may characterise the production side of a country.
In order to bring the analysis down to manageable proportions, some
selectivity in the choice of alternative systems is unavoidable. In
this chapter we relax the assumption of perfect factor markets, while
product markets continue to be perfect. The discussion of the
product market distortions is postponed until the next chapter.
The kind of factor market imperfections analysed here concerns
the existence of a stable factor-price differential between the two
industries. Such distortions may be the result of taxation, particu-
larly of corporate income taxation which boosts up the cost of capital
to the corporate sector over and above its cost to the non-corporate
sector (Harberger [7]); they also may arise from the existence of
the wage differential between the industrial and the subsistence
sector of an underdeveloped economy (Lewis [14]), or from the
introduction of unionisation which raises union wages over non-
union wages. What is particularly noteworthy is that this type of
factor-price differential is compatible with perfect factor mobility
which we continue to assume for the better part of this chapter. In
order to simplify our analysis we assume that capital continues to
Factor Market Imperfections 241

earn the same reward in both industries, but that the wage rates
differ. The different rates of return in the two industries can be
analysed without any difficulty. The crucial factor that prompts this
analysis concerns the differential between factor-price ratios, irres-
pective of whether or not both factor prices differ in the two indus-
tries. The rest of the assumptions are the same as those made at the
beginning of Chapter 2. Thus product markets are still perfect, pro-
duction functions exhibit constant returns to scale and diminishing
returns to factor proportions, factors are perfectly mobile and factor
prices perfectly flexible, and so on.

10.1 The Model with the Wage Differential


The model is now described by the following equations:
CLlX1+CL2X2 = L (1 0.1)
cKlxl + CKzXz = K ( 10.2)
CL1w1+CK1r =Pt (10.3)
CL2w 2 + CK 2 r = p2 (10.4)
and
(!X =F 1) (10.5)
where wi is the wage rate in the jth sector and <X is a parameter
measuring the extent of the wage differential. By differentiating these
equations totally, we obtain
A.LtXi+A.LzX! = L*-(A.LtC!t +A.LzC!z) (10.6)
A.KtXi +A.KzX! = K*-(A.Ktc~~ +A.KzC~z) (10.7)
OL1 w'f + OK1r* =Pi (10.8)
OLz w'f + OKzr* = P!- OLz<X* (10.9)
and
<X*+wi = w! (10.10)

where A.Li = CLiX)L and OLi = CLiwi/Pi• and so on. As usual, in


deriving (10.8) and (10.9) we have made use of the minimum unit
cost conditions
(10.11)
242 Studies in the Pure Theory of International Trade

and
(10.12)
The next step involves the introduction of the elasticity of sub-
stitution in each sector, so that

(). =
q.-cr.
J J (} = 1,2).
1 wt-r*

These, together with (10.10)-(10.12), yield

C! 1 = -OK 1a 1(w!-r*)
C11 = eL1()1(wf-r*)
C!2 = - [OK2u2(w!- r*) + eK2()2~·]
and

Substituting these in (10.6) and (10.7), we obtain

ALtXf+,I.L2X! = L*+PL(wf-r*)+,I.L28K2u2~* (10.13)

AKtXi +AK 2 X! = K*- PK(w! -r*)-AK 2 8L2a 2 ~* (10.14)


where

and

It must be clear by now that we are solving the system of equations


presented above in terms of the wage/rental ratio in the first industry.
If~ is unchanged, the wage/rental ratio in the the second industry is
immediately determined. If~ is not constant, then the wage/rental
ratio in the second industry can be determined with the help of
(10.10). Thus once we solve for (wtfr), we can obtain values for the
rest of the variables.
The structural relations presented above can be used to answer a
number of interesting problems that were explored in the absence
of the wage differential; but first a few remarks on the factor
intensities in the present model are in order.
Factor Market Imperfections 243
10.2 Factor Intensities and the Wage Differential
As before, let A. be the matrix of coefficients on the left-hand side
of (10.13) and (10.14), and let () be the corresponding matrix in
(10.8) and (10.9). Obviously

(10.15)
and
I() I = -w -r · CLl CL2(k
1
2- rxk 1 ) (10.16)
P1Pz
where ki = Ki/ Li is the capital/labour ratio in the jth sector. At the
same time
(10.17)
and
(10.18)
The main feature of the undistorted two-sector model is that IA.I
and 101 have the same sign, for then rx = I. Now the sign of IA.I
indicates the ranking of industries in terms of the factor intensities
in the 'physical' sense, for here the comparison between the employ-
ment of the physical units of the factors in the two sectors is
involved; the sign of I() I on the other hand reflects the ranking of
industries in terms of factor intensities in the 'value' sense, for here,
as is evident from (I 0.18), we compare the relative share of any
factor in the two sectors in determining whether or not an industry
is capital- or labour-intensive relative to the other. This may also
be seen directly by using (10.5), so that from (10.16)

kz-rxkl = Wz [ rKz - rK! ]·


r w 2 L 2 w1 L 1
In other words, the sign of 101 indicates that, instead of comparing
the ratio between physical units of capital and labour employed in
the two sectors, which comparison is reflected by the sign of lA. I, we
compare the ratio between the total value of capital and labour em-
ployed in the two industries. In the absence of the wage differential,
IA.I and 101 have the same signs. With X 2 capital-intensive relative
to X~> for example, this implies from (10.17) and (10.18) that not
only does the proportion of labour exceed the proportion of capital
used in the first industry, but also that the relative share of labour
244 Studies in the Pure Theory of International Trade

in the first industry exceeds that in the second. This, however, may
no longer be true in the distorted model. For example, if tx > 1, the
sign of (k 2 -txk 1 ) and hence of 101 may be opposite to the sign of
(k 2 -k 1) and IA.I when X 2 is capital-intensive relative to X 1 in the
physical sense. From (10.5), tx > 1 implies that w 1 < w2 , that is,
the differential is paid by the second commodity. With labour being
paid a premium in the second industry, it is not difficult to visualise
a situation where the relative share of labour in the second industry
may exceed that in the first, so that 101 < 0, even though X 2 employs
a greater proportion of capital than labour, i.e. A.K 2 > A.L 2 •
What if tx < 1, that is, w 1 > w2 • Here lA. I and 101 must possess the
same sign when k 2 > k 1 . The premium to labour is now paid by the
first industry which is also labour-intensive in the physical sense.
From this we conclude that if an industry pays the differential on
its intensive factor (in the physical sense), IA.I and 101 have the same
sign, that is, the factor intensities in the physical sense cannot get
reversed in the value sense.
The picture is further blurred by the fact that, as first demonstrated
by Johnson [I OJ, the physical factor-intensity relationships may also
change sign in the presence of the differential, even though in its
absence they are non-reversible. This point can be illustrated with
the help of the box diagram presented in Fig. 10.1, where X 1 in the
absence of distortions is shown to be labour-intensive relative to x2

Figure 10.1
Factor Market Imperfections 245

in the physical sense. In the undistorted case, the contract curve,


0 1 A0 2 , is defined by the tangency of the isoquants for X 1 and X 2 ,
where the slopes of the two isoquants and hence the wage/rental
ratios are the same. In the presence of the factor market distortion,
the contract curve will pass through points where the isoquants
representing X 1 and X 2 intersect. In Fig. 10.1 the distorted contract
curves are given by the dotted curves. If IX > 1 so that (wdr) <
(w 2 jr), the distorted efficiency locus lies below 0 1 A0 2 and passes
through a point such as B, for only at a point like Bare the slope of
the x1 isoquant x1x1, and hence (wdr), less than the slope of the
X 2 isoquant X'2 X'2 and hence (w 2 jr). Here the presence of the dif-
ferential causes an increase in the spread between k 2 and k 1 . In the
opposite case where IX < 1, the efficiency locus will lie above 0 1 A0 2 •
Three possibilities emerge in this case. The distorted efficiency locus
may pass through a point such as C, be identical with the diagonal
0 1 0 2 , or pass through a point such as E which lies not only above
0 1 A0 2 but also above the diagonal. In the first case IX < 1 but k 2
is still greater than k 1 , although the spread between the two factor
intensities has declined; in the second case, for some value of
IX = IX 0 < 1, the spread between k 2 and k 1 falls to zero and the
contract curve becomes identical with the diagonal; finally, in the
third case IX < IX 0 < 1, the relationship between k 2 and k 1 is re-
versed. Clearly, this result is attributable to the adverse substitution
effect that takes place in the first industry against labour when IX < 1
and w 1 > w2 • In the absence of the differential k 1 < k 2 • However,
when the first industry pays the premium to labour, substitution
takes place against labour in the first industry and as a result k 1
begins to rise. Therefore, if the premium paid by labour is sufficiently
high, k 1 may rise above k 2 , in which case the factor intensities in
the physical sense will get reversed. From the discussion so far in this
section, we may derive the following theorem:
Theorem 10 .1. The necessary condition for the factor intensities to
get reversed in the physical sense is that an industry pays the wage
differential on its intensive factor. On the contrary, the necessary
condition for the reversal of factor intensities in the value sense is
that an industry pays the differential on its non-intensive factor (in
the physical sense).
It follows from this theorem that factor intensities may get re-
versed in both the physical and the value sense, in which case jA.j and
246 Studies in the Pure Theory of International Trade

101 will again have the same sign. In the diagram, for example, k 2
becomes less than k 1 for IX < a: 0 ; then k 2 may also be less than a:k 1 •
Thus factor intensities may get reversed in both the physical and
the value sense.
The distinction introduced above between factor intensities in the
value and the physical sense is very important. It is clear from (10.6)
and (10.7) that the relationship among the physical variables of the
system, such as commodity outputs, factor endowments, etc., is
influenced by the factor intensities in the physical sense, i.e. by the
sign of IA.I, whereas the relationship among the financial variables,
like factor and commodity prices, is regulated by the factor inten-
sities in the value sense, i.e. by the sign of 101. If IA.I and 101 have
opposite signs, the usual positive link between financial and physical
variables established in the previous chapters will be reversed.

10.3 Factor and Commodity Prices


Let us first consider the relationship between factor and commodity
prices at a constant wage differential, so that a:* = 0. From (10.8)
and (10.9), we can obtain
* * _ 8Kj(P!-p!) (10.19)
wl -pi - IOI

(10.20)

and
• •
w* -r* = P1 -p2 (10.21)
1 IOI
remembering that, with a:* = 0, wt = w!.
As expected, 101 determines the relationship between commodity
and factor prices. If lA. Iand 101 are both positive, then it is clear that
a rise in the relative price of the first commodity raises the relative
and the real reward of its intensive factor, labour, in both industries
(because wt = w!) and lowers the relative and the real reward of the
other factor, capital. The Stolper-Samuelson theorem continues to
hold in the presence of the given wage differential. However, if
101 < 0 when IA.I > 0, the Stolper-Samuelson theorem is no longer
valid in the presence of the factor market distortion. The reason
once again lies in the fact that the sign of lA. I has nothing to do with
Factor Market Imperfections 247

the interrelations among financial variables. Furthermore, 101 < 0


means that the first industry is capital-intensive in the value sense.
The result may be illustrated in another way. Suppose the wage/
rental ratio rises in the economy; then the relative price of that
commodity will rise where the relative share of labour is higher; in
other words, for w!- r* > 0, (p!- p 2 *) ~ 0 if eLl ~ OL 2 , so that
101 ~ 0 (see (10.21)).
10.4 The Slope of the Transformation Curve
In the undistorted two-sector model (with or without intermediate
goods) we have become accustomed to two facts: (i) theMRTreftects
the commodity-price ratio, and (ii) the transformation curve is
concave towards the origin, a result that derives directly from the
positive price-output response. In the distorted model, however,
we are in for some jolts. It is no longer true that (i) M RT equals the
commodity-price ratio, (ii) the transformation curve is necessarily
concave towards the origin, (iii) the price-output response is always
positive, and (iv) the shape of the transformation curve can be
inferred from the price-output response.
All these results can be derived by solving the structural relations
presented above and by setting ex* to zero.
To derive the slope of the transformation curve, we follow a
circuitous route which contributes to the relative simplicity of the
calculations. Instead of using the procedure employed in the last
three chapters, we utilise the one used in Chapter 4. From the pro-
duction functions for xl and x2 we can get

v 1dL 1 +z 1dK 1
v 2 dL 2 +z 2 dK 2

and since dL 1 = -dL 2 and dK 1 = -dK2 from the maintenance of


full employment, and since p 1 z 1 = p 2 z 2 and, in the presence of the
differential, cxp 1 v1 = p 2 v2 , we can write

p[(v 2 dL 2 /cx) + z 2 dL 2 ]
-f3p (10.22)
v 2 dL 2 +z 2 dL 2
where
248 Studies in the Pure Theory of International Trade

x,

Figure 10.2a

0 H'
Figure 10.2b
Factor Market Imperfections 249

It is clear that p ~ 1 according as oc ~ 1. A glance at (10.22) suggests


that theMRT is no longer equal to - p, unless, of course, P = 1, so
that there is no wage differential. Two effects of the wage differential
are depicted in Fig. 10.2, where the solid curve HH' is the trans-
formation curve in the undistorted model. The presence of the
differential introduces inefficiency in the system, and as a result the
transformation curve shrinks towards the origin to the dotted curve
HPH' except at the points of complete specialisation, Hand H',
where the wage differential becomes irrelevant. In Fig. 10.2A the
differential is shown to be against the first industry and the price
line AB cuts the distorted transformation curve from below to show
that the M RT at point P exceeds the relative price of the second
commodity. The converse is true in Fig. 10.2B, where the differential
is paid by the second industry, the price line cuts the distorted trans-
formation curve from above and the M RT at P falls short of the
price ratio. Let us defer for the time being our remarks concerning
the shape of the distorted curve.

10.5 The Price-Output Response


Let us now consider the effect of a rise in the relative price of the
first commodity on the two outputs at constant factor endowments.
From (10.13) and (10.14) and (10.21), we obtain
n _ ().K2pL +).L2pK)(P1 -pt) (10.23)
1 - 1).1·191
and

(10.24)

Evidently, a rise in the relative price of the first commodity raises


the output of X 1 and lowers that of X 2 only if 1;,1 and 101 possess
identical signs. The rationale behind this result is obvious. A rise in
the relative price of the first commodity lowers the wage/rental ratio
if the first industry pays a lower share to labour than the second
industry (so that 101 < 0). The decline in the wage/rental ratio in
turn tends to lower the physical capital/labour ratio in both indus-
tries and thus create an excess demand for labour and an excess
supply of capital. In order to restore the full-employment equi-
librium, the output of the industry that is capital-intensive in the
250 Studies in the Pure Theory of International Trade

physical sense rises and that of the other industry falls. With
k 2 > k 1 , and jA.j > 0, the rise in the output occurs in the second
commodity at the expense of the output of the first, even though to
begin with it was the relative price of the first commodity that in-
creased. In other words, if jA.j and jej have dissimilar signs, the
price-output response is negative or 'abnormal', and the long-run
supply curves are negatively sloped.

10.6 The Shape of the Transformation Curve


We are now analytically equipped to deal with the shape of the
transformation curve which has been the subject of analysis in recent
articles by Bhagwati and Srinivasan [5], Herberg and Kemp [8]
and Jones [13]. The algebra in this case is highly complicated, so we
shall rely mainly on the verbal discussion. We have demonstrated
above that, in the presence of the given wage differential, the price-
output response may be negative. If the M RT equalled the com-
modity-price ratio, this in itself would be sufficient to cause the
distorted transformation curve to become globally convex towards
the origin. The same would be true if {3, the wedge between - p and
the MRT, were constant. This becomes evident when (10.22) is
differentiated with respect to x2 to obtain

(10.25)

Thus if df3/dX2 = 0, the shape of the distorted transformation curve


can be inferred directly from the sign of dp/dX2 . In general, df3/dX 2
may be positive or negative even if IX is constant. This factor gives
rise to a rich variety of possibilities. First, (d 2 XtfdXD may be
negative, whatever the sign of (dp/dX2 ), in which case the distorted
transformation curve will be concave to the origin as in Fig. 10.2A.
Second, (d 2 X 1/dXD may be positive even if dp/dX 2 > 0, provided
df3/dX 2 is sufficiently negative, that is, the distorted transformation
may be locally (or globally) convex towards the origin as in Fig. 10.2B
even if the price-output response is positive. The converse is also
true. The distorted transformation curve may be concave to the
origin even if the price-output response is negative.
The crux of this discussion is that, in the presence of the given wage
differential, the transformation curve may have any shape irrespective
of the nature of the price-output response.
Factor Market Imperfections 251

10.7 The Heckscher-Ohlin Theorem


In Chapter 3 we derived sufficient conditions for the validity of the
Heckscher-Ohlin theorem, which asserts that a country exports the
commodity which is intensive in the use of its relatively abundant
factor. Two definitions of relative factor abundance were utilised,
namely the price and the physical definition. We shall now show that
in terms of the price definition the Heckscher-Ohlin theorem may
not be valid in the presence of distorted factor markets even if the
degree of distortion, ex, is the same in the two countries.
The demonstration of this point in terms of the price definition
requires the use of the unpredictable relationship that exists between
factor and commodity prices if the factor intensities get reversed in
the value sense,t even if in the physical sense, as depicted in Fig.
10.3, they are non-reversible. The wage/rental ratio in the first in-
dustry, w 1 , is related to the capital/labour ratio in each industry as
well as to the commodity-price ratio, p. In the undistorted case,
k 2 > k 1 at all w 1 ; in the presence of the differential, however,

t It is possible to derive precise conditions for the reversal of factor intensities in


the value sense for a given a. Since in Fig. 10.3 k 2 > k 1 , the necessary condition for
the factor intensities to get reversed in the value sense is that

k! < k!.

From the expressions concerning C;j,

k! = Ch -q, = u 1(w!-r*)
and

Therefore, k! < k! for a given a implies that

If the elasticities of substitution are the same in the two sectors, factor intensities
can never get reversed; if they are constant, there can at most be one reversal; if they
are variable, there may be more than one reversal points like R; finally, the necessary
condition for this reversal is that u 2 < u 1 . On the other hand, the precise con-
ditions for the reversal with a given w 1 can be obtained by differentiating k 2 jak 1 to
obtain
k!-a*-k! = a*(u 2 -l).

Thus fork 2 /ak 1 to decline, it is necessary that u 2 + I. For further discussion of these
points, see Herberg et a/. [9].
252 Studies in the Pure Theory of International Trade

Figure 10.3

k 2 < ct.k 1 (ct. > 1) after the factor-intensity reversal point R is


reached.t The relationship between w 1 and pis described by ATB,
which appropriately undergoes a change in slope at point T corres-
ponding to the reversal point R. It is easy to realise now that if the
autarky w 1 in each country lies on the different side of R', the pattern
of trade may violate the Heckscher-Ohlin dictum.t It is worth point-
ing out here that the factor-price equalisation will also be disrupted
when the unique relationship between commodity and factor prices
hreaks down.
If the home country is capital-abundant relative to the foreign
country, then from the factor-price definition of factor abundance,

t What this really means is that the sign of jej is reversed after point Rand so is
the relation between w 1 and pat a given :x.
! It is worth noting that the discussion of the Heckscher-Ohlin theorem illustrated
b) Fig. 10.3 is identical with the corresponding discussion in Chapter 3 in terms of
Fig. 3.6. The only difference is that the factor-intensity reversal in the earlier chapter
was caused by the multiple intersections of the unit isoquants of the two com-
modities. whereas now it arises from the existence of the differential.
Factor Market Imperfections 253

If in Fig. 10.3 each country's w 1 lies on the same side of R', as is the
case given by the pair Ow 1h and Ow 11 ,
Ph< PJ
so that the home country will export the capital-intensive com-
modity X 2 and import the labour-intensive commodity X 1 • The
Heckscher-Ohlin theorem continues to hold. However, if each
country's w 1 lies on different sides of R', as is described by the pair
Ow'1 h and Ow 11 ,

p~ > PJ
so that the home country will export X 1 and import X 2 , thereby
contradicting the Heckscher-Ohlin theorem. But this result is by no
means necessary even if each country's w 1 lies on different sides of
R'. This can be visualised if the home country's w 1 was to be slightly
below Ow'lh but above R'.
Coming now to the physical definition of'relativefactor abundance,
the Heckscher-Ohlin theorem cannot be disproved if IX is the same in
the two countries. This is because the Rybczynski theorem, as will be
shown later, continues to hold in spite of the differential. Whether
or not the factor intensities get reversed in the value sense makes
little difference to the discussion of the Heckscher-Ohlin theorem in
terms of the physical definition, for, as stated before, the definition
of factor intensities in terms of distributive shares does not control
the relationship between commodity outputs and factor endow-
ments. Of course, if rx is not the same in the two countries, or if one
country has no distortions in factor markets, then the factor inten-
sities in the value sense will also play a role in determining the vali-
dity of the theorem in question. As we show in the subsequent dis-
cussion, the level of output in each commodity at given commodity
prices becomes a function not only of the country's factor endow-
ments but also of IX, and the effect of the latter on the output levels
is determined by the signs of both I)-I and 101. This is how the factor
intensities in the value sense participate in determining whether the
theorem under question continues to be valid in the presence of the
differential. If IX is the same in the two countries, the effect of the
distortion on the two outputs is also the same, which means that
the sign of 101 is of no consequence to the validity of the theorem.
Thus we conclude that the Heckscher-Ohlin theorem continues
to be valid in terms of the physical definition of relative factor
254 Studies in the Pure Theory of International Trade

abundance if IX is the same internationally. However, if IX is not the


same then the theorem may not hold, simply because a relatively
capital-abundant country may then be producing relatively more of
the labour-intensive product at the given commodity prices than the
labour-abundant country. The discussion in this section will become
transparent when we explore the implications of a change in IX on
the output levels at constant commodity prices.

10.8 The Implications of the Wage Differential


Up to now we have assumed that the wage differential is constant.
In this section we examine the consequences of a change in the wage
differential for production, distribution and the terms of trade.
Solving for wt and r* from,(l0.8)-(10.10), we get
* * OKiPi-p!)+8KlOL21X*
wl -pi = IBI (10.26)

* * OKiPi- p!) + (}K2 OLliX*


Wz -pi = IOI (10.27)

() Li(P! - P!) + () Ll () LziX*


r*-pt = (10.28)
IOl
* * (pj- P!) + OLziX*
wl -r = IBI (10.29)

Substituting (10.29) in the expressions for C~, we get

k* _ C* -C* _ a 1 [(pj-p!)+8L21X*]
1 - Kl Ll - IOI (10.30)

and

k * _ C* -C* _ Uz[(p'j-p!)+OL11X*]
2 - K2 L2 - IBI (10.31)

Next, we solve for Xj and X! from (10.13) and (10.14) to obtain

(10.32)

and
()'Llf3K+ )..Klf3L)(p'j- p!) + BIX*
X!= (10.33)
IA.I·IBI
Factor Market Imperfections 255

where

and
B = eLlu2(A.xlA.L2eK2 +A.LlA.K2eL2)+0L2A.LlA.KlO"l > 0.
The conclusions that emerge from (10.26)-(10.33) are categorical
for a small country facing given terms of trade, so that (pT- p!) = 0.
The implications of a rise in tx for real factor rewards, capital/labour
ratios and the two outputs at constant commodity prices are the
same as those of a rise in the relative price of the first commodity
at the constant wage differential. The reasoning is straightforward.
A rise in pifp 2 alone directly serves to benefit the first industry
relative to the second. A rise in tx alone tends to move the differential
against the second industry and thus indirectly benefits the first. In
both cases the relative gain accrues to the first industry; hence a
rise in tx or pifp 2 alone generates similar repercussions in the
economy.
In our discussion in section 10.2 on factor intensities, we showed
that the presence of the differential may cause a reversal in the factor
intensities in both the physical as well as the value sense. Following
Magee [15], we are now in a position to show that factor intensities
in the physical sense can never get reversed if either tx or p( = p 2 /p 1 )
alone gets altered. Consider Fig. 10.4, which retains some features
of Fig. 10.1. Let us begin with the case where tx = 1 and the efficiency
locus is given by 0 1 A0 2 , A being the actual production point. A
decline in tx serves to move the efficiency locus towards the diagonal
and, for some tx < tx 0 < 1, moves it to the opposite side (see Fig.
10.1). However, from (10.30) and (10.31), a decline in tx, with 101 > 0,
leads to a decline in the capital/labour ratio in both industries if the
commodity prices are constant. All this implies that a fall in tx will
move the production point to somewhere above 0 1 A0 2 , below 0 1 A
(so that k 1 declines) and above 0 2 A (so that k 2 declines). On the
contrary, the rise in tx will move the production point to somewhere
below the undistorted locus 0 1 A0 2 , below 0 2 A and above 0 1 A.
These considerations suggest that the distorted efficiency locus is
bounded from above by the two rays 0 1A and 0 2 A when com-
modity prices are unaltered. The dashed curve 0 1 CAB0 2 in Fig.
10.4 is one such curve where the new production point is given by
either B or C, depending on whether tx* ~ 0. It is evident from the
256 Studies in the Pure Theory of International Trade

Figure 10.4

diagram that the distorted contract curve cannot cross the diagonal
if the terms of trade are constant, whatever the level of a below
unity. In the limit there will be complete specialisation in X 2 , but
factor intensities in the physical sense will never get reversed. This
result derives directly from the fact that when commodity prices
are fixed, a change in a moves the capital/labour ratios in the same
direction. For this reason, the sign of 101 is unimportant in this
matter. For the same reason alone, if a is given, a change in com-
modity prices, which also moves the capital/labour ratios in the
same direction, can never cause a reversal in the physical factor
intensities. However, if commodity prices and a are changing simul-
taneously, k 1 and k 2 may move in the opposite direction, in which
case it is possible for the contract curve to move to the opposite
side of the diagonal. tIt is worth pointing out here that factor inten-
sities may still get reversed in the value sense, even if only one of a
and p is altered.
The implications of a shift in a for the real factor rewards have
important and interesting policy prescriptions for trade unions. To
t From (10.30) and (10.31) it is clear that k 1 and k 2 move in the opposite direction
only if (8L2a•- p*) and (8L 1a*- p*) possess conflicting signs. Evidently, the necessary
condition to ensure this result is that
Factor Market Imperfections 257

be specific, assume that the second sector is the unionised sector and
tx > 1 so that w 1 < w 2 • In other words, suppose that the producers
in the unionised sector pay a premium to labour over the labour
force employed in the non-union, first sector. Suppose further that
the union in the second sector seeks to secure an increase in this
premium, thereby increasing the value of tx. When the repercussions
on the rest of the economy are taken into account, however, the
union may end up with a lower real wage rate for its members. As
(1 0.26) and (I 0.27) suggest, with (p!- p!) = 0, the real wage rate
will decline in both sectors if I01 is negative, that is, if the unionised
sector pays a higher share to labour than the non-union sector and
hence is labour-intensive in the value sense. The reason should be
obvious from (10.30) and (10.31). With 101 < 0, a rise in tx induces
substitution of labour for capital in both sectors; resulting in a
decline in k 1 and k 2 which in turn leads to a general decline in the
wage rate and a rise in the return to capital. The policy prescription
is then unmistakable. If the union discovers that the non-union sector
pays labour a lower relative share, it should seek a reduction in the
inter-industry wage differential. The union policy could prove self-
defeating if it always pressed for higher wages.
The change in the wage differential affects not only the factor
rewards, but has repercussions for the two outputs as well. Here
again the identity of the signs of JA.I and 101 plays the crucial role.
Specifically, a rise in tx raises the output of X 1 and Jowers that of X 2
at constant commodity prices if 1.?.1 and 101 possess the same sign.
The converse is true if 1..1.1 and 101 have opposite signs.

10.9 The Wage Differential and Welfare


As with other results, the implications of a change in the differential
for welfare do not conform to one's intuition. One would expect that
a rise in the wage differential would lower national income and
welfare. This, however, turns out to be valid only if 1.?.1 and 101 have
the same sign.
Differentiating the utility function U(D 1 ,D 2 ) totally with respect
to tx, and remembering that D 1 =X 1 -E 1 , D 2 =X2 +E2 and
p£2 = £ 1 , we obtain

I dU dY dX 1 dX 2 dp
- - = - = - + p - - E2 - .
ul dtx dtx dtx dtx dtx
258 Studies in the Pure Theory of International Trade

dx. ax. dp ax.


_) = _) ·-+-).
da. ap da. aa.

From (I 0.22),
axl _ f3p aX2 and aX 1 -f3p a;z.
aa. aa. ap

Using these,

dY
da.
= P ax2 (I-/3)+ dp
aa. da.
[P ax2 (I-{3)-Ez]·
ap
(10.34)

If terms of trade are constant, dpjda. = 0, so that the sign of(dYjda.)


is determined by the sign of (aX2 jaa.) and the value of {3. Let us first
consider the case where IA.I and 181 have the same sign, so that
aX2 jaa. < 0. Then if f3 < I, (dYjda.) < 0. Now f3 < I implies that
a. > I or w 1 < w2 . A rise in a. here means a rise in the wage differen-
tial. In other words, a rise in the wage differential has resulted in a
decline in welfare, and vice versa. On the other hand, if f3 > I,
dYjda. > 0. Here a rise in a. implies a decline in the differential. Thus
our conclusion remains unchanged, that is, an increase in the dis-
tortion level leads to a decline in national income when terms of
trade are constant. It should be clear from this discussion that if I.A. I
and 181 have opposite signs so that (aX 2 jaa.) > 0, an increase in the
wage differential will lead to a rise in national income, and conversely.
A geometrical demonstration contributes much to the compre-
hension of these results. Consider Fig. 10.5, where for the sake of
simplicity the distorted transformation curve is drawn concave to
the origin. A rise in the wage differential triggered by a rise in a.
moves HH' inwards to the dotted curve; if I.A. I and 181 have the same
sign, production shifts from P toP', consumption from C to C, and
welfare declines from U2 to U1 when terms of trade are kept constant
at the slope given by FP parallel to FP' and FP". If I.A. I and 181 have
opposite signs, production switches to P", consumption to C", and
welfare improves to U 3 . Thus we see that a rise in the differential
may lead to a rise or a decline in welfare, depending on its effect on
the output of the two commodities at constant terms of trade.
Factor Market Imperfections 259

x,
~<1.0(>1

Figure 10.5

10.10 The Wage Differential and the Terms of Trade


The analysis of the preceding two sections has been conducted under
the assumption that the country with distorted factor markets is a
price-taker and has no influence on world prices. However, if the
country possesses monopoly power in trade, p will normally change
as a result of the change in IX, in which case the results derived above
may have to be revised.
Our immediate task, then, is to obtain an expression for the re-
lationship between IX and the terms of trade. With this objective in
mind, consider the effect of a change in IX on the demand for the
home country's imports (£2 ) at constant terms of trade. The balance-
of-trade equilibrium is now described by
pE2(p,1X) = Et
so that
260 Studies in the Pure Theory of International Trade

or

whence
dp P o£2/orx.
(10.35)
drx. E 2(a1 +ah-1)
where, as before, a1 and ah are, respectively, the foreign and home
elasticities of demand for imports. For stability, a1 +ah > I. Now

so that
o£2 oD2 ay ax2 ay oX2
p - = p - · - - p - = mh--p-
orx. oy arx. arx. arx. arx.
ax2
= p orx. [ mh(I - p)- 1]

because from (10.34) dYjdrx. = (poX2jorx.)(l- {3) when pis constant.


Here mh is the marginal propensity to consume importables. Sub-
stituting this in (10.35) then furnishes
dp p(oX2forx.)[mh(I-P)-l]
(10.36)
drx. E 2(a1 +ah-l)
This expression suggests that the effects of a change in rx. on the
terms of trade depend on three factors, namely, (oX2jorx.), mh and p
(or rx.). In the absence of inferior goods, 0 < mh < 1, which means
that, whatever the value of p, the square-bracketed term in the
numerator of (10.36) is negative. These considerations suggest that
the sign of (dpjdrx.) depends crucially on the sign of (oX 2jorx.). If
initially there is no wage differential, or if I.A. I and I01 possess identical
signs, (oX2jorx.) < 0, which implies that (dpjdrx.) > 0. Now rx. > 1
means that w1 ·< w2 , so that a rise in rx.leads to a rise in the differen-
tial paid by the import-competing industry; if rx. < 1, w 1 > w 2, so
that a rise in rx. causes a decline in the differential paid by the export-
able good. All this leads us to the conclusion that a rise in the differen-
tial paid by the import-competing good or a decline in the differential
paid by the exportable good gives rise to a deterioration in the terms
of trade. These results are evidently reversed if IA.I and 101 have
opposite signs so that (oX 2/orx.) > 0.
Factor Market Imperfections 261

For a graphical illustration of these results, let us revert to Fig.


10.5, where a rise in rx shifted the consumption point from C to C'
when 1)-I·IBI > 0 and to C" if IA.I·IOI < 0. The home country's de-
mand for imports in the former case rises at the constant terms of
trade given by the slope of FP, because the decline in the output of
X 2 , equal to GP, exceeds the decline in its consumption, equal to
AC', whereas in the latter case the demand for imports declines
because the rise in output of X 2 , given by RP", exceeds the rise in
its consumption, equal to BC. A rise in the demand for imports at
constant terms of trade would eventually generate a deterioration in
the terms of trade, whereas the decline in the import demand would
switch the terms of trade in favour of the home country, provided,
of course, that the foreign trade market was stable.
When terms of trade are variable, the conclusions obtained in the
last two sections under the assumption of constant commodity
prices may have to be modified, for variables in the economy are
now regulated not only by a change in a but also by a concomitant
movement in p. As established above, (dpjdrx) is positive if the factor
intensities are identical in the physical as well as the value sense,
thereby implying that (p!- p!) and rx* are negatively related.
Clearly, then, the implications of a change in a for the capital/labour
ratios, factor rewards and output levels are unpredictable from
(1 0.26)-( I 0.33). If the magnitude ot (p!- p!) is sufficiently large, all
the results derived in section 10.8 under the assumption ofgiven terms
of trade may actually be reversed.
Of greater interest is the case where the factor intensities in the
physical and the value sense are in conflict, so that (p!- p!) and a*
have similar signs. Here the change in the wage differential not only
continues to generate unambiguous repercussions on the variables
in the system, but also the results attained in the presence of un-
altered terms of trade are reinforced. The reason is not far to seek.
As argued before, a rise in a creates the same effects in the economy
as a rise in the relative price of the first commodity. Now if the factor
intensities in the physical and the value sense are identical, ptfp 2
moves in the opposite direction to the change in a when terms of
trade are variable. All the variables in the system are then subjected
to two conflicting pulls so that the final outcome cannot be pre-
dicted. In the contrary case where the physical and value factor in-
tensities are dissimilar, both the relative price of the first commodity
262 Studies in the Pure Theory of International Trade

and a shift in the same direction, and thus each shift strengthens the
ramifications of the other.
The implications of a change in the differential for welfare deserve
special consideration when terms of trade are variable, for there the
inspection of (1 0.34) makes it clear that even if the factor intensities
in the physical and the value sense are divergent, so that (dp/da) < 0,
the change in the differential does not give rise to unambiguous
shifts in community welfare, because of the ambiguity involved in
the sign of [p(aX 2 /ap)(l- {:1)- £ 2 ]. This is rather surprising because
the latter expression displays the effect of a change in the terms of
trade on welfare, which effect was demonstrated to be unambiguous
in Chapter 5. However, this premise, a detailed discussion of which
is postponed to the next section, turns out to be invalid in the
presence of distortions. The full impact of a change in the differential
on welfare can be divided into two components: (1) the effect of a
change in a at constant terms of trade, plus (2) the effect of a change
in p at a given a weighted by the magnitude of (dp/da). These com-
ponents may move in the same or the opposite direction, so that the
sign of (dY/da) is uncertain. To obtain a better perspective, let us
rewrite (10.34) as

dY = p(l-P) [ax2+ ax2 • dp]-£ 2 dp. 00 .34A)


da aa ap da da
It may be observed from (10.33) that (aX 2 jaa) and (aX 2 jap) are
always opposite in sign, which implies that the square-bracketed
expression in (10.34A) possesses a categorical sign only if (dp/da) < 0,
which in turn requires that (aX 2 jaa) > 0 or that the physical and
the value factor intensities are divergent. With (aX 2 jaa) > 0,
(aX 2 /ap) < 0 and (dp/da) < 0, the square-bracketed and the last
terms in (10.34A) are positive. However, the entire expression will
be unambiguously positive only if p < 1 or if the differential is paid
by the import- competing industry, X 2 • The conclusion that emerges
from this discussion is that (dY/da) > (aY;aa) > 0. In other words,
under the conditions cited above, the rise in the real income consequent
upon a rise in the differential exceeds the rise in income at constant
terms of trade. Another interesting result is that the real income may
rise as a result of the rise in the differential even if the physical and
the value factor intensities are identical, provided the terms of trade
are variable, because (dY/da) may be positive even if 1;-l·lel > 0
and hence (aX 2 jaa) is negative and (aX 2 jap) and (dp/da) are positive.
Factor Market Imperfections 263
~<1.ot.>1
FP 11

0
Figure 10.6

This last result is illustrated graphically in Fig. 10.6, where a rise in


IXat a constant p shifts the production point from P toP' and welfare
declines from U2 to U1 . In the new equilibrium, however, the terms
of trade move against the home country because p rises, the pro-
duction point at the new given level of IX moves to P'' and welfare
rises to U 3 , which lies not only above U1 but also above U 2 • Thus it
is clear that welfare may be positively related to IX even ifi.A.I·IOI > 0,
provided the terms of trade are variable.

10.11 The Wage Differential and the Gains from Trade


Up to now our analysis has proceeded along the assumption that
the country with distorted factor markets follows the policy of
laissez-faire. A good deal of trade-theory literature in the theory of
wage differentials has, however, raised questions as to the optimality
of the free trade policy. Several other results in the theory of gains
from trade have also been subjected to closer scrutiny. This section
is concerned with an examination of these issues under the assump-
tion that the differential is constant.
The most appropriate way to tackle the issues offree trade versus
no trade and the optimality of /aissez-faire is to differentiate the
264 Studies in the Pure Theory of International Trade

utility function totally with respect to the rate of tariff. The equations
involved in this procedure are U = U(D 1 D 2 ), D 1 = X 1 -E 1 ,
D 2 = X 2+E2,pE2 = E 1 ,ph =p(I+t 2)and(dXtfdX2) = -fJp(1+
t 2 ). Using these equations and following the procedure used many
times before, we obtain

__!_ dU = dY = PPh dX 2 (I-fJ)+p 2t 2 d£2 (10.37)


u1 dt2 dt2 dPh dPh
remembering that under the assumption of constant terms of trade
- an assumption usually made while exploring such issues -
dPh/dt 2 = p, where Ph is the price ratio in the home country. In the
absence of the differential, this expression reduces to p 2t 2(dE 2/dph)
which is necessarily negative if inferior goods are absent, so that
community welfare and the rate of tariff are negatively related;
furthermore, free trade turns out to be the optimal policy because if
t 2 = 0, (dU/dt 2 ) = 0. These results stand in need of drastic revision
in the presence of the differential, even if we assume that the price-
output response is positive as is the case in the economy without
distortions. In this spirit, let (dX2/dph) > 0; then (d£ 2/dph) < 0.
However, (dY/dt 2 ) < 0 unambiguously only if fJ > I or rx < I, so
that w 1 > w 2. In other words, only when the differential is paid by
the exportable good can we assert categorically that social welfare is
uniquely and negatively related to the rate of tariff, given that the
price-output response is positive. If fJ < 1, or if the differential is paid
by the import-competing industry, x2, social welfare and the tariff
rate are no longer uniquely related. An increase in the tariff rate may
then augment community welfare, and vice versa. It is a simple
matter now to deduce that free trade (which is the special case of
the tariff policy in that t 2 = 0) may be inferior to no trade or to a
prohibitive tariff.
Figs. 10.7 A and I 0. 7B resolve this issue geometrically. Fig. I 0. 7A
displays the case where the differential is paid by the exportable
good, X 1 , the free trade price ratio is given by FP, production is at
P, consumption at C and welfare at U 2 • A prohibitive tariff shifts
the price line from FP to DP, both production and consumption
move to S and welfare declines to V1 , indicating that free trade is
necessarily superior to no trade. In Fig. 10.7B the outcome is not so
definite, because it depends on where the self-sufficiency equilibrium
point lies. The diagram depicts the case where the no-trade produc-
tion and consumption are given by S, so that the autarkic welfare
Factor Market Imperfections 265
x, ~ >1. 0(<1 (w1 > w 2 )

0 H'
Figure !0.7a

Figure 10.7b
266 Studies in the Pure Theory of International Trade

level, U 2 , lies above the free trade level of welfare given by U1 • Thus,
we conclude that if the price-output response is positive, free trade
may be inferior to no trade, provided the differential is paid by the
import-competing industry.
Needless to say, the necessary condition for free trade to be inferior
to no trade in the presence of the differential and the negative price-
output response is that the differential is paid by the producers of
the exportable good.
These results can be explained in terms of the familiar concepts of
production and consumption losses (or gains) arising from the inter-
vention in the free inter-country flow of goods and services. t As the
rate of tariff is increased, the consumer has to pay increasingly high
prices for the importables, and this leads to a consumption loss
irrespective of the presence or the absence of the wage differential.
The consumption loss is maximised when the tariff wall is high
enough to be prohibitive. In the case of production there are two
possibilities, depending on whether the differential is paid by the
producers of X 1 or X 2 • When the differential is absent, the intro-
duction of the tariff gives rise to a decline in the output of the export-
able good and hence to a decline in the country's specialisation or
to a production loss, given that the price-output response is positive.
In general, the output structure is biased against the industry which
pays the differential, for at the prevailing market prices the output
of that industry would be higher if factor markets were undistorted.
Therefore, if the output of the industry suffering from the differential
rises consequent upon a policy shift, the economic inefficiency de-
clines in spite of the constancy of the wage differential, because the
output structure moves closer to what that structure would be in the
undistorted economy. This loss of inefficiency, then, constitutes a
production gain. The converse is true when the output of the industry
paying the differential declines. On balance, the policy switch from
free trade to no trade may give rise to the production gain or loss,
depending on which industry pays the differential. In terms of Fig.
I 0. 7A, the shift from free trade to no trade generates a production
loss as the output of X 1 , the industry paying the differential, declines,
and this loss adds to the consumption loss, so that free trade is un-
ambiguously superior to no trade. In Fig. l0.7B, on the contrary,
such a policy shift results in a production gain, and if this gain out-

t These concepts were first introduced in Chapter 4.


Factor Market Imperfections 267
weighs the consumption loss, free trade may be inferior to no trade.
This is what is pictured by pointS in Fig. 10.7B.
What, then, is the optimal policy in the presence of distorted
factor markets? Two considerations suggest themselves in this con-
nection, namely (I) the welfare-maximising policy that retains the
wage differential, and (2) the policy that eliminates the differential
to the producers. The first type of policy will generate a 'second-best'
optimum, because the inefficiency caused by the differential in
shrinking the undistorted transformation curve still remains,
whereas the second type will produce a 'first-best' optimum. The
second-best policy follows directly from our discussion concerning
the production gain (loss) in the presence of distorted factor markets.
Evidently, the production gain rises as the output of the industry
paying the differential augments. The welfare in the presence of the
differential is then maximised by introducing, in addition to free
trade, a policy of production tax-cum-subsidy such that the output
of the industry suffering from the differential is at the maximum.
Such a policy takes production toP' in Figs. 10.7A and 10.7B, where
FP', parallel to FP, is tangential to the distorted transformation
curve, consumption is at C' and welfare improves to U 3 , which lies
above U1 and U 2 • The tax-cum-subsidy policy emerges from are-
consideration of (10.37). Suppose the initial situation is one of
laissez-faire and t 2 = 0. It can easily be seen that in this case

dU = dY = pdX2 (!- /3) (1 0.38)


ul
which, of course, demonstrates that the initial situation of free trade
is not optimal in the presence of the differential, because f3 =f I,
so that dY =F 0. Clearly, then, the second-best policy will be one
that makes dY > 0 while the differential persists. Such a policy
requires that dX 2 ~ 0 if f3 ~ I, which confirms the fact that the
output of the industry paying the differential be raised to its maxi-
mum level.
An interesting query may be raised at this stage. Why should this
objective be achieved through a tax-cum-subsidy policy and not
through any other alternative such as tariffs or import subsidies,
etc.? The reason is attributable to the existence of the distortion not
in the foreign trade sector but in domestic factor markets, so that
there is a divergence between the marginal rate of transformation
and the commodity-price ratio. A production tax-cum-subsidy
268 Studies in the Pure Theory of International Trade

eliminates such divergence, whereas a tariff or import subsidy could


do this job but not without generating a divergence between the
marginal rate of substitution and the given foreign-price ratio.
Hence the production tax-cum-subsidy is superior to taxes or sub-
sidies on imports and to any other policy.
The second-best policy maximises welfare subject to the constraint
of the wage differential, and for this reason will not lead to the
optimum optimorum, which can be achieved only if the value of f3 to
the producers is equated to unity by means of a suitable factor tax-
cum-subsidy policy. Thus if f3 is equated to unity for producers,
dY = 0 from (10.38). The factor tax-cum-subsidy policy may be
administered by any one or all of the following ways which tend to
equate (wjr) in both sectors: (l) subsidise the use of labour or tax
the use of capital in the industry paying the differential, and/or (2)
tax the use of labour or subsidise the use of capital in the other
industry.
The operation of the first-best policy is illustrated in Fig. 10.8,
where the differential is shown to be paid by the X 2 producers. The
grant of the factor subsidy to X 2 in addition to free trade takes
production from P to P' on the undistorted transformation curve,
consumption moves from C to C, and welfare improves from U1 to
U 2 , which under the circumstances is the maximum attainable level
of welfare.
Let us now proceed to analyse the effects of a change in the terms
of trade on welfare. The expression for the implications of such a
change has in fact already been derived in section 10.1 0, where the
consequences of a change in IX for welfare were explored. The ex-
pression, of course, can be derived via a direct route by totally
differentiating the utility function with respect top to obtain
_1 dU = dY = p dX 2 ( 1- /3)-£2 (10.39)
ul dp dp dp
where IX is assumed to be constant. The economic content contained
in (10.39) is clear. In the undistorted case, the expression reduces to
(dYjdp) = -£2 , confirming the fact that in the absence of distor-
tions the real income rises as a result of the improvement in the
terms of trade, and vice versa. However, with (dX 2 jdp) and (1- /3)
possessing any sign in the presence of the differential, the terms of
trade and real income may no longer have a monotonic relationship,
giving rise to the possibility of an increase in real income as a result
Factor Market Imperfections 269
x,

Figure 10.8

x,
FP
FP 1

Figure 10.9
270 Studies in the Pure Theory of International Trade

of a worsening in the terms of trade, and conversely. This para-


doxical result is attributable solely to the possibility of a production
loss as the terms of trade improve and a production gain as the
terms of trade deteriorate. Fig. 10.9 depicts the case where the terms
of trade improve from the initial free trade price ratio given by the
slope of FP to that reflected by the slope of FP', production moves
from P to P', which increases the production loss as the output of
the industry paying the differential, X 2 , declines owing to a decline
in p, consumption shifts from C to C', and welfare actually declines
from u2 to ul.

10.12 The Wage Differential and Immiserising Growth


The concept of immiserising growth, which signifies that the real
income of the growing country may decline as a result of growth,
was first introduced in Chapter 6. Under the assumption of undis-
torted markets, the necessary condition for this result was shown to
be a deterioration in the terms of trade of the growing country. In
the presence of distortions, however, this condition is no longer
necessary, a result that was first discovered by Johnson [12] and
subsequently generalised by Bhagwati [3]. Even if the terms of trade
are unchanged, growth can be immiserising, provided it is accom-
panied by a production loss in the sense defined above.
The demonstration of this result consists in totally differentiating
the utility function with respect to the growth agent, G, which may
be identified with changes in factor supplies and/or technical im-
provements. There is a very simple way to obtain an expression for
dY/dG. All we need to do is to supercede da. by dG in (10.34) to obtain

_I dU = dY = oX 1
U1 dG dG oG
oX2 [ oX2 (l-p)-E
+ P oG + P op 2
JdGdp (10.40)

where (oXIfoG + poX2/oG) = o Y/oG, which as far as (10.34) is con-


cerned is equivalent too Yjoa.. With constant terms of trade, the last
term in (10.40) disappears and oYjoG is positive if factor markets
are undistorted. In the presence of the wage differential, however,
this conclusion may not be valid, as is shown in Fig. 10.10, where
growth results in an outward shift of the distorted transformation
curve from HH' to GG', the production point shifts from P to P',
consumption from C to C', and welfare declines from U2 to U1 , even
though the terms of trade are unchanged because FP is parallel to
Factor Market Imperfections 271

[!>1, 0(<1(w1 >w2J

Figure IO.IOa

H 1 G'

Figure IO.IOb
272 Studies in the Pure Theory of International Trade

FP'. In both Figs. 10.10A and 10.108 growth results in a production


loss, which requires that growth be ultra-biased against the industry
that pays the differential.
For the sake of illustration, let us consider the effect of a change
in factor supplies on national income at constant terms of trade. Let
dG = dL, that is, let there be an increase in the supply of labour
alone. At constant prices and the given distortion level, the factor
prices from (10.28) and (10.29) remain unaltered. Bearing this in
mind, (10.13) and (10.14) can be solved to yield
V* _ )..K2L*-)..L2K*
(10.41)
Ai - IA.I

and

(10.42)

which are, of course, exactly identical with (2.37) and (2.38) obtained
in Chapter 2, which shows that the Rybczynski theorem continues
to be valid in the presence of the wage differential. From these
equations,

so that
oY oX1 oX2 X1K2 -pX2K1
- = - +p - = --=-=--=-=--~----=.--__,..: .
oL oL oL L1L2(k2 -kl)'

From (10.3)-(10.5), (wtfp 1) = (X1K 2 -pX2K 1)/[L 1L 2 (k 2 -txk 1)],


which implies that
oY w1 (k 2 -txk 1)
(10.43)
oL = P1. (k2-kl).

This expression corroborates the fact that, in the presence of the


differential, (o Y/oL) may be negative; the necessary condition for
this result is that the physical and the value factor intensities be
divergent. This particular condition, however, turns out to have
stemmed from the fact that the source of growth is the rise in supply
of the factor that earns a premium in one sector. For example, if
Factor Market Imperfections 273

growth occurs owing to the rise in the supply of capital alone, then,
from (10.41) and (10.42),

so that
oY pL1X2 -L2X1
oK L1L2(k2 -a.kl)"
Now, from (10.3)-(10.5),
r pL 1X 2 -a.L 2X 1
P1 L1L2(k2-a.k1)
whence
oY r (k 2 -a.k 1) (pL 1 X 2 -L 2X 1)
oK = P1. (k2 -kl) (pL1X2 -a.L2Xd
which makes it clear that (o Y/oK) will be negative even if the
physical and the value factor intensities are identical, provided
(pL1X2-L 2X 1 ) and (pL 1X 2 -a.L 2X 1)"have opposite signs. Thus
the conclusion is unmistakable. For growth to be immiserising with
unchanged terms of trade, the reversal of the physical and the value
factor intensities is necessary only if output expansion occurs because
of an increase in the supply of the factor that earns a premium in one
sector.
Having demonstrated that growth may be immiserising in the
presence of distorted factor markets, even when terms of trade are
constant, a simple inspection of (1 0.40) reveals that this immiserisa-
tion may be reinforced as a result of the growth-induced improve-
ment in the terms of trade, or that if(oYjoG) was positive, (dY/dG)
may be negative in spite of, or rather because of, the terms of trade
improvement fostered by growth and so on. t

The Prebisch Hypothesis


The results derived in this section have a close bearing on the still
inconclusive debate initiated by Prebisch [16] over the hypothesis
of the secular deterioration in the terms of trade facing the under-
developed countries producing primary products. The proponents

t For further discussion of this resu1t, see Batra and Scully [2].
274 Studies in the Pure Theory of International Trade

of this thesis argue that the cause of poverty in the developing nations
lies in the deterioration in their terms of trade which started in the
latter part of the nineteenth century and persisted until the beginning
of the Second World War. Participants on the other side of the
debate have strongly challenged the empirical and analytical sound-
ness of this hypothesis.
We have no intention of getting involved in this still mooted issue.
But participants on both sides of the debate have displayed at least
tacit consensus on the fact that the deterioration in the terms of
trade by itself is a bad thing, that it leads to a decline in the rate of
growth of income per se. However, we have shown that such a de-
terioration is no longer directly related to the rate of growth if factor
markets are imperfect. To the extent that factor markets in the
underdeveloped countries are very likely to be imperfect, it is neces-
sary first to decide whether or not a deterioration in the terms of
trade is necessarily harmful before arguing for it or contesting it as a
sufficient explanation for ubiquitous poverty in the underdeveloped
world.

10.13 Other Types of Factor Market Distortions: Factor Immobility


and Factor-Price Rigidity
Distortions in factor markets can arise not only from inter-industry
factor-price differentials, but also from imperfect mobility of factors
and downward rigidity of factor prices. The implications of the latter
type of distortions for gains from trade have been analysed by
Haberler [6], Johnson [II] and Batra and Pattanaik [1], and this
section is devoted to an examination of their results. Throughout
the book so far we have continued to assume that factors are per-
fectly mobile between the two sectors and that factor prices are
perfectly flexible. Both of these are extreme assumptions and it will
be useful to see how the relaxation of these assumptions modifies
the results. For expository purposes, let us assume that factors are
perfectly immobile and factor prices perfectly rigid, notwithstanding
the fact that in general there is some degree of factor mobility and
factor-price flexibility. The interested reader is referred to Haberler
[ 6] to see that the conclusions derived under the extreme assumptions
remain qualitatively unmodified when these assumptions are re-
laxed. Furthermore, whether or not factor prices differ between
sectors turns out to have no bearing on the results under our extreme
assumptions.
Factor Market Imperfections 275
The method of analysis followed by Haberler as well as Johnson
is to start from a position of no trade, introduce free trade, and then
compare the level of social welfare attained in these two situations.
In actual practice, however, the need for protection arises when the
country under consideration, following a policy of free trade, con-
siders it detrimental to its welfare. That Johnson himself is aware of
this will be clear from the following passage:
For the analysis of arguments for protection derived from im-
mobility of factors and downward rigidity of factor prices, it is
convenient to pose the problem in terms of whether the opening
of the opportunity to trade makes a country worse off when these
conditions exist, so that a prohibitive tariff would secure a higher
level of welfare than could be attained under free trade, even
though in reality the argument for protection usually arises when
trade is already established and the international price of imports
suddenly falls. ([II] p. 14)
Johnson further asserts that 'the difference of assumptions merely
simplifies the problem without altering the conclusions' (ibid.). One
of the objectives of this section is to show that 'the difference of
assumptions' does alter the conclusions. Specifically, in contrast to
the Haberler-Johnson thesis, it will be shown that free trade is
necessarily superior to no trade even in the presence of inflexible
factor prices, provided we start from a situation of free trade and
then introduce a prohibitive tariff.

From No Trade to Free Trade


To begin with, let us see what the transformation curve looks like
when factors of production cannot move from one sector to another.
Consider Fig. I 0.11, where H H' is the transformation curve derived
under the assumption that factors are perfectly mobile between the
two industries X 1 and X 2 , S is the point of self-sufficiency
equilibrium, the autarky price ratio is given by the slope of DP, and
U1 reflects the level of welfare. If factors are completely immobile
but factor prices continue to be flexible, the transformation curve
reduces to the rectangle formed by ASB, reflecting that the output
of X 1 and X 2 is fixed as no factor moves from one industry to another;
the fact that factor prices are still flexible ensures that factors con-
tinue to be fully employed and that the production point remains
unaltered, irrespective of the level of relative commodity prices. A
276 Studies in the Pure Theory of International Trade

Figure I 0.11

switch from no trade to free trade at the given world prices reflected
by the slope of FP takes the consumpti on point to C, but under the
postulated conditions leaves the production point unchanged at S.
However, the immobility of factors does not interfere with the
superiority of free trade over no trade as U 2 lies above U1 • This
occurs because, although production gain is zero, the consumpti on
gain remains.
Consider now the case where factor immobility is accompan ied
by factor-pric e rigidity; the switch from no trade to free trade would
cause unemploy ment of some factors employed in X 2 as a result of
the decline in its relative price from that given by the slope of DP to
one furnished by the slope of FP. As a consequen ce the output of X 2
will decline, although that of X 1 will remain unchanged as unutilised
factors cannot move from X 2 to X 1 . In other words, the consumpti on
gain in the presence of factor-pric e inflexibility will be accompan ied
by the production loss, and if the latter is large enough, free trade
may be inferior to no trade. This is the possibility depicted in Fig.
10.11, where free trade takes production toP, consumpti on to C'
and welfare to U0 which lies below the autarky level, U1 • It is this
Factor Market Imperfections 277

possibility which was discovered by Haberler and examined further


by Johnson.

From Free Trade to No Trade


Let us now reverse the reasoning and consider the case where free
trade is already established and the country decides to introduce a
prohibitive tariff. Let us go to Fig. I 0.12, where the rectangular
transformation curve is given by APB, and where Pis the free trade
production point lying on the transformation curve H H' (not drawn),
C is the corresponding consumption point and U2 is the level of
welfare. A prohibitive tariff unaccompanied by factor-price rigidity
leads to production and consumption at P and to the welfare level
given by ui; if factor prices are also rigid, the production of xi will
decline as a result of the switch from FP to D P, whereas the output
of X 2 will be unchanged as no factor can move from Xi to X 2 • This
case is then just the reverse of the one discussed above. Suppose the
new production point is given by S. This will also be the consump-
tion point, so that the welfare level will be given by U0 which lies
below U 2 • Free trade is then necessarily superior to no trade, in spite
of rigidity of factor prices. The reasoning is clear. If the initial situ-
ation is one of full employment, any change in relative prices causes

x,

8
Figure 10.12
278 Studies in the Pure Theory of International Trade

a production loss as the output of one industry declines and that of


the other remains unchanged. Under this setting, a movement from
no trade to free trade gives rise to a consumption gain and a produc-
tion loss, so that the final outcome for welfare is uncertain, whereas
a switch from free trade to no trade causes losses in production as
well as consumption, so that free trade is necessarily superior to
no trade.

Initial Unemployment
Both Haberler and Johnson assume full employment in the absence
of trade in spite of inflexibility of factor prices, and unemployment
of a fraction of given factor supplies is caused in their framework
only by the introduction df trade. But why should not the rigidity
of factor prices result in some unemployment even in the absence of
trade, or for that matter in the initial situation of free trade? The
analysis of rigid factor prices will not be complete unless this con-
sideration is taken into account.
In the presence of unemployment, the initial production point
will lie inside the transformation curve. For the sake of brevity, we
consider only the case involving a shift from free trade to no trade,
because in the other case of moving from no trade to free trade the
results are qualitatively unchanged. Consider Fig. I 0. I 3, where P'
is the production point that would have prevailed if there was full
employment at free trade prices; the actual production point, how-
ever, is given by P, consumption is at C and welfare under free trade
at U 2 • The consumption loss resulting from the imposition of a
prohibitive tariff leads to the level of welfare given by ul, with p
now being the production as well as the consumption point. At the
no-trade price ratio given by DPthe output of X 1 will decline, so that
there will be an increase in unemployment of factors specific to X 1 ,
whereas the output of X 2 will rise, which it can because of the exist-
ence of a pool of unemployed factors specific to it. As a consequence,
the new production point will lie in the rectangle BPRT. Evidently,
then, the level of no-trade welfare may lie above or below the free
trade level or even remain the same, depending on the final produc-
tion point. Fig. 10.13 pictures the case where the self-sufficiency
equilibrium is at Sand the autarky level of welfare, given by U3 ,
lies above the free trade level U 2 . It is here that the conclusion is
substantially different from the case where there is full employment
initially.
Factor Market Imperfections 279
x,

r-------------------------------~pt
DP 1

0 8 T

Figure 10.13

10.14 Concluding Remarks


One truth which emerges again and again in the preceding analysis
is that virtually all the results derived from undistorted factor
markets may be reversed when factor market distortions are taken
into account. A Pandora's Box of paradoxes is opened, the moment
the assumption of undistorted factor markets is relaxed. However,
when we consider the intuitive explanations that are available for
all the paradoxical results, the semblances of paradox should dis-
appear. Nevertheless, the novelty of these results catches one's
imagination, and perhaps for this reason the recent trade-theory
literature abounds in the discussion of factor market distortions.

REFERENCES
[I] Batra, R. N., and Pattanaik, P. K., 'Factor Market Imperfections and Gains
from Trade', Oxford Economic Papers, XXIII (July 1971) 182-8.
[2] ---,and Scully, G. W., 'The Theory of Wage Differentials: Welfare and Im-
miserizing Growth', Journal of International Economics, 1 (May 1971) 241-7.
[3] Bhagwati, J. N., 'Distortions and Immiserizing Growth: A Generalization',
Review of Economic Studies, XXXV (Oct 1968) 481-5.
280 Studies in the Pure Theory of International Trade

[ 4] Bhagwati, J. N., and Ramaswami, V. K., 'Domestic Distortions, Tariffs and the
Theory of Optimum Subsidy', Journal of Political Economy, LXXI (Feb 1963)
44-50.
l5J --,and Srinivasan, T. N., 'The Theory of Wage Differentials: Production
Response and Factor Price Equalization,' Journal of International Economics,
I (Feb 1971) 19-35.
[6] Haberler, G., 'Some problems in the Pure Theory of International Trade',
Economic Journal, LX (June 1950) 223-40.
[7] Harberger, A. C., 'The Incidence of the Corporation Income Tax', Journal of
Political Economy, LXX (June 1962) 215-40.
[8] Herberg, H., and Kemp, M. C., 'Factor Market Distortions, the Shape of the
Locus of Competitive Outputs, and the Relation between Product Prices and
Equilibrium Outputs', in J. N. Bhagwati eta/ (eds.), Trade, Balance of Pay-
ments, and Growth (Amsterdam: North-Holland, 1971 ).
[9] - - , - - , and Magee, S. P., 'Factor Market Distortions, the Reversal of
Relative Factor Intensities, and the Relation between Product Prices and Equi-
librium Outputs', mimeographed (1970).
[10] Johnson, H. G., 'Factor Market Distortions and the Shape of the Transforma-
tion Curve', Econometrica, XXXIV (July 1966) 686-98.
[II]--, 'Optimal Trade Intervention in the Presence of Domestic Distortions',
in Trade, Growth and the Balance of Payments (Chicago: Rand McNally, 1965).
[12] --,'A Note on Distortions and the Rate of Growth of an Open Economy',
Economic Journal, LXXX (Dec 1970).
[13] Jones, R. W., 'The Structure of Simple General Equilibrium Models', Journal
of Political Economy, LXXII (Dec 1965) 557-72.
[14] Lewis, W. A., 'Economic Development with Unlimited Supplies of Labour',
Manchester School of Economics and Social Studies, XXII (May 1954) 139-91.
[15] Magee, S. P., 'Factor Market Distortions, Production, Distribution and the
Pure Theory of International Trade', Quarterly Journal of Economics, LXXXV
(Nov 1971) 623-43
[16] United Nations, Department of Economic Affairs, Relative Prices of Exports
and Imports of Underdeveloped Countries (1949).
SUPPLEMENTARY READINGS
[17] Batra, R. N., and Pattanaik, P. K., 'Domestic Distortions and the Gains from
Trade', Economic Journal, LXXX (Sep 1970) 638-49.
[18] --.and--, 'Factor Market Imperfections, the Terms of Trade, and Wel-
fare', American Economic Review, LXI (Dec 1971) 946-55.
[19] --.and Casas, F. R., 'Factor Market Distortions and the Two-Sector Model
of Economic Growth', Canadian Journal of Economics, IV (Nov 1971) 524-42.
[20] Hagen, E. E., 'An Economic Justification of Protectionism', Quarterly Journal
of Economics, LXXII (Nov 1958) 496-514.
[21] Magee, S. P., 'Factor Market Distortions, Production and Trade: A Survey',
Oxford Economic Papers (forthcoming).
[22] Pearce, I. F., 'The Theory of Wage Differentials: The n x n Case', Journal of
International Economics, I (May 1971) 205-14.
11 Product Market
Imperfections: The Theory of
Monopoly in General
Equilibrium

The previous chapter was devoted to the analysis of factor market


imperfections in the presence of international trade. The quintes-
sence of the chapter was that the existence of distortions in factor
markets could cause reversals in nearly all the results derived from
the standard undistorted model. Throughout our analysis there,
product markets were still assumed to be perfect. In this chapter we
turn to the other side of the exercise and assume that different
goods are produced by different monopolists, but that factor
markets continue to be perfect. It is worth pointing out here that
within the theory of market imperfections it is the distortions in
factor markets that have borne the brunt of attack from the trade
theorist, whereas the existence of distortions in product markets,
perhaps because of the intractability of the problem, has by and
large been ignored.
The implications of monopoly in general equilibrium constitute
the subject of analysis in two recent articles by Melvin and Warne
[2] and Batra [I]. Our analysis here runs parallel to theirs. To the
delight of those who continue to have undying faith in the existence
of competitive markets, the presence of product market distortions
alone, in contrast to factor market distortions, leaves virtually all
the results derived from the undistorted model unchanged. This is
perhaps the major difference between product and factor market
distortions. The existence of monopoly does, however, generate a
rich variety of new possibilities in terms of the changes in monopoly
profits - absent under competitive conditions - that take place
consequent upon any change in the system.
282 Studies in the Pure Theory of International Trade

11.1 The Model with Pure Monopoly


The assumptions of this chapter are the same as those maintained
under the standard model, except that product markets are now
characterised by pure monopoly. In order to ensure the continued
existence of monopoly, we assume that there is no freedom of entry
into the product markets. tAll the other assumptions specified at the
beginning of Chapter 2 are unaltered. All this implies that the basic
modification introduced by monopoly occurs in the price equations,
whereas the full-employment equations remain unscathed. In this
spirit, we write
cLixl + CL 2 X 2 = L (11.1)

cKlxl + cK2x2 = K. (11.2)

The price equations now include monopoly profits. Let C,i stand
for monopoly profits per unit of output in thejth sector. Then

(11.3)
and
(11.4)

As regards the determination of C,i, we note that under monopoly


the reward of each factor equals the marginal revenue product,
which in turn equals the marginal revenue (M R) times the marginal
product. Now, as is well known in micro-theory,

MR.= P·
J J
(1-_!_)
6j

where si is the price elasticity of demand for thejth product. In other


words, under monopoly

t Freedom of entry may be restricted by assuming that either all production takes
place under government franchise, patent laws protect the monopolist, or the
monopolist has such an efficient technology that no other producer can possibly com-
pete with him.
Product Market Imperfections 283

where MPK and MPL denote, respectively, the marginal product


of capital and labour. Since the excess profit equals total revenue
(TR) minus total cost (TC), we may write

C.= TR.-
J
TC.) = p.X.-(rK+wL)
J J J J
"J X X
J J

= Pi~-pJI-(1/t)][MPKiKi+MPLiLi]
(11.5)
~
because, from Euler's theorem, MPKiKi+MPLiLi = ~· Under
competitive conditions ei = oo, so that the excess profits are zero.
With monopoly, however, C,.i > 0, and this brings us directly to the
problem of determining ei in the context of our general equilibrium
system. This indeed is the key to the solution of the system. Perhaps
the main reason for the neglect of monopoly in the traditional
models has been the difficulties encountered in specifying func-
tional forms for ei in the context of general equilibrium. Melvin and
Warne have provided an ingenious way out by assuming a homo-
thetic social utility function, one implication of which is that the
elasticity of demand for each product becomes a function of the
commodity-price ratio and nothing else. In this chapter we assume
that the social utility function is of the constant elasticity of substi-
tution (C.E.S.) variety. In a closed economy,

where a, b and p are the parameters such that a > 0, b > 0,


- I < P < oo, P 1= 0 and u0 = 1/(1 + p), where u0 is the elasticity
of substitution between the two goods in consumption. Purely for
the sake of simplicity of calculation, we assume that afb = I. For
this utility function, the demand functions (derived in the appendix,
section 11. 7) are

whence
xl _ p(l +p-/II1D) _ (1-flao) - p"D
x2- (l+pflan) -p -
284 Studies in the Pure Theory of International Trade

where Y is national income. From these demand functions, ei is


given by
_I +uopfl"n
el - I +pfl"n (11.7)

and
(11.8)

where
ei = _a~ .Pi (j = 1,2).
opj ~
These expressions confirm the fact that, under the homothetic
utility function, ei is determined solely by p. Furthermore, since the
numerator and denominator of (11.7) and (11.8) differ only by u0 ,
the magnitude of ei is determined only by the magnitude of u 0 .
Specifically, ei ~ I, if u0 ~ 1.

11.2 The Consumer-Producer Equilibrium


The major problem arising from the introduction of monopolistic
product markets into the general equilibrium analysis of a two-
sector model stems from the interdependence of supply and demand
decisions. In other words, the fundamental difference between
perfect competition and monopoly equilibria is that while aggregate
demand and supply for products are independent and jointly deter-
mine outputs and prices in perfect competition, in monopoly this
dichotomy does not exist. The general equilibrium problem under
competitive conditions has three aspects. Given a production func-
tion and fixed factor and product prices, individual firms attempt to
maximise their profits and in the process generate demand for factor
services. Their decisions determine demand and supply functions
for factor services and outputs, respectively. On the other side, given
a utility function and taking the factor and product prices as fixed,
the individual households offer factor services and buy goods in such
a way as to maximise their flow of satisfaction derived from the
consumption of commodities. Their decisions determine supply
functions for factor services (which we assume are inelastic) and
demand functions for products. Given these demand and supply
functions, equilibrium prices are determined in such a way as to
clear all product and factor markets.
Product Market Imperfections 285

However, under monopoly conditions in product markets alone,


although the households make their decisions independent of the
production side, the individual monopolists need data about
demand for their products in order to make profit-maximising price-
output decisions. This is evident from the expressions for factor
rewards, which yield
p2 MPL 1 [1-(l/e 1 )] MPK 1 [1-(l/e 1 )]
p = P1 = MPL 2 [1-(l/e 2 )] = MPK2[1-(lje 2)J' (ll. 9)
In perfect competition, on the other hand,
MPL 1 MPK 1
p=--=--
MPL2 MPK2

and it is clear that the producers' price ratio is determined indepen-


dent of the consumers' price ratio furnished by (11.6). However,
(11.9) shows that, under monopoly, producers' price ratio depends
not only on marginal factor productivities but also on the elasticities
of demand for the two products. This dependence of monopolist's
price on consumer's demand raises an important question about the
conditions necessary to ensure the existence of equilibrium. In this
section we shall show that, for any given capital/labour ratio, k, all
the variables in the model are uniquely related to the wage/rental
ratio, w. It is evident from ( 11.9) that pis positive if each elasticity of
demand is greater than unity. This is, of course, the well-known
result in monopoly theory, for the profit-maximising monopolist
produces an output where marginal revenue equals positive marginal
cost, and the former is positive only if the elasticity of demand ex-
ceeds unity. Thus we may conclude that a necessary condition for the
monopoly equilibrium to exist is that both price elasticities of demand
are greater than unity, or that uD > 1.
We have already stated that, under monopoly, producers' price
ratio depends, among other things, on the two demand elasticities,
which in tum can be shown to depend on the price ratio facing the
consumers. Suppose we impose any price ratio on the consumers.
From (11.6) we know that the ratio in which commodities may be
consumed is uniquely determined by the product-price ratio. Cor-
responding to this price ratio there also exists a set of elasticities of
demand, and given these and the marginal factor productivities
from the production functions, the producers' price ratio will be
286 Studies in the Pure Theory of International Trade

determined uniquely from (11.9). The question is whether or not,


and under what conditions, the producers' price ratio will be the
same as the price ratio that was initially imposed on the consumers.
We begin by showing that, under the homethetic utility function,
the change in the commodity-price ratio facing the consumers (Pc)
is uniquely related to the change in the output ratio (Xd X2 ). Differ-
entiating (11.6) totally, we obtain
(11.10)
The next step then is to obtain an expression for (Xi- Xi) in terms
of the change in the wage/rental ratio, for the latter can be easily
seen to be uniquely related to the price ratio encountered by the
producers. This can be accomplished by differentiating the entire
system of equations in section 11.1. From ( 11.1) and ( 11.2)
A.L1Xi+A.L2Xi = L*-(A.Llqt +A.uC!2) (11.11)
A.xt Xi+ A.x2Xi = K*- (A.Kt Ctt + A.K2Cl2) (11.12)
where A.ii is as before. Differentiating the price equations (11.3) and
(11.4) totally and using the minimum unit cost condition,t
eLjctj+eKjctj = o
t If factor markets are perfect, then the monopolist's profits are maximised when
the given factor prices are equated to their marginal revenue productivities, which
means that
w MPLi
r MPKi
But this is a condition for unit cost minimisation, which is given by
wdCLi+rdCKi = 0.
The assumption of perfect factor markets, however, needs some justification, for the
monopolist is likely to be a monopsonist and therefore aware of the fact that the factor
supply curves facing him are not perfectly elastic, in which case the factor prices can-
not be considered as given to the producers. This assumption is needed mainly for
simplifying the mathematical calculations. Actually, in order to utilise the minimum
unit cost condition, what we need to assume is that w and r are considered given by
those who hire the factors of production. Suppose the producer is a multi-plant
monopolist, and although the price-output decisions are made at the top level of
managerial hierarchy, the decisions to hire factors are made at the plant level; if there
are numerous plants, each contributing its share of the total output, then for all prac-
tical purposes the desiderata for perfect factor markets will be satisfied.
Another solution to the problem could be to assume that factor markets are
unionised and the monopolist can hire any amount of inputs at the factor prices
which the union agrees to accept. Here again the producer can consider the factor
prices as given.
Product Market Imperfections 287
we obtain
OLlw* +0K 1 r* = pf(l-0,. 1)+0l<lei (11.13)
and OL 2 w*+OK 2 r* = p~(l-0,. 2 )+0,. 2 e~ (11.14)
where Oii is the relative share of the ith factor in the jth good, and
where i includes labour, capital and the monopolist. For example,
0,. 1 = C,.tfp 1 = l/e 1 • In other words, O,.i is the share that goes to
the monopolist's coffers. Clearly, OLi+OKi+O,.i = 1. As before, let
A. and 0 be the matrices of production coefficients in (11.11) and
(11.12) and (11.13) and (11.14) respectively. Then

2 (k 2 -k
I L 1 L L·K
IA.= 1)
=(A.Ll-A.Kt)=(AKz-ALz) (11.15)

and

As expected, the expression for IA.I is the same as that obtained in


the undistorted model. Furthermore, the signs of lA. Iand 101 are the
same. However, when factor intensities are defined in terms of
distributive share, i.e. lJ's, some interesting possibilities come to
light. Suppose IA.I and 101 are positive, so that k 2 > k 1 • Under the
competitive conditions where O,.i = 0, this would imply that
OLl > OL 2 and OK 2 > OKt· With monopoly, however, this need not
be the case, for 101 could be positive even if OK 2 < OKl• provided
(1- 0><1) is sufficiently greater than (1- 0,. 2 ). In other words, even if
a commodity is capital-intensive, it may have a lower share of capital
than the labour-intensive commodity, provided the former pays a
higher share to the monopolist than the latter. Similarly, a labour-
intensive commodity may actually pay a lower share to labour than
the other.
As usual, the expression for Cli and c:i can be obtained by the
interaction of the minimum cost conditions and the definition of the
two elasticities offactor substitution t1j· In the presence of monopoly
Cli = -0Kit1j(w*-r*)/(1-0,.)
and
C1i = 0Ljt1j(w*-r*)j(1-0,.).
288 Studies in the Pure Theory of International Trade

Substituting these in ( 11.11) and ( 11.12) yields


AL1Xt+A.L2X! = L*+PL(w*-r*) (11.17)
and A.K 1Xt + A.K2X! = K*- PK(w*- r*) (11.18)
O
where PL = AL10K1U1
(1-0,.1)
+ AL20K2U2
(l-0,.2)
>

and p _ A.K10L1U1 AK20L2U2 0


K - (1-07<1) + (l-0,.2) > .
Subtracting (11.18) from (11.17) and using the expression for IA.I
from (11.15), we obtain

X •-x• _ L*-K*+(PL+PK)( •- *) (11.19)


1 2 - IA.I IA.I w r .

Substituting (11.19) in (11.10), we get

• _ (PL+PK)( •- *) (11.20)
UoPc - IA.I w r

given that factor endowments are constant, so that L • = K* = 0.


One glance at (11.20) reveals that the price ratio facing the con-
sumers, Pc• is uniquely related to the wage/rental ratio and that this
relationship is determined by the sign of lA. I. It is a relatively simple
matter now to observe that a similar relationship also exists between
the price ratio facing the producers (call it Pp) and the wage/rental
ratio. This can be obtained by solving (11.13) and (11.14) simul-
taneously to derive
IOI(w*-r*) = -[(l-07tl)(l-0,. 2)p;
- 07<1 (1- O><l)ef + 0,. 2(1- 07tl)eU
The required expression for et
in turn can be obtained by totally
differentiating (11.7) and (11.8). It turns out that each demand
elasticity is an increasing function of its relative price, becauset

e• - -
P2 UoPp
2 {ian
• - -A • (A 0) (11 21)
1 - [l + p:an] [l + UoP:an] Pp - 1Pp 1 > ·

t Equations (11.21) and (11.22) simply show that as the monopolist, facing one of
the demand curves derived from the utility function, raises his price, the elasticity of
demand for his product rises and vice versa. It is for this reason that e; is shown to be
related to the producers' price ratio rather than the consumers' price ratio, although
on the consumption side e; is related top,.
Product Market Imperfections 289

p2u2 p-flav
e* - D P p* - A p* (A 2 > 0). (11.22)
2 - [1 +pp flan][I +UnPp flav] P - 2 P

Substituting these in IeI(w.- r *) then yields


w* = w*-r*

(E > 0). (11.23)

This expression provides corroboration of the fact that the pro-


ducers' price ratio is also uniquely related to the wage/rental ratio.
With this last expression, we have reached the stage where we can
derive conditions necessary to bring about the equality of pP and Pc
in equilibrium. A comparison of(ll.20) and (11.23) reveals that the
relationships between p c and w on the one hand and that between p P
and won the other are completely opposite simply because 101 and
IA.I possess the same sign. This ensures that there is only one wage/
rental ratio at which Pp and Pc are equal. The result can best be
illustrated in terms of Fig. 11.1, which is drawn under the assump-

Pc

Figure 11.1
290 Studies in the Pure Theory of International Trade

tion that lA. I and 101 are positive, so thatpP is shown to be a decreasing
and p, an increasing function of w. The equilibrium commodity-
price ratio is Pe• and corresponding to it the equilibrium wage/rental
ratio is we. It is manifestly clear from the diagram that there is only
one wage/rental ratio at which the producers' and consumers' price
ratios are identical. It must be remembered, however, that the
monopoly equilibrium exists only under the constraint of an ex-
ceeding unity, for otherwise pP will not be positive.t

11.3 The Price-Output Response and the Nature of Autarky


Equilibrium
Once the matter of commodity prices facing the consumers and
those facing the producers being equal is settled, we can proceed
to analyse other questions which until now have been explored
under the conditions of perfect competition in product markets.
This section is concerned with the change in the two outputs in
response to the shift in relative commodity prices. If the factor en-
dowments are kept constant, so that L * = K* = 0, the solution of
(11.1 7) and (11.18) yields
X*_ (A.K2fh+A.L2f1K)(w*-r*)
I - IA.I
and
X* _ -(A.LifJK+A.KlfJd(w*-r*)
2 - 1.-1.1
Substituting (11.23) in these, we obtain

* -E(A.K2fJL +A.L2fJK)(p! -pi)


(11.24)
xl = IA.IIOI
and
E(A.LI PK+ A.Kl{JL)(p!- pi)
( 11.25)
IA.IIOI
Since IA.I and 101 possess the same sign, and since E > 0, it is clear
that supply curves are positively sloped, that is, a rise in the relative
price of a commodity raises its output at the expense of the output of
the other commodity.
t For further exposition of the existence and the uniqueness of monopoly equilib.
rium, see Batra [1] and Melvin and Warne [2].
Product Market Imperfections 291

This discussion leads us directly to the questions of the slope and


shape of the transformation curve in the present framework. One
point should be clear at once. Since factor markets are still undis-
torted, the position of the transformation curve remains unaltered.
Nevertheless the slope of the transformation curve no longer reflects
the commodity-price ratio, for the former is given by the ratio of the
marginal costs in the two industries which, under competitive con-
ditions, are equal to the respective prices, but which under monopoly
are equal to the respective marginal revenues. In other words, under
monopoly
dX1 MR 2 p 2 [l-(l/e 2 )]
(11.26)
- dX2 = MR 1 = p 1 [l-(l/e 1 )] = 1Xp

where IX = 1 - ( l/e 2 ). It can be readily seen that IX ~ l, according as


l-(l/e 1)
e2 ~ e1 . Thus it is only when the elasticities of demand for both
goods are the same that the M RT equals the commodity-price ratio.
As already noted, the absence of distortions in factor markets
guarantees that the position of the transformation curve itself is not

x,
45°

0
Figure 11.2
292 Studies in the Pure Theory of International Trade

affected. At the same time the assumption of full employment en-


sures that the equilibrium production point will in fact lie on the
production locus. Is it then possible to determine uniquely the
location of the equilibrium production point under monopoly? The
pursuit of a satisfactory answer to this question takes us to Fig. 11.2,
where HH' is the transformation curve. If there were no monopoly,
the equilibrium production point in autarky would be determined
by the tangency of the utility function to the production possibility
locus HH'. Such a production point in the diagram is given by C,
and corresponding to it the level of welfare is given by Uc. As
another point of reference, let us locate the point where e 1 = e2 •
This, from (II. 7) and (11.8), is possible only if

Substituting this value of p in (11.6), we find that

xl =I
x2 .
Thus, when the two elasticities of demand are equal, the output of
the two goods is also the same. This rather unusual result is attribu-
table to our particular assumption made for simplifying the cal-
culations, namely, (a/b) = 1. In general, the output ratio cor-
responding to the point where e 1 = e2 is determined by the value
of the parameters a, band {3. Draw a 45° line from the origin which
intersects the production possibility locus at E. Then along any
point on the ray OE, e 1 = e2 • One characteristic of the production
point given byE is that at this point theM RT equals the commodity-
price ratio. Thus if the social indifference curve were to touch HH'
at E, the autarky equilibrium under monopoly would be identical
to that under competitive conditions. However, E is one of an
infinity of possible production points and there is no a priori reason
to believe that community preferences will be such as to lead to the
competitive solution, especially when the production and utility
functions are independent. Let an indifference curve Ue pass through
E. Then it is evident that atE, MRT = -pP > MRS= -Pc· A
movement away from E in either direction creates a divergence be-
tween the M RT and pP" As we move along HH' towards H, X1 rises
and X 2 declines, so that the price ratio facing the consumers rises (as
can be confirmed from ( ll.l 0)) and hence e2 rises and e1 declines.
This in turn leads to a rise in oc above unity, and from (11.26) the
Product Market Imperfections 293

M RT comes to exceed Pr This is true of any point between E and


H (excluding E and H); but at C, MRT = Pn which implies that
there Pc > Pp· Thus we have two points of reference. At E, pP > Pc
and at C, Pp < Pc· Hence the production point where Pp = Pc will
necessarily lie between E and C. This is how Melvin and Warne have
been able to reach the conclusion that the monopoly output ratio
is bounded by the competitive output ratio and the one correspond-
ing to the point representing the equality between e1 and e2 . Suppose
such a production point is given by M, through which passes another
indifference curve Um. Fig. 11.2 then also indicates that the monopoly
welfare level lies below the competitive level of welfare.
Until now .we have maintained the distinction between Pp and Pc
in view of the fact that they are equal only in equilibrium. Since the
subsequent sections are concerned with comparative statics where
two different equilibrium situations are compared, this distinction
will no longer be necessary.

11.4 The Standard Trade Theorems


Up to now our main concern has been with the demonstration that a
unique equilibrium under monopoly situation in both sectors does
exist, provided aD > l. The model developed in the previous sections
can be easily applied to the standard theorems in international trade.
Let us begin with the Rybczynski theorem, which is concerned
with the implications of changes in factor supplies on commodity
outputs at constant terms of trade. If commodity prices are constant,
then the wage/rental ratio is constant, which implies that the last
expression on the right-hand side of ( 11.17) and ( 11.18) disappears
so that we are left with the usual equations that were encountered in
several previous chapters where monopoly was absent. The in-
terested reader can easily see for himself that the Rybczynski theorem
holds in the present model without any qualification. This can also be
confirmed from (11.19) by setting (w*- r*) to zero.
The validity of the Rybczynski theorem coupled with the assump-
tion of international identity of consumption patterns ensures the
validity of the Heckscher Ohlin theorem in terms of the physical
definition. Furthermore, since commodity and factor prices continue
to be uniquely related, given of course the non-reversal of factor
intensities, the Heckscher-Ohlin theorem also continues to be valid in
terms of the factor-price definition of inter-country relative factor
abundance.
294 Studies in the Pure Theory of International Trade

Closely related to the Heckscher-Ohlin theorem is the theorem


concerning international equalisation of factor prices. In free trade,
prices are everywhere the same, which, with similar inter-country
utility functions, implies that e1 and e2 will be the same in the trading
partners. Given the unique relationship between commodity and
factor prices, the wage/rental ratio will be the same in each country.
From the expressions concerning c:i and C[i, one can see that
kj = c:i-Cti = ai(w*-r*)
which is the same as the one existing in the undistorted markets.
Given the international similarity of production functions, the
capital/labour ratio in each commodity will be the same in both
trading countries; this in turn will ensure the equality of real factor
rewards in the two countries, for under linearly homogeneous pro-
duction functions the marginal product of each factor depends
solely on the capital/labour ratio in thejth industry. Since in addition
each industry will have the same marginal revenue in both countries,
factor rewards will also be the same everywhere.
An interesting question concerns the profits per unit of output in
each country. With so much information at hand, this matter can be
resolved by recourse to the definition of C"i' which from (11.5) is
given by

With Pi and ei identical in the two countries, there is no reason why


C"i is not also identical. If the rate of monopoly profit is defined as
total profits (ni) per unit of the capital stock (presumably owned by
the monopolist), then
1ti 1ti Xi Li C"i.APLi PrAPLi
Ki = Xi Li Ki = ki = eiki
where APLi is the average product of labour in the jth industry.
Since APLi and MPLi are interrelated, and since the latter is the
same in the two countries, nil Ki is also internationally identical. By
the same token ni/Li will also be similar internationally. Thus, in
whatever way we define them, we find that the rates of monopoly profit
in the two sectors will be internationally identical, provided the other
conditions usually assumed for the validity of the factor-price equalisa-
tion theorem are satisfied.
Product Market Imperfections 295

Finally, we examine the validity of the Stolper-Samuelson


theorem in the presence of monopolised product markets. The two
equations (11.13) and (11.14) can be solved to yield

w*-p! =
-[(1-err2)eKI +eK2err!AI +eK!err2A2] ( * *) (11.27)
lei P2 -p~

-[(I-erri)(}K2+eK2err!AI +eK!err2A2] ( * *)
w*-p! = lei P2- P! (11.28)

An inspection of these four equations reveals that real factor rewards


are uniquely related to the relative commodity prices. Suppose there
occurs a rise in the relative price of the second commodity, so that
(pi-p!) > 0. If 1e1 > 0, that is, if the first industry is labour-
intensive relative to the second, the wage rate declines and the
return on capital rises in terms of both commodity prices, because
w*- pj < 0 and r*- pj > 0. This is a clear demonstration of the
continued validity of the Stolper-Samuelson theorem in spite of the
presence of monopoly elements in the product markets.
A change in commodity prices will also manifest itself through
changes in monopoly profits via the change in the outputs and the
elasticities of demand. The desired expressions can be obtained by
totally differentiating ( 11.5) and then substituting for Xi* and sj.
However, we can follow a simpler route by simply analysing the
changes in the components that constitute the monopoly profits.
From (11.5), we know that rei = piXijsi. Let us first consider the
effects of a rise in the relative price of the second commodity on
monopoly profits in the same industry. Now

A rise in p causes a rise in X 2 , but since each demand elasticity is an


increasing function of its own relative price, it also causes a rise in
s 2 • Since both the numerator and the denominator of the above ex-
pressions rise as a result of a rise in p, the outcome on the real rate
296 Studies in the Pure Theory of International Trade

of profit in the second industry is uncertain in terms of both goods.


The same kind of fate awaits the position of the monopolist in the
first industry, where

Here a rise inp leads to a decline not only in X1 but also in e 1 , so that
once again we cannot predict how the rise in p will affect the real
income of the monopolist in the first industry. Thus we arrive at the
conclusion that a change in the relative commodity prices exerts a
determinate influence on the real incomes of the primary factors but
not on those of the monopolists.
The same is true of the relative returns of the monopolists, which
are now given by
7t2
-=p-•-
x2 el

7t1 xl e2
because with a rise inp, (X2 /X1) rises but (ede 2 ) declines, so that the
final result is indeterminate.

11.5 Monopoly and the Gains from Trade


The theory of gains from trade has kept company with us in almost
every chapter. We have demonstrated again and again that in the
presence of distortions, whether in the foreign trade sector or in the
domestic markets, free trade is not the optimal policy. The presence
of monopoly, which comes into the category of a domestic distor-
tion, constitutes no exception to this rule. Yet monopoly in the local
product markets gives rise to novel situations which make the dis-
cussion of the gains from trade in the present model command more
than merely academic interest.
First of all, there is the question of whether domestic monopolies
can exist in the face of foreign competition, for if the country in
question is relatively small, so that the commodity prices in the free
trade equilibrium are fixed, the monopolistic advantages to domestic
producers must disappear, for they no longer possess the power to
influence prices. Furthermore, since the consumers now face a given
set of prices, the elasticities of demand approach infinity and IX
approaches unity, so that from (11.26) the MRTcomes to equal the
international-price ratio. Under this setting, the introduction of
Product Market Imperfections 297

trade serves to eliminate the domestic distortion. Thus three types of


gains stem from free trade: the consumption gain, the production
gain and the gain from the elimination of the domestic distortion
because the M RT is no longer different from the commodity-price
ratio.
Things get really involved when both trading partners possess
monopolistic elements in their local product markets as well as in
the foreign trade market, so that the free trade prices can no longer
be taken as given. Several conceptual difficulties need to be resolved
before any progress can be made towards our objective. With the
existence of monopolies in both countries, each commodity is now
produced by two producers instead of one. Therefore in the pre-
sence of trade we are in effect faced with a situation of duopoly and
the multifarious solutions that present themselves under these
circumstances, for, as is well known, duopoly, unlike monopoly,
has numerous solutions, depending upon the assumptions we make
concerning the reactions of one producer to the expected actions of
the other. Melvin and Warne have a simple solution for this dilemma.
Following them, we assume either that each producer of a com-
modity behaves as a monopolist and takes its share of the world
market as the total demand for his own output, or production is
carried out by multinational firms which control the production
and distribution of a commodity in both countries. The analysis is
the same under any one of these cases.
There still remains the question of the pertinence of the elasticity
of demand when each commodity is being sold locally and abroad.
If we assume that the utility functions are similar internationally,
the elasticities of demand, being functions of commodity prices
only, will be the same in both countries under free trade. In autarky
equilibrium, the elasticities of demand pertaining to consumption
and production are the same. This is no longer tenable in the pre-
sence of trade, for the local output of each firm is different from the
total consumption of any commodity. Therefore the elasticities of
demand relevant to each country;s production point on its trans-
formation curve are the ones associated with the local or the world
consumption of the two commodities, rather than the elasticities
consistent with the local production. This point gains further
clarification when we consider the welfare implications of free trade,
to which we now turn.
298 Studies in the Pure Theory of International Trade

FP

Figure 11.3

Consider Fig. 11.3, where theM RT is shown to exceed the relative


price of the second good at all production points, reflecting the fact
that e2 > e 1 so that a > 1 (see equation (11.26) ). The home country,
the one under consideration, is assumed to import X 2 and export
X 1 • The self-sufficiency equilibrium is given by Sand the domestic-
price ratio by the slope of DP; the free trade price ratio is furnished
by the slope of FP, the production point shifts toP, the consumption
point to C, and the level of welfare rises from U 1 to U2 • Free trade
in this case is superior to no trade. At point S the elasticities of
demand facing the consumers and those relevant to production are
the same. As trade is introduced, the relative price of the first good
rises and so does its output and e 1 ; on the other hand the output of
X 2 and e2 declines, but the new elasticities of demand are the ones
consistent with the consumption point C and not with the free trade
production point P, simply because in the presence of trade local
consumption and production are not identical. Furthermore, in
computing the elasticities at P, account must be taken of the elasti-
cities of foreign supply or demand in addition to the elasticities at C.
It can be easily seen that the demand elasticities pertaining toP are
the weighted average of those associated with C and the elasticities
of foreign supply and demand, and there is no reason to suggest that
the two sets of elasticities will be the same. This is how the elasticities
Product Market Imperfections 299
accompanying the total consumption and local production levels
differ when trade is allowed. It may be emphasised here that since
the world consumption is equal to world production in equilibrium,
the elasticities associated with world output must be the same as
those associated with world consumption. This line of reasoning
suggests that under our assumptions concerning monopoly be-
haviour in the two countries and the international similarity of
homogeneous utility functions, the existence of the world equilib-
rium can be established in the same manner as we established the
existence of autarky equilibrium.
What we have shown up to now is that free trade is necessarily
superior to no trade, provided the country imports the commodity with
the higher autarkic elasticity of demand (e 2 > e 1 ). Let us now pro-
ceed to the other case where e1 > e2 , a case that is depicted in
Fig. 11.4, where the M RT is shown to be less than the relative price
of the second commodity. In general, the M RT is lower (higher)
than the relative price of the commodity with the lower (higher)
elasticity. As before, the introduction of free trade, arising from a
shift from the autarky price ratio DP to the foreign-price ratio FP,
takes production and consumption from StoP and C respectively.
As regards welfare, there are two possibilities. If the shift in the

x1

Figure 11.4
300 Studies in the Pure Theory of International Trade

production point is not very large, and/or if FP is much flatter than


DP, free trade will continue to be superior to no trade. For these
conditions will give rise to high consumption gain and low produc-
tion loss. In the contrary conditions, however, the production loss
may outweigh the consumption gain, in which case free trade may
be inferior to no trade or a prohibitive tariff. Such a possibility is
illustrated in Fig. 11.4, where welfare declines from U2 to U1 after
the introduction of free trade. Thus we conclude that if a country
exports the commodity with the higher elasticity (in autarky), free
trade may be inferior to no trade.
One may ask the question: how is FP determined? Partly, the
answer has already been given. Fig. 11.5, however, makes it crystal
clear. As stated earlier, demand elasticities are identical in the two
countries under free trade. The converse of this statement is also
true. Free trade prices are those at which demand elasticities are the
same internationally. Fig. 11.5 consists of two transformation curves,
of which H H' belongs to the home country and FF' to the foreign
country; in view of the international similarity of utility functions,
a common indifference curve U1 intersects these transformation
curves at Sh and S1 , the former being the autarky equilibrium point

x,

H' F'
Figure 11.5
Product Market Imperfections 301

in the home country and the latter in the foreign country. It is also
evident that ph, the autarkic relative price of the second good, which
is reflected by the slope of U 1 at Sh, exceeds pf, the corresponding
price in the foreign country, indicated of course by the slope of U 1
at Sf. Furthermore, in autarky e2 h > e2f and e 1h < e1f, where the
subscripts h and f refer as usually to the home and the foreign
country. With Ph > pf, the home country will import the second
commodity and export the first. Returning now to the original ob-
jective for which Fig. 11.5 was introduced, it is easy to see that as Ph
and pf move closer to each other when trade is introduced, (e 2 je 1h
and (e 2 /e 1 )f also shift towards each other and at some point are
equalised. Such a point is given byE in the diagram, and the slope of
U1 at E furnishes the world-price ratio FP. Indeed, if the world
elasticity ratio is known, the free trade price ratio can be computed
directly from (II. 7) and ( 11.8), because then

e2 aD+pPan
el I +aDpPan·

Let e be the elasticity ratio in the free trade equilibrium. Then


e = (aD+pPan)j(l +aDpPan)
_ [(]D-e ]liPan
whence p- --
eaD-1

which reveals that, once the world elasticity ratio is known,p can be
determined immediately because f3 and aD are given parameters.
This formula also suggests that aD can be equal neither toe nor to
lje, for otherwise p will be either zero or infinity. Furthermore, in
order to avoid the possibility of a negative p, aD is such that it exceeds
or falls short of both e and 1/e.

11.6 Concluding Remarks


The model described in the preceding sections has incorporated the
distortions arising from monopoly elements in the product markets,
while the distortions occurring in the factor markets, which were
explored in the previous chapter, have been ignored. It is heartening
to note that, unlike the case with factor market distortions, virtually
all the results derived from the undistorted two-sector model con-
tinue to hold in the presence of monopoly. Thus the theorems of
302 Studies in the Pure Theory of International Trade

Rybczynski, Stolper and Samuelson, and Heckscher and Ohlin all


continue to be valid in the present model, which is a very general
model in the sense that the competitive system springs from it as a
special case whe,re both the elasticities of demand equal infinity.
When trade is introduced, free trade still turns out to be the optimal
policy if the country with monopolistic product markets in autarky
faces a given set of world prices. In the large country case, however,
where monopolies exist in both countries, free trade, unlike in the
competitive system, may not be superior to no trade. The foregoing
results remain unchanged if the monopoly exists in one sector instead
of both, except that now the monopoly equilibrium cannot be
identical with the competitive equilibrium even by chance, for the
two elasticities can never be equal.

11.7 Appendix
Some of the equations presented in the text without proof will now
be derived.
Consider a utility function with a constant elasticity of substitu-
tion, av, where av = 1/(1 + {3), -I < {3 < oo. With two sectors in
the economy producing two goods, xl and x2, the utility function
can be represented as

(All.l)

where ajb is assumed to be unity for the sake of simplicity of cal-


culation. Given the constraint that Y = p 1 X1 +p 2 X 2 , we form the
Lagrangian function

Maximising, we obtain

az
- ~ (aX!fl +bX:Zfl)-lffJao( -af3X!lfao)-ply = 0 (All.3)
oX1

az
-!{3 (aX!fl +bX:Zfl)-11/lao( -bf3X21/ao)-P2Y = 0 (All.4)
ax2

az
Y -p1X1 -p2X2 = 0. (All.5)
oy
Product Market Imperfections 303

From (All.3) and (All.4) we have


aX1tlav(aX1fJ +bXifJ)-tlfJav = PtY
bXit/av(aXifJ+bXifJ)-11/Jav = P2Y·

Dividing, using (All.5) and solving for X 2 , we obtain

(All.6)

(All.7)

where, as in the text, p = p 2 /Pt·


From these demand functions, the 'own' price elasticities are
derived as follows:
Differentiating (All.6) partially with respect to p 2 , we have

- Y[I +crv(P2)-fJ" 0 (p 1 )fJ" 0 ]


(All.8)
[p2 + (p2)" 0 (Pt)/J"n] 2

whence
P2[l +crv(P2)-/J"n(Pt)/J"n]
P2 +(p2)"D(Pt)fJ"D

so that

Similarly,

It may be observed that, in the derivation of demand elasticities,


we have assumed that money national income, Y, is constant at all
points on the transformation curve. The validity of this procedure
stems from the fact that, for any product-price ratio, individual
prices can always be so adjusted as to keep income unchanged. The
elasticities will be unaltered, because they depend only on the price
ratio.
304 Studies in the Pure Theory of International Trade
REFERENCES
[1] Batra, R. N., 'Monopoly Theory in General Equilibrium and the Two-Sector
Model of Economic Growth', Journal of Economic Theory, IV (June 1972) 355-71.
[2] Melvin, J. R., and Warne, R. D., 'Monopoly and the Theory of International
Trade', Research Report No. 7032, Department of Economics, Univ. of Western
Ontario, London, Ontario.
12 Non-Traded Goods

Every country produces goods which cannot be traded at all either


because of the nature of the goods, like houses, services, etc., or
because of political barriers preventing, for example, the export of
certain strategic military equipment, or because of artificial trade
barriers like prohibitive tariffs. In all the preceding chapters we
ignored the presence of such products, much at the expense of
economic reality. The objective of this chapter is to examine how
and to what extent the results derived from the two-traded-goods
model are modified when a non-traded final good is incorporated
into the framework.
To the two traded goods, X1 and X 2 , assumed so far, we add
another, non-traded good, X", the demand for which always equals
local supply. Thus we now have a three-good, two-factor model;
this non-equality of the number of goods and factors introduces
some very unexpected problems which must be satisfactorily re-
solved before we can proceed direct to some of the theorems
derived before. The rest of the assumptions continue to be the same
as those of the standard two-sector model. That is to say, at least
for the time being, we ignore the complications arising from market
imperfections which were the subject of analysis in the two previous
chapters.

12.1 The Model with Non-Traded Goods


In the presence of an additional good, the full-employment relations
are given by

(12.1)
and

(12.2)
306 Studies in the Pure Theory of International Trade

where, as before, cij is the amount of the ith input divided by the
jth good (i = L,K; j = 1,2,n), for example, CLn = (Ln/Xn). The
price equations, on the other hand, now become
CL1w+CK 1 r = p 1 (12.3)
CL2w+CK2r = P2 (12.4)
and CLnw+ CKnr = Pn (12.5)
where Pi is the price of the jth good. As before, the input-output
coefficients can be solved from the expressions relating them to
factor prices. But we still have one too many unknowns, for there
are five equations and six unknowns, X 1 , X2 , Xn, w, rand Pn· The
system can be given a determinate solution, however, by introducing
another equation concerning the demand-supply equality for Xn.
Then
(12.6)
which shows that the local supply of the non-traded good equals its
demand, which in turn depends on p, q and Y, where p and q are,
respectively, the relative prices of the second and the non-traded
good in terms of the first, and where Y is national income. This gives
us a completely determinate system of six equations and six un-
knowns, given that the production coefficients can be solved from
the factor prices.
The solution of the model with the non-traded good proceeds by
first solving for w and r from the two price equations (12.3) and
(12.4) in terms of the input-output coefficients and p 1 and p 2 , which
are determined internationally. The values of w and r so obtained
can be plugged in (12.5) to solve for Pn in terms of the production
coefficients of all three commodities. Since production coefficients
themselves are functions of wand r, we get a determinate solution
for factor rewards, the price of the non-traded good and capital/
labour ratios in all three goods, once the values of the parameters
p 1 and p 2 are internationally determined. Note that this solution is
obtained independently of demand conditions and factor endow-
ments. If the country in question is a small country and hence faces
given prices of traded goods, the demand conditions and factor
supplies exert no influence on factor prices and prices of non-
traded goods. If in addition we assume that production functions
are similar internationally, wand r will also be the same in the two
countries. This follows clearly from the solution of (12.3) and
Non- Traded Goods 307

(12.4). Furthermore, if the international identity of production func-


tions extends also to the domestic goods, the prices of these goods
will also be the same everywhere in the world in spite of the fact that
these goods are not traded. The validity of this result does not
depend on the number of domestic goods.
Until now our diagnosis has proceeded on the side of price equa-
tions and factor rewards. Let us now explore the system from the
side of factor endowments and outputs and see what new features
are introduced by the presence of a third, domestic good X". It is
worth emphasising at this point that many of the new properties of
the model stem chiefly from the inequality of the number of goods
and factors and not from the fact that one of the goods is purely a
domestic good. First of all, when there are three goods and two fac-
tors, we should expect indeterminacy in the system if all goods are
traded, the reason being that in such a model the production possi-
bility surface becomes a ruled one like ABC in Fig. 12.1, for any

x,

Figure 12.1
308 Studies in the Pure Theory of International Trade

amount of output of a commodity is compatible with a given set of


prices. This can be seen most readily by solving for X 1 and X 2 in
terms of L, K and X, from (12.1) and (12.2) to obtain

X _ CL2L(k2-k)+CL2CLit(kn-k2)Xn (12.7)
1 - ICI
X _ CL1L(k-k1)+CL1CL1t(k1-kn)Xn (12.8)
2- ICI
where, as usual, k1 is the capital/labour ratio in the jth good
(j = 1,2,n) and ICj is the determinant of the matrix of production
coefficients involved in X 1 and X2 , that is, ICI = C" 2CL 1-CL2CK1·
If commodity prices are constant, (12.3)-(12.5) show that factor
prices and production coefficients will remain unaltered. But this is
not sufficient to ensure that X 1 and X 2 will also be unaltered so long
as Xn is positive and all capital/labour ratios differ. For then the out-
put of one or both goods will vary with the variation in Xn even
though commodity prices were invariant. Thus when the number of
goods exceeds the number of factors, any level of commodity out-
puts is consistent with a given set of prices. The production possi-
bility surface is then described by straight lines (as in Fig. 12.1) and
each line is a locus of possible output configurations for the three
commodities corresponding to a given set of commodity prices. No
wonder, then, that when all three goods are traded, the system is
indeterminate, for the specification of international prices is insuffi-
cient to determine the output levels. The characterisation of one of
the goods as non-traded, however, solves the problem. For suppose
that the output of X, is fixed at some level furnished by domestic
demand; then X 1 and X 2 can be easily obtained from (12.7) and
(12.8). Thus there is no indeterminacy if one of the three goods is
non-traded, although the production possibility surface continues
to be a ruled surface.
A richer understanding of these issues emerges from an examina-
tion of Fig. 12.2, which is constructed on the basis of an ingenious
geometrical technique discovered recently by Melvin [5]. Fig. 12.2
comprises two boxes, one given by 0,.0 2 and the other by 0 1 0 2 •
There are three origins, each representing one commodity; for On
represents X,, 0 1 represents X 1 , and so on. The output level of each
good is given by the distance of the production point from its origin.
Suppose the output of Xn is fixed at the level 0 1 0,., then the remain-
Non-Traded Goods 309

Figure 12.2

ing factor endowments can be allocated between the two traded


goods in such a way as to trace a contract curve constrained by factor
supplies equal to (K- K,) and (L- L,). Such a contract curve be-
tween X 1 and X 2 is given by 0 1P0 2 • Let P be the actual production
point, then the output of X1 equals 0 1 P and that of X 2 equals 0 2 P,
with their capital/labour ratios given by the slopes of 0 1 P and 0 2 P,
respectively. Thus once the output of the domestic good is given, the
other two outputs are readily determined.
Suppose that, instead of 0,.0 1 , the output of X,. is fixed at o,.o;;
draw o;r parallel to 0 1 P. The diagram makes it clear that even
though the capital/labour ratios and hence the commodity prices
are constant, the output of X1 and X 2 has declined when X, is fixed
at a higher level. In other words, there is no unique relationship be-
tween commodity outputs and prices when goods exceed factors.
We will discover later that, for this reason alone, the output of the
traded good may actually decline in response to a rise in its relative
price.

12.2 The Price-Output Response of Traded Goods


In order to obtain the expressions for the price-output response of
traded goods, the system of equations must be differentiated totally,
310 Studies in the Pure Theory of International Trade

to obtain
A.LIXi+A.L2X! = L*-(A.LIC!t +A.L2Ct2)-A.Ln(Ct,.+X:) (12.9)

(12.11)

(12.12)

(12.13)

and Xn* = En2P*- Enq* +'In Y* (12.14)


where £..2 = (p/Xn)(oXnfop),
En = -(q/Xn)(oXnfoq),
'In= (Y/X,.)(oXnfoY)

and where, in obtaining (12.11 )-(12.13), use has been made of the
cost-minimising condition OLiCti+ OKic:i = 0, with O's as before
denoting the relative share of the relevant factor. In (12.14), En 2 is
the cross-elasticity of demand between the domestic and the second
good, En is the price elasticity and 'In the income elasticity of demand
for the non-traded good. From (12.11) and (12.12)

IOI(w*-r*) = (pf-p!) = -p*


and from (12.11) and (12.13)

IBnl(w*-r*) = (pf-p~) = -q*

whence
q*
p* =wJOnJ (12.15)

where JOI = lOLl eKtl and IBnl = lOLl eKll


0L2 0K2 (}Ln (}Kn

and JOJ ~ 0 if k 2 ~ k 1 and JOn! ~ 0 if kn ~ k 1 . From (12.15) it is


clear that the effect of a change in the prices of traded goods on the
price of the domestic goods depends on the signs of JOJ and JOn I or
on the capital/labour ratios in the three industries. As usual, Cti
and c:i can be obtained by solving simultaneously the equations
Non-Traded Goods 311

representing the elasticity of factor substitution and cost-minimising


condition in each sector.t Thus
C"t = -lJ"iuJ(w*-r*)
and c:.j = (}LjUJ(W*- r*).
Substituting these expressions in (12.9) and (12.10), assuming that
L • = K* = 0, and solving them yields

X • _ ()..K2pL + ).L2pK)(w*- r*) + x:().L2).K,- ).K2).Ln)


1 - 1).1
X* _ - ().L1pK + )." 1Pd(w*- r*) + X,*().K1).L,-).Ll).K,)
and 2 - 1).1
where 1).1 = I).Ll ).L21
).K1 ).K2
PL = ).L1(}K1U1 +).L2(}K2u2+).Ln(}K,u, > 0
and
PK = ).K1 (}Llu1 + ).K2(}L20'2 + ).K,(}Lnu, > 0.
Substituting for (w*- r*) in these expressions, we get
X* _ -().K2pL +AL2pK)p* 1).,21 X,*. •
(12.16)
1 - 1;.1101 + 1;.1 p• P
x• _ ().LlPK+).K1pJp* + 1;.,11 X,• •
and (12.17)
2 - IAIIOI 1).1 p* p
where l).n21 = ().L2).K,-).K2).Ln) ~ 0 if k, ~ k2
and l).n11 = ().K1).L,-).Ll).K,) ~ 0 if k1 ~ k,.

These expressions confirm what has already been anticipated,


namely, that there may not be a unique relationship between prices
of traded goods and their outputs. Consider (12.16), for example.
If x,• = 0, then Xrfp* < 0 as usual because 1).1101 > 0. However,
if x,• =I= 0, the last expression may have a sign opposite to that of the
first, in which case X1*jp* may not be negative. The change in p
affects the demand for the domestic good, which in tum leads to a
change in its output and hence in the amount of resources available

t This was accomplished in Chapter 2, and the expressions for Cl1 and Cl1 derived
there remain unchanged in the present model because each good is still produced with
the help of two factors.
312 Studies in the Pure Theory of International Trade

to the traded goods. Suppose that all goods are gross substitutes, so
that (X"* jp*) is positive. We still have to reckon with the sign of lA. I
and IA.n 2 l. Suppose IA.I > 0, so that k 2 > k 1 ; if IA.n 2 l is also positive,
so that k" > k 2 , then (X.*fp*) may not be unambiguously negative.
On the other hand, if the domestic and the second good are com-
plements, so that (X"*/p*) < 0, then with both 101 and IA.n 2 l > 0,
(Xt /p*) is unambiguously negative and the price-output response
in the first industry is normal. Similar considerations apply to the
price-output response in the second industry. The gist of this dis-
cussion is that, in the presence of purely domestic goods, the price-
output response of the traded goods is no longer predictable. The
outcome depends not only on the substitution effects on the produc-
tion side, but also on the nature of the demand for three commodi-
ties as well as their relative factor intensities.
This result generates far-reaching ramifications for several results
which we derived in the chapter on the theory of nominal tariffs.
First, the offer curve may not only be negatively sloped at some
volume of trade, but may also bend back towards the origin. In other
words, the price elasticity of demand for imports may no longer be
negative. This factor tends to create instability in the model. Second,
the introduction of the tariff may result in a rise in the import
demand at constant terms of trade, and hence eventually cause a
deterioration in the terms of trade of the tariff-imposing country. t

12.3 The Stolper-Samuelson Theorem


The effect of protection on factor rewards is straightforward in a
two-good model. In our three-good model, however, the logic be-
comes more involved, for a rise in the price of the importable good
entails a change in the price of the non-traded good. Is it then
possible that factor rewards moving unambiguously in terms of the
prices of traded goods may not do so in terms of the prices of purely
domestic goods? The answer turns out to be a definite no. This can

t It is true that, in the presence of the domestic good, the demand for the import-
able good is also influenced by the price of X. which may rise or decline as a result of
the tariff, and this factor, absent in the usual two-good model, may strengthen or
attentuate the resultant decline in the demand for the importable good. Nevertheless,
in the absence of the categorical relationship between the output of X2 and the tariff-
induced rise in its local relative price, it is no longer certain that the rate of tariff and
the demand for imports are negatively related when world prices are kept constant.
For this reason, tariffs may or may not cause an improvement in the terms of trade
of the tariff-imposing country. For further discussion, see Komiya [ 4].
Non- Traded Goods 313

be seen immediately by solving for w• and r• from (12.11) and


(12.12) to obtain

w• = (12.18)

• (}Ll
r = TOTP2• (12.19)

where we assume that the tariff-imposing country is small, so that its


tariff on X2 leaves p 1 unchanged. Let 101 > 0, then w• < 0 and
r• > O.Furthermore,(OLl/IOI) > 1,sothatr• > pf.Inotherwords,
the wage rate declines and the return on capital rises in terms of the
second good and also in terms of the first good, because PT = 0.
Now a decline in wand a rise in r will lead to a change inpn. How-
ever, since from ( 12.13) the change in Pn is a weighted average of the
changes in wand r, the weights being (}Lit and (}Kn which sum up to
unity,
r• > p:
> w•.
This implies that r rises and w declines in terms of the domestic good
also. In other words, factor rewards move unambiguously in terms
of all commodities including the non-traded good as a result of the
tariff. What is interesting is that since Pn varies with p 2 , a rise in Pn
will worsen and a decline inpn will improve the real income position
of both factors in comparison to what their positions would be in the
absence of the domestic good.

12.4 The Rybczynski Theorem


The implications of a change in factor supplies for output levels have
kept recurring in several previous chapters. This one is no exception.
In exploring the effects of changes in factor supplies on the two out-
puts at constant terms of trade, we have to approach the problem in
a slightly different manner. A rise in factor supplies culminates in a
rise in national income when terms of trade are unchanged. This
raises the demand for all products including the domestic good. To
match the increase in demand, the output of the domestic good must
rise to the same extent. In other words, a part of the additional factor
supplies must be absorbed in the domestic good. Thus we conclude
that whatever the direction of change in the output of the traded
goods, the output of the non-traded good must rise as a result of the
increase in factor supplies, when terms of trade are kept constant.
314 Studies in the Pure Theory of International Trade

As regards the traded goods, the change in their outputs can be


deciphered by solving for X1* and X2* from (12.9) and (12.10), while
remembering that at constant commodity prices (w*- r*) = 0. Thus

X* _ (2K2L* -2L2K*)+(2L22Kn-2K22Ln)Xn*
(12.20)
1 - 121

(12.21)

From (12.6) we know that when terms of trade are unaltered,


X"*= '1nY*. WithY= wL+rK,and with w* = r* = 0, Y* = fhL*+
()KK*, where ()Land ()K denote respectively the share of labour and
capital in national income. This means that
Xn* = '7n(()LL*+8KK*).
Substituting this in (12.20) and (12.21) yields

I21Xf = [2K2(1-'7n(JL2Ln)+'7n(JL2L22Kn]L*
- [ AL2(l- '1n0K...1.Kn) + '7n0K...1.K2ALn]K*

IA.IX! = [2Lt0-'7n()K2Kn)+'7nOKA.Kl2Ln]K*
- [2Kl(I-PJn()LA_Ln)+PJn()L2LlAKn]L*

Now

where m" is the marginal propensity to consume the domestic good


and lies between zero and unity. With so much information at hand,
the effects of changes in factor supplies on X1 and X2 can be easily
derived. Suppose K* = 0. Then with IA.I > 0, that is, with k 2 > k 1 ,
XNL* > 0 and Xt/L* < 0. In other words, an increase in the
supply of labour alone raises the output of the labour-intensive
traded good and lowers the output of the capital-intensive traded
good, when terms of trade are unaltered. The increase in the supply
of one factor alone then leads to an ultra-biased growth. This is
nothing but a straightforward generalisation of the Rybczynski
theorem to the model incorporating the non-traded good. Similarly,
for L* = 0, (Xf/K*) < 0 and X~./K* > 0 when 121 > 0.
Non-Traded Goods 315

12.5 The Gains from Trade Once Again


Most of the results derived in Chapter 4, which was concerned with
the analysis of gains from trade, did not ignore the presence of purely
domestic goods, except the theorem comparing the welfare implica-
tions of higher versus lower tariffs. There we promised a reappraisal
of this theorem in the present chapter. The reason must be palpable
by now, for, as we have seen above, the relationship between the
demand for imports and the rate of tariff is no longer monotonic,
that is, (d£2 /dt) may be negative, positive or even zero. Since the
sign of the change in welfare is controlled solely by the sign of
(d£2 /dt), the effect of an increase in the tariff rate on welfare becomes
ambiguous when purely domestic goods are introduced in the
model.t
Next, we turn to the question of the optimal policy in the presence
of purely domestic goods when factor markets are distorted by the
presence of inter-industry wage differentials. The choice of the dis-
tortion is of course arbitrary, but the results derived below permit
application to any kind of distortion.
In the presence of three goods, the problem may be formulated as
one of maximising the social utility function
U = U(D1,D2,Dn)
subject to the constraints
D 1 = X 1 -E1 = X 1 -pE2 (12.22)
D 2 = X 2+E2 (12.23)
and
(12.24)
where D" is the consumption of the non-traded good and £ 1 = p£2
from the balance-of-payments constraint. On the production side,
(j = 1,2,n) (12.25)
(12.26)
w1 w2 w"
and -=-=-
(X 1 (X2 (Xn

t See the expression for (l/U1 )(dUfdt) in Chapter 4. For a full discussion of gains
from trade in the presence of non-traded goods, see Batra [I].
316 Studies in the Pure Theory of International Trade

or (12.27)

where w1 is the wage rate in the jth sector and where rx1 ~ l is a
constant. With no wage differential, rx 1 = rx 2 = rx 3 . In the presence
of the differential, however, the following possibilities arise:
l. rx 1 = rx 2 < rx,., and the differential is paid by the non-traded
good.
2. rx 1 = rx, < rx 2 , and the differential is paid by the import-
competing good.
3. rx 2 = rx,. < rx 1 , and the differential is paid by the exportable
good.
In addition to these, there are six more possibilities if the differential
exists among all the industries. These are:
rx 1 ~ rx 2 ~ rx,. and w1 ~ w2 ~ w,;
rx 1 ~ rx,. ~ rx 2 and w1 ~ w, ~ w2 ;
rx 2 ~ rx 1 ~ rx,. and w2 ~ w1 ~ w,.
For expository purposes, we rewrite the full-employment equations
as
L 1+L 2+L, = L and K 1+K2+K, = K.
Our first task here is to obtain an expression concerning the relation-
ship between the outputs of all three industries, something similar
to the slope of the transformation curve in the two-good case.
Totally differentiating (12.25) and dividing through by dX2 , we
obtain
dX1 FK1dK1 +FL1dL1
(12.28)
dX2 = FK2dK2+FL2dL2.
From the full-employment equations
dL 1+dL 2+dL, = 0 and dK 1+dK2+dK, = 0.
Using these and (12.26) and (12.27), (12.28) can be written as
dX1 = -p[FK2(dK2+dK,)+(rxdrx2)FL2(dL2+dL,)J
dX2 FK2dK2 + FL2dL2

(12.29)
Non- Traded Goods 317

where fJ = FK2dK2+ (a.t!a.2)FL2dL2 ~ l


F1udK2+ FL 2dL 2

according as (a.tfa. 2 ) ~ l. In the analogous manner

dXn = J!. [FK2dKn+(a.nfa.2)FL2dL"]


dX2 q (FK2dK2 + FL2dL2)
= J!. [FK2dKn + (a.1ja.2)FL2dL,.J
q y(FK2dK2+FL2dL2)

where ~ = FK2dKn+(a.n/a.2)FL2dL,. ~ l if
y FK2 dKn + (a. tfa. 2)FL2 dLn

or y ~ l , according as (a. t1a.,.) ~ l.


It may now be seen that the second term within the brackets of
(12.29) equals (yqjp)(dXn/dX2). Hence

dX1 = -p[fJ+ yq. dX"]


dX2 p dX2
or dX1+ {Jp dX2 + yq dX,. = 0. (12.30)

Two special cases deserve further consideration. In the absence of


the differentials, fJ = y = I, so that (I2.30) reduces to dX1+ pdX2+
qdX,. = 0. If, in addition, there is no non-traded good, we arrive at
the familiar relationship that (dXtfdX2) = -p.
Reverting to our original objective for deriving (I2.30), let us
totally differentiate the utility function and the constraints given by
( I2.22)-( 12.24), to obtain ~ = dX1+ pdX2 + qdX". Substituting for
dX1 from (12.30) in this expression then yields

dU
U1 = pdX2(l-fJ)+qdX,.(I-y). (I2.3I)

As usual, we have assumed that the initial situation is one of free


trade and that the country in question is a price-taker. This last
equation confirms the result that in the absence of the wage differen-
tial, so that fJ = y = I, free trade continues to be the.first-best policy.
318 Studies in the Pure Theory of International Trade

Let us now consider the case where factor markets are distorted
but the domestic goods do not exist, so that (12.31) becomes
dU = pdX2(I- {3).
ul
Here {3 reflects the divergence between the marginal rate of trans-
formation and the international price ratio. As expected, the optimal
policy in the presence of the differential requires a change in X 2 such
that the divergence between the MRT and the given international
price ratio is eliminated, so that the value of {3 to producers comes
to equal unity. The second-best policy in other words requires the
imposition of production tax-cum-subsidy on the second andjor
the first commodity. The first-best policy of course consists in the
elimination of the wage differential to the producers by means of the
grant of a factor tax-cum-subsidy, such that (rxtfrx 2 ) and {3 are
equated to one. None of these results is new, but they have been
reproduced to present a contrast to the results to be derived below.
It is noteworthy here that these results are not modified even if the
non-traded good exists, so long as y = I or rx 1 = rx., so that wages
are the same in the exportable and the purely domestic-good
industries.
Next, consider the case where the non-traded good exists and the
wage differential is paid by either X 2 or X •. In the first case
rx 1 = rx. < rx 2 , which means that y = I but {3 < I. Here again the
second-best policy calls for a production tax-cum-subsidy to be
granted to the traded goods. In the second case where the differen-
tial is paid by the non-traded good, rx 1 = rx 2 < rx. and y < 1 but
{3 = I, so that the change in welfare reduces to

dU = qdX.(l- y).
ul
Here the second-best policy calls for a consumption subsidy to be
granted to the domestic good, such that the divergence to the pro-
ducers between - (dXtfdX2 ) and p[l + (qjp)(dX./dX2 )] is eliminated,
that is to say, the effective value of y to the producers is equalised to
unity. Note that this policy does not apply to the traded goods. Since
the distortion is in the domestic production sector of the economy,
the thrust of the cure should also lie in removing the inequality be-
tween -(dXtfdX2 ) and p[l +(qjp)(dX./dX2 )], just as in the two-
good model the second-best optimum is achieved by changing the
output of X1 and X 2 in such a way that the difference between
Non- Traded Goods 319

-(dXtfdX2 ) andp is eliminated. If the consumption tax (or subsidy)


were to be imposed on one of the traded goods, there will not be any
change in outputs, for p, which is exogenously determined, will
remain unchanged for the producers. However, in the case of the
non-traded good the domestic output of X" equals its domestic con-
sumption. Therefore, if the wage differential is paid by X" producers,
the second-best optimum will be attained only by increasing its out-
put to such an extent that to its producers y comes to equal unity.
However, in order to raise the output of X" without creating a dis-
crepancy in its domestic demand and supply, a consumption subsidy
to the non-traded good, rather than a production subsidy, is
required. It may also be noted from (12.31) that when}' < l and
f3 = l, so that the wage differential is paid by the non-traded good,
a positive sign of dD" = dX" will raise welfare, because (dU/U1 ) is
then positive.
On the other hand, if a production subsidy was granted to the
non-traded-goods industry, the rise in its output will not necessarily
be matched by the rise in its demand. Indeed, after the grant of the
production subsidy which keeps the prices unchanged to the con-
sumers, the demand-supply equilibrium for the non-traded good
will be maintained only through coincidence. There is nothing in the
model to guarantee this. The value of production at given world
prices will rise as the production loss arising from the wage differen-
tial is eliminated through the provision of the requisite amount of
the production subsidy to X". This by itself would lead to a rise in
D" even if prices to the consumers are constant, but there is nothing
to guarantee that the rise in D" will match the initial rise in X". On the
other hand, if a consumption subsidy was conferred on the non-
traded good, D" could be raised to the required level of X", and the
increase in the output of the domestic good would follow auto-
matically, even if prices to the producers were unchanged, for, as
demonstrated before, any output configuration is attainable with a
given set of prices in our model with three goods and two factors.
Let us now consider the case where the differential is paid by the
X 1 producers. Here ex 2 = ex" < ex 1 , so that both f3 andy exceed unity.
An interesting but a chance result emerges here. It is possible for
(dUjU 1 ) to be zero, if from (12.31)

q (l-y) dX2
(12.32)
p(l - /3) = dX"
320 Studies in the Pure Theory of International Trade

in which case free trade alone is the second-best policy and the im-
position of the production tax-cum-subsidy is not needed. Although
this result is purely of academic interest, for only by coincidence will
y and f3 be such as to satisfy (12.32), yet the condition does suggest
an interesting possibility. In any case, if the difference between the
left- and right-hand sides of(l2.32) is small, the gain to be derived
from the introduction of the production tax-cum-subsidy may not
be more than negligible and may not be worth the administrative
headaches and costs. A necessary condition for (12.32) to be satis-
fied is that (dX2 /dX,) be negative, because both y and f3 exceed unity
when the differential is paid by X1 producers. The chances of(l2.32)
being satisfied improve if the wage differential exists among all three
industries; nor is it necessary any more that (dX2 /dX,) be negative,
because the differentials may be such that y ~ I when f3 S I. The
essential point is that there exists a certain configuration of wage
differentials which will satisfy (12.32). Nevertheless, in all the cases
discussed in this section, the first-best policy requires the grant of
factor tax-cum-subsidy so that the differential to the producers is
eliminated.
An interesting policy device to attain the second-best welfare sug-
gests itself from this discussion. Suppose the level of the inter-
industry factor-price differential itself depends on government
policy. For example; Johnson [3], referring tQ Harberger [2], has
suggested that the inter-industry factor-price differential will occur if
the government imposes a corporation income tax (which is a tax on
the use of capital in the corporate sector alone) that drives a wedge
between the gross rewards of capital in the corporate and the non-
corporate sector. If the government must impose the corporation
income tax in order to raise revenue, etc., then our argument sug-
gests that it should be imposed in such a way as to satisfy condition
(12.32). Although y and f3 represent the wage differentials among
the three industries under the assumption of similar reward of
capital in all industries, the essential nature of our analysis is un-
modified in the case where the rewards of capital differ among in-
dustries but the wage rates are similar. Depending on the situation,
the level of the tax on the use of capital could be determined in the
following manner:
(i) The entire revenue can be raised from taxing the use of
capital alone in the first commodity. If the other two in-
Non- Traded Goods 321

dustries constitute the non-corporate sector, this tax can


be passed on as the corporation income tax.
(ii) The use of capital can be taxed in the second and the non-
traded-good industries, but in order to attain the second-
best optimum the tax rate will have to be different in the two
industries. This may require taxing the use of capital in the
corporate as well as the non-corporate sector. The need for
different tax rates arises from the need to keep both p and y
from unity. This enables us to hazard a conjecture that the
single corporation income tax rate on all incorporated in-
dustries in vogue in some countries (such as the U.S.,
Canada, etc.) might have had distortionary effects on the
second-best level of social welfare.
(iii) Finally, the differential rate of tax on the use of capital can
be imposed on all three industries. It may be emphasised
again that the tax-rate differential should be such as to select
those levels of p andy that satisfy condition (12.32).
Thus there are several revenue-generating measures at the dis-
posal of the government which will satisfy the dual objective of
raising the revenue without lowering the level of welfare below the
second-best level.

REFERENCES
[I] Batra, R. N., 'Non-Traded Goods, Factor Market Distortions and the Gains
from Trade', American Economic Review (Sep 1973).
[2] Harberger, A. C., 'The Incidence of the Corporation Income Tax', Journal of
Political Economy, LXX (June 1962) 215-40.
[3] Johnson, H. G., 'Factor Market Distortions and the Shape of the Transformation
Curve', Econometrica, XXXIV (July 1966) 686-98.
[4] Komiya, R., 'Non-Traded Goods and the Pure Theory of International Trade',
International Economic Review, VIII (June 1967) 132-52.
[5] Melvin, J. R., 'Production and Trade with Two Factors and Three Goods',
American Economic Review, LVIII (Dec 1968) 1249-68.
13 International Investment

The barter theory of international exchange remains incomplete


until the complications arising from the presence of international
investment are explored in our standard two-good, two-factor
model. This is the task assigned to this chapter. In all, this chapter
is concerned with two burning issues. First, it turns out that under
certain conditions free international movement of capital is a perfect
substitute for free inter-country movement of goods. In other words,
the hitherto assumed absence of international immobility of factors
turns out to have been made not for convenience but for ensuring
that trade will in fact take place among countries. However, the
conditions under which capital movement becomes a perfect sub-
stitute for goods movement are quite stringent, and it is not at all
unlikely to encounter the presence of international movement of
both goods and capital. This gives rise to our second question,
namely, what are the optimal policies in the presence of trade and
international investment? The model described below is a natural
extension to the case of international investment of the two-good,
two-factor model that we have been using in the past several
chapters.

13.1 Capital Mobility as a Substitute for International Trade


Mundell [5] was the first to demonstrate rigorously that under con-
ditions of factor-price equalisation the same international equilib-
rium can be achieved by either free trade or the unimpeded mobility
of capital. In other words, if trade is distorted by tariffs, the uninter-
rupted international flow of capital will supersede trade and leave
unimpaired the efficiency of /aissez-faire. In this chapter, however,
we shall follow the demons~ration given by Rakowski [7], who not
only derives Mundell's result under less restrictive assumptions,
but also points ouf the serious difficulties that normally stand in the
way of capital mobility becoming a perfect substitute of free trade.
International Investment 323

Before we proceed to the demonstration, a few remarks concern-


ing the concept of international capital mobility are in order, if only
to avoid the possibility of confusion in the subsequent analysis. By
international mobility of capital is meant the physical movement
of the capital goods in response to different rentals of capital in the
two countries. These goods are not subject to any tariffs or transport
costs. Furthermore, the owners of capital goods remain in their own
country and can consume only the goods that are located there.
The consumption of the capital owners depends on the marginal
product of capital in any good which, of course, can be exchanged
for the equivalent value of the other good. In addition, we assume
that labour is still immobile internationally and that the tariffs and
taxes are imposed only by the home country.
In a world where capital is mobile, equilibrium is achieved only
when the marginal product of capital (MPK) in any commodity is
the same in the two countries. For example, consider the case where

but

where the subscript/stands for the foreign country and distinguishes


the variable from that in the home country. Now since the marginal
product of capital in the second good is higher at home than abroad,
foreign capital will move to the home country free of cost and will
be employed in the production of the second good until MPK2 h
and MPK2f are equalised. Hence in equilibrium,

and (13.1)

Once this is recognised, it is a relatively simple matter to show that


goods mobility and capital mobility are perfect substitutes, provided
-and this is a crucial point- goods obtained through the repatriation
of capital earnings are not subject to any tariffs. In our barter model,
where money is absent, the repatriation of capital earnings requires
the movement of goods, and if such goods are exempt from tariffs,
it is clear that the introduction of any tariff would lead to the
absence of all trade in goods that come under the purview of taxation.
324 Studies in the Pure Theory of International Trade

For the sake of illustration, let us assume that the home country
imposes a tariff (t11) on the imports of the second good, so that
p, = p(l +t,)
where, as before, p is the terms of trade and p, the domestic-price
ratio in the home country in terms of the first good. However, in
the presence of the tariff the capital market cannot remain in
equilibrium. For example, let
MPK 111 = MPK 11 (13.2)

that is, let the return to capital in the first industry be the same in
the two countries. However, when exchanged for the equivalent of
the second good, MPK 111 = p,MPK211 and MPK 11 = pMPK21 , so
that in view of (13.2)
(13.3)

Since, in the presence of the tariff, p,. > p, it follows from (13.3)
that

which evidently violates the equilibrium condition (13.1). For (13.1)


to be satisfied, it is clear that p, must equal p, which is impossible
so long as the tariff is positive and operative. The only way in which
p,. can be equal top in the presence of the tariff is that there occurs
a complete absence of international trade, so that the tariff is in-
operative. In other words, any impediment to free trade must lead to
no trade when capital is internationally mobile. It is in this sense
that free trade and international capital mobility are perfect sub-
stitutes. Free international movement of either goods or capital
leads to the equality of commodity prices in the trading countries.
It may be noted here that this result does not depend on the con-
ditions necessary to ensure international factor-price equalisation.
Thus, for example, even if one or both countries are completely
specialised, so that factor-price equalisation is not possible, the
result derived above can be established. In this spirit, suppose that
the home country is specialised in the first good and the foreign
country in the second. A unit of capital invested in the home country
earns MPK 111 of good I or MPK 111 /p 11 of the second good. Similarly,
International Investment 325

a unit of capital invested in the foreign country earns MPK21 of


the second good or pMPK21 of the first. All this implies that
MPK 111 /p11 = MPK 21
and
MPK 111 = pMPK21
which can simultaneously hold only if p = p,. Thus even if both
countries are completely specialised and factor-price equalisation
fails to occur under free trade, any tariff must eventually become a
prohibitive tariff if the capital market is to remain in equilibrium.
Note further that in arriving at this result we did not have to make
the assumption of international identity of production functions,
as done by Mundell.
Although Mundell's original proposition can be derived from
much weaker conditions, there still remains one restrictive con-
dition which is crucial for capital mobility to be a substitute for
goods mobility and which is very unlikely to be satisfied in the real
world. The requirement that goods moving across the countries
through the repatriation of foreign earnings are not subject to the
home country's tariff is really a special one and is unlikely to be
satisfied. Thus goods mobility and capital mobility are not perfect
substitutes, for capital mobility gives rise to some level of goods
mobility. Nevertheless, Mundell's weaker proposition that an
increase in trade implements results in increased capital movements
is still valid simply because such an increase would normally raise
the inter-country differential in the earnings of capital.

13.2 Optimal Tariffs and Taxes


In the preceding section we have seen that, since capital and goods
mobility are not perfect substitutes, international trade and foreign
investment can coexist in equilibrium. This gives rise to several
interesting questions concerning optimal policy facing a large or a
small country in a world where movement takes place in both goods
and capital across the countries. The presence of foreign investment
introduces another variable in the model, and the link between the
earnings of capital and the world terms of trade introduces an ele-
ment of interdependence in trade and investment policies. Fig. 13.1,
which depicts the free trade equilibrium in the presence of inter-
national investment, highlights the importance of this interdepen-
dence between trade and investment. This diagram is a simple
326 Studies in the Pure Theory of International Trade
x,

Figure 13.1

extension of the usual free trade diagram drawn in the absence of


international investment. In Fig. 13.1, HH' is the home country's
transformation curve, the slope of DE (parallel to AB) equals the
world terms of trade, Pis the production point, Cis the consumption
point, and the level of welfare is given by U1 • The home country ex-
ports PQ of X 1 in exchange for QC of X 2 , where RC amount of X 2
represents the home country's foreign earnings in terms of the second
commodity. This presentation serves to show the interdependence of
trade and investment policies, because any policy change, by chang-
ing the world terms of trade, may also alter the foreign rate of return
and hence the length RC and the position of the social indifference
curve in the diagram. Furthermore, the position of the transforma-
tion curve also depends, among other things, on the level of foreign
investment.
The link existing between trade and investment policies has been
demonstrated by several authors including MacDougall [ 4],
Negishi [6], Kemp [3], Jones [2] and Gehrels [I] among many
others. The approach adopted here draws chiefly upon the work
of the last three authors.
International Investment 327

The setting of the problem is quite simple. We wish to maximise


the social utility function
U = U(D 1 ,D 2 ) (13.4)
under the constraints given by
D1 = X1-£1 (13.5)

(13.6)

(13. 7)
and
(13.8)

where (13.5) and (13.6) have been described several times before;
(13.7) specifies the production constraint which takes into account
the amount of home country's capital(/) invested abroad; and (13.8)
describes the balance-of-payments constraint in the presence of the
foreign investment, that is, the value of imports equals the value
of exports plus the earnings on foreign investment which equal rii,
where 'I is the foreign rate of return on capital expressed in terms
of the second commodity. If the home country is a net creditor, I is
positive; it is negative if the home country is a net debtor. Using
(13.5), (13.6) and (13.8), (13.4) can be written as

(13.9)

The problem now reduces to one of maximising (13.9) subject to the


production constraint (13.7). This can be solved by the well-known
way of setting up the Lagrangian multiplier to obtain

Z = U(X 1 -E 1 ,X2 +rii+Edp)-y¢J(X1 ,X2 ,/)


= Z(£ 1 ,/) (13.10)
because

and

If the foreign country is incompletely specialised, 'I depends only


on p; otherwise it depends on I and orI jiJI < 0.
328 Studies in the Pure Theory of International Trade

The Optimal Tariff


In the presence of both trade and foreign investment, the function
Z is to be maximised with respect to both £ 1 and I. A full optimum
can be obtained only when the policy-makers are free to vary either
E 1 or I without fear of affecting the level of the other, that is, when
£ 1 and I can be treated independently. Keeping this in mind, let us
differentiate (13.10) with respect to £ 1 and set the partial derivative
to zero. Thus

(13.11)

where
(} = 1,2).

Since the marginal rates of transformation and substitution reflect


the domestic commodity-price ratio under conditions of perfect
competition,

(13.12)

Again, because of the equality between the marginal rate of


transformation and the domestic-price ratio, ¢ 1(oXIfo£ 1}+
¢ 2 (oX2 /oE 1 ) = 0. Therefore, for oZjo£ 1 to be equal to zero.

(13.13)

From (13.13), the policy prescription is clear. If the country in


question is a small country, so that o(Ijp)fo£ 1 = 0, the commercial
International Investment 329

policy should be such as to equate Ph with p. In other words, free


trade continues to be the optimal policy for a small country even when
foreign investment is present. For a large country, however, where
8(1/p)/8£ 1 =I= 0, the optimal policy requires the imposition of an
optimal tariff, which can be derived from ( 13.13) by noting that
(ph-p)jp = th. Thus

where ai = - [8Ed8(1jp)][(1jp)jE 1 ] is the foreign elasticity of de-


mand for imports, lli is the ratio of foreign earnings to the value
of exports, and YI = [8rij8(ljp)][(ljp)irJ] is the elasticity of the
foreign rate of return on capital with respect to the world terms of
trade. From this expression, the optimum tariff is given by

(13.14)

What is interesting in the case of (13.14) is not only that the optimal
tariff formula is different in the presence of international investment,
but that there also arises the possibility that the optimal tariff may
be zero or even negative, for llJ and y1 may possess any sign. If the
home country is a net creditor, lli > 0; YI on the other hand depends
on whether X 1 is capital- or labour-intensive in the foreign country.
If X 1 is capital-intensive abroad, YI > 0, and if it is labour-intensive,
YI < 0. This follows from the Stolper-Samuelson theorem, but only
if the foreign country is incompletely specialised. Thus we conclude
that in the case of incomplete foreign specialisation the optimal tariff
may be of any sign and may even be zero, even if the country in
question possesses monopoly power in international trade. The reason
for this surprising result is not far to seek. In the presence of foreign
investment two terms of trade are involved, namely,p and ri, and as
a result of the tariff p and ri may or may not move in the same
direction, depending on whether the first good is capital- or labour-
intensive abroad. Now the home income rises owing to a decline in
p, but declines owing to a decline in 'I· Thus if the two effects cancel
out, the optimal tariff is zero. Nevertheless, in the presence of
complete foreign specialisation when YI = 0, the optimal tariff
reduces to the standard formula derived in Chapter 5.
330 Studies in the Pure Theory of International Trade

The Optimal Tax


The optimal rate of taxation on foreign earnings or the earnings of
foreign capital can be obtained in an analogous manner. Given that
£ 1 and I are independent,

oz oX1 [oX2 o(I/p) or1 ]


of = ul TI+ u2 of +El --a~+rf+I of
oxl oX2 )
-y ( 4J1 TI+4J2 TI+4Jf = o. (13.15)

In (13.15), 4J1 # 2 is the marginal contribution of a unit of capital to


the home country's national product expressed in terms of the
second commodity, and equals rh, the rate of return on capital in
the home country, and
4J1 oX1 oX2 ul oX1 oX2
4J 2· of+ of = U 2·TI+TI = -rh
that is, a unit of capital flowing abroad reduces home income by rh
when commodity prices are kept constant. All this implies that the
last term in (13.15) equals zero. Thus (13.15) reduces first to

and then to
(13.16)

Before we derive the optimal rate of taxation on international invest-


ment, it is necessary to be clear about the relationship between r1
and rh in the presence of such taxation. Suppose the home country
is a net creditor and she imposes a rate of tax equal to lOOt per cent
on her foreign earnings; then the equilibrium condition (13.1) gives
way to

and
MPK 2 h = MPK 21 (1-rh).
But rh = MPK1h/Ph and r1 = MPK 11 /p, so that
MPK 11 (l-rh) pr/1-rh)
rh = = .
Ph Ph
International Investment 331

But if the home country already has a tariff on her imports,


Ph= p(l + th). Therefore
1 -rh
rh = 'J I+ th. (13.17)

On the other hand, if the home country is a net debtor, and imposes
a tax r1 on the earnings of foreign capital, the capital market is in
equilibrium if

and
MPK 2h(l-r1 ) = MPK 21 .
Using rh and r1 , however, this becomes
rh(l-r1 )(l+th) = r1 . (13.18)
Substituting for rh from (13.17) and (13.18) in (13.16), the two
optimal rates of taxation are given by

and

where y1 = - (or1 jol)(Ijr1 ) is the elasticity of the foreign rate of


return to capital with respect to international investment and y1 has
been defined as before. Using the expression for the optimum tariff,
these formulae become

(13.19)

and
0 +Ji.JYJ)(flfyf-afyi)
'J = . (13.20
a1 [JJ.1 y1 -yi(1 +JJ.1 Y1 )]
As before, the signs of rh and r1 depend on the signs of y1 and Ji.J
which in turn depend on the inter-industry factor-intensity relation-
ships in the foreign country and whether I ~ 0. To begin with,
suppose that the optimum tariff is zero, that is, 1+ Ji.JYJ = 0. This,
from (13.19) and (13.20), means that the optimal taxes on foreign
332 Studies in the Pure Theory of International Trade

investment must also be zero. In other words, welfare maximisation


in the presence offree trade also calls for the absence of intervention
in international capital movements. This policy applies to both the
small and the large country cases. For example, suppose that the
foreign country is incompletely specialised, so that r1 is influenced
only by p and not by I. Then y1 = 0, and if the home country is
small, a1 = oo; one can see immediately from (13.19) and (13.20)
that both -rh and -r1 then reduce to zero. However, if the optimum
tariff is not zero and the home country possesses monopoly power
in trade, then it is clear that the optimal taxes on foreign investment
may also be non-zero. As an illustration, let us suppose that the
home country is a net creditor, so that I and Jl.J are positive; if X1
is the capital-intensive commodity in the foreign country, then in
the presence of incomplete foreign specialisation y1 > 0 and y1 = 0.
Furthermore, assume that a1 > 1 + Jl.JYJ· Then, under these circum-
stances, it pays the home country to impose not only an optimum
tariff but also an optimal tax (-rh) on its foreign earnings. However,
if y1 is negative, and Jl.J still positive, the optimal tax on foreign
earnings is indeterminate. Similar conclusions can be derived for
the case where the home country is a net debtor, so that Jl.J < 0
and the relevant tax is -r1 . The important point is that, in the presence
of international investment, optimal policy may require the im-
position of both tariffs on imports and taxes on foreign investment.
Up to now we have assumed that the volume of imports and
international investment are independent, so that full optimisation
is possible. Quite often, however, policy-makers are not free to vary
both imports and foreign investment, in which case £ 1 and I no
longer remain independent. However, the new constraints need not
alter the intrinsic nature of the results derived above, although the
optimal tariffs and taxes will be quantitatively altered. The main
objective of this chapter was to outline the method of formulating
the optimal policy in the presence of foreign investment, and this
seems to have been done. The interested reader, however, can easily
derive alternative tariff and tax formulae in the case of partial
optimisation.

REFERENCES
[I] Gehrels, F., 'Optimal Restrictions on Foreign Trade and Investment', American
Economic Review, LXI (Mar 1971) 147-59.
International Investment 333
[2] Jones, R. W., 'International Capital Movements and the Theory of Tariffs and
Trade', Quarterly Journal of Economics, LXXVII (Feb 1967) 1-38.
(3] Kemp, M. C., 'The Gain from International Trade and Investment: A Neo-
Heckscher-Ohlin Approach', American Economic Review, LVI (Sep 1966)
788-809.
[4] MacDougall, G. D. A., 'The Benefits and Costs of Private Investment from
Abroad: A Theoretical Approach', Economic Record, XXVI (Mar 1960) 13-35.
[5] Mundell, R. A., 'International Trade and Factor Mobility', American Economic
Review, XLVII (June 1957) 321-37.
[6] Negishi, T., 'Foreign Investment and the Long-Run National Advantage',
Economic Record, XLI (Dec 1965) 628-32.
[7] Rakowski, J. J., 'Capital Mobility in a Tariff-Ridden International Economy',
American Economic Review, LX (Sep 1970) 753-60.
14 International Trade
in a Dynamic Economy

Recent years have witnessed a steadily growing application of the


standard two-factor, two-good model to the solution of questions
that arise in a growing economy, questions that were first raised in
a seminal contribution by Oniki and Uzawa [5]. Although some
insights as to what happens in the economy under conditions of
growing factor supplies and technology were gained in Chapter 6,
the nature of our analysis there was essentially comparative statics,
for we were concerned primarily with the implications of exogenous
and once-for-all growth in factor supplies. In the present chapter,
however, our concern will be the development of a dynamic model
where labour grows exogenously at a certain given rate, but where
capital grows endogenously as a result of the savings habits of the
factor owners. Oniki and Uzawa and, following them, otherst have
used a technique which, for want of a better substitute, I call the
'production function' technique. The mathematical calculations
under this technique are quite lengthy and oppressive, and a super-
ficial glance is sufficient to discourage the student from getting
seriously involved with the problem. Fortunately, the technique
suggested by the activity-analysis approach used in the past chapters
cuts down the length of the derivations and makes the issues suscep-
tible to better comprehension.

14.1 Assumptions and the Model


In order to transform the standard two-good model into a dynamic
model of international trade, it is necessary to assume that one of
the traded goods is the capital good whose demand is determined
by the quantum of savings generated in the economy. The avail-
ability of capital goods in the current period contributes in the next
period to the total stock of capital which, if full employment is to

t See, for example, Bardhan [1,2], Kemp [3] and Khang [4].
International Trade in a Dynamic Economy 335

be maintained, must be fully utilised in conjunction with a labour


force that is exogenously growing at a certain rate. The problem is
to find that rate of growth of capital which takes the economy to the
steady-state equilibrium where both capital and labour grow at the
same rate and the capital/labour ratio remains constant over time.
Throughout this chapter it is assumed that the second good is the
capital good and the first good is the consumption good. As far as
the savings behaviour of the community is concerned, we assume
that a certain proportion of income is saved in each time period at
a constant rates. Up to now we have assumed that X 1 is exported
and X 2 imported by the home country. However, in a dynamic
setting where factor supplies change over time, there is no guarantee
that the same pattern of trade will be maintained for ever. Since any
commodity can be the import good, we shall assume that Ej =
Dj- Xj is the excess demand for the jth good (j = 1,2), which is
positive if the good is imported and negative if it is exported. t For
the sake of simplicity it is assumed that capital is everlasting, so that
there is no depreciation factor to worry about. Except for these
alterations, the rest of the assumptions are the same as those pre-
sented in Chapter 2.
Let us start with the case where the home country imports the
capital good, so that its domestic consumption is given by

( 14.1)

but the value of D 2 is constrained by the total amount of savings in


the economy, that is,

p 2 D2 = sY = s(rK+wL). (14.2)

Substituting (14.2) in (14.1), we can derive an expression for the


import demand for the capital good. Thus

£ 2 = s(!_ K+~L)-x2 • (14.3)


P2 P2

t Up to this chapter we have assumed that £ 1 equals the quantity exported of X 1


and hence equals X 1 -D 1 , whereas £ 2 is the quantity imported of X 2 and equals
D 2 - X 2 . Such specific labelling of the goods was possible in the static model where
factor supplies are constant. In a growing world economy, where factor supplies
grow over time, this labelling is not possible. That is why both £ 1 and £ 2 are defined
in the same way, remembering that their signs are always opposite.
336 Studies in the Pure Theory of International Trade

In addition to (14.3), which specifies the excess demand function for


the importable good under conditions of incomplete specialisation,
there are two more possibilities, depending upon whether the home
country is specialised completely in X 1 or X 2 • If the world terms
of trade (p) fall below a certain level, the home country will be com-
pletely specialised in X 1 , and X 2 will decline to zero. Let us call such
a p as Pmin· Then for p :-=;; Pmin• the excess demand function reduces
to

£ 2< · > = s ( -r K +-
w )
L = sX-
1
(14.4)
mm P2 Pz P

because Y = rK+wL = p 1 X 1 . On the other hand, ifp rises above a


certain level, the home country will be specialised in X 2 • Let us call
such a p as Pmax· Then for p 2: Pmax

so that the export of the capital good will be given by

£2(max) = -(_!_K+!!!__L)(1 -s) = -(1-s)Xz. (14.5)


Pz P2

14.2 The Uniqueness of Momentary Equilibrium


At any moment of time, factor supplies in the home and the foreign
country are given, so that the excess demand depends solely on the
terms of trade. It follows then that if the relationship between p
and £ 2 is monotonic, the momentary equilibrium will be uniquely
determined. In the case of complete specialisation in X 1 , (14.4) makes
it clear that the excess demand curve becomes a rectangular hyper-
bola, for s is given and X 1 remains constant so long as p :-=;; Pmin·
In the case of complete specialisation in X 2 , however, the excess
demand function becomes a negative constant, as is evident from
(14.5), because X 2 is unchanged so long asp 2: Pmax·
The relationship between £ 2 and pis not so straightforward when
both goods are being produced. Differentiating (14.3), we have'
International Trade in a Dynamic Economy 337

where OK and eL are respectively the shares of capital and labour in


the economy and 8K+8L = 1. From equations (2.35) and (2.36) in
Chapter 2 we know that

Substituting these, we have

Now eK can be written as

where ej is the share ofthejth good in the economy and 8 1 + 8 2 = 1.


Substituting for eK yields

(14.6)

because, from (2.17) in Chapter 2, IOI = eK2-eK1· Since X'5./P* is


positive, (14.6) shows that £ 2 and p are negatively and uniquely
related. All this discussion suggests that the excess demand curve
would look like the curve AB depicted in Fig. 14.1, where for p
between 0 and Pmin• AB is a rectangular hyperbola, for p :2: Pmax•
AB is a straight line parallel to the horizontal axis, and for p lying
between Pmin and Pmax• AB reflects a simple negative relationship.
For plying betweenpmin and Fthe capital good is imported, whereas
for the world terms of trade lying above OF the second good becomes
too expensive to be imported; instead it becomes an export good
of the home country.
In what follows, we assume that the variables without the country
subscript refer to the home countries, whereas those with the sub-
script f pertain to the foreign country. The solution for p can be
338 Studies in the Pure Theory of International Trade

Figure 14.1

obtained from the condition that in equilibrium the world excess


demand for any good equals zero, that is,
(14.7)
Fig. 14.2 makes use of this condition, depicts the excess demand
curve for both countrues, HH' for country Hand FF' for country F,
and shows that equilibrium is achieved when the two curves inter-
sect at Q. The equilibrium terms of trade are given by OA, E 2 equals
QA and E21 = -QA, so that E 2 +E21 = 0.
In order to tackle some more difficult questions in a dynamic
economy where labour is growing exogenously at a certain rate, the
exposition is simplified if we present the variables in per capita terms.
Let the lower-case letters denote the variables in per capita terms.
International Trade in a Dynamic Economy 339

\
\
\
\
\
\
\
\
\
\
0~------~~~----~~----~'T---------------~p
\
\
\
\
\
\
~------------~--H'
\
\
\
~----

Figure 14.2

Thus ei = Ei/L, xi = Xi/L, y = Y/L, and so on. At the outset, it


may be noted that the relationship between the excess demand
function and the terms of trade remains unchanged even when it is
expressed in per capita terms. The reason for this is attributable to
the fact that (14.6) has been obtained under the assumption that K
and L are given at any moment of time, so that K* = L* = 0. Using
this, (14.6) can be written as

* _ -p
e2- •[d2
- (} +-
x2 ·----.
x!J
1 (14.7*)
e2 e2 p
where e! = £!- L* and x! = X!- L*. Thus the relationship
between the excess demand function and the terms of trade changes
little when it is stated in per capita terms, provided the factor sup-
plies are constant. The uniqueness of the momentary equilibrium
340 Studies in the Pure Theory of International Trade

shown above can be demonstrated in exactly the same manner from


(14. 7*). The only difference that occurs in Figs. 14.1 and 14.2 con-
cerns the labelling of the vertical axes; they would now measure e2
and pe 21 instead of £ 2 and £ 21 , where p = L1 /L is a constant (see
equation (14.15), which appears below and is derived from (14. 7)).
Fig. 14.2 depicts the case where the equilibrium terms of trade are
such that each country produces both commodities. However, this
is only one among a number of other possible solutions. In general,
the resulting pattern of specialisation depends on the levels of k and
kj(k = K/L). In order to determine the influence of k and k1 on the
pattern of specialisation, it is necessary to see how the excess de-
mand curve shifts as a result of a change in k when the terms of
trade are kept constant. In other words, we are interested in the
sign of oe 2 jok.
The three equations (14.3)-(14.5) may be rewritten in per capita
terms as
(14.8)

s
e2(min) = -X 1
p
(14.9)

and
e2(max) = -(1-s)Xz (14.10)
where w is the wage/rental ratio. In the case of complete specialisa-
tion, the sign of oe 2 jok is apparent from one glance at (14.9) and
(14.10). Evidently,

oe2(min) - s ox 1 0 (14.11)
~-pak>

and
8e2(max) = -(I-s) OXz < 0 ( t4.12)
ok ok
because (ox)ok) > 0. The case of incomplete specialisation, how-
ever, is not susceptible to such straightforward results. Differenti-
ating (14.8) partially with respect to k and remembering that wand
r are constant with a given p, we have

oe2 sr ox2
ok = P2 -8k (14.13)
International Trade in a Dynamic Economy 341

Now p 1 (oxtfok)+p 2(ox 2/ok) = oyjok = r. Substituting this in


(14.13) yields
oe 2 p 1 (oxtfok)-(l-s)r
(14.14)
ok P2

The sign of oe 2jok depends clearly on the factor-intensity relation-


ship between the consumption good and the capital good. If X 1 is
labour-intensive relative to X 2 , then from the Rybczynski theorem
oxtfok and hence oe 2jok are negative. On the other hand, if the
consumption good is capital-intensive relative to the capital good,
oxtfok is positive and so is oe2jok, because with (ox2jok) < 0,
p 1 oxtfok is clearly greater than rand hence (1-s)r. This last case
is one of special interest. The hypothesis requiring the capital good
to be labour-intensive relative to the consumption good is widely
known as the capital-intensity condition, and has been very frequently
used in the past in the stability analysis of two-sector growth models.
We are now in a position to show the influence of k and k 1 on the
international patterns of specialisation. Suppose there is a once-
for-all rise ink, but that k 1 is unchanged. Under the capital-intensity
condition, oe 2jok > 0, so that the home country's excess demand
curve HH' in Fig. 14.2 shifts upwards to the dotted curve, the
equilibrium point shifts to Rand the terms of trade rise to OS. The
pattern of ~pecialisation at R is such that the home country produces
both goods, but the foreign country is completely specialised in its
exportable good, X2 • Thus even though k1 was unchanged, a change
ink produced a change in the foreign country's level of specialisa-
tion. As indicated above, a rise in k also leads to a rise in p. The
generalised expression for the sign of opjok can be obtained from
the following equilibrium condition:

(l4.t5)

where p = L1 jL is the ratio between the labour force available in


the two countries. Hereafter we assume that the labour force in both
countries grows at the same rate, so that p remains unchanged over
time. Differentiating (14.15) partially with respect to k, we obtain

op
= (14.16)
ok
342 Studies in the Pure Theory of International Trade

Similarly,

(14.17)

Since the denominator in (14.16) and (14.17) is unambiguously


negative, the sign of ( opjok) or (opjok1 ) depends completely on the
sign of (oe 2jok) or (oe 21 jok1 ). Under the capital-intensity con-
dition, oe 2/ok and oe 21 /ok1 are positive, and so are op/ok and opjok1 .
However, this need not be the case if the country is completely
specialised. If the home country is specialised in the first good and
the foreign country in the second, then opjok > 0 but opjokf < 0,
because from (14.11) and (14~12) one can see that oe 2 <min)/ok > 0,
but oe 21<max/ok1 < 0. The signs of op/ok and opjok1 are reversed
when the home country is specialised in the second good and the
foreign country in the first.

14.3 The Stability of the Long-Run Equilibrium


The long-run growth path of an economy is determined by the rate
of capital accumulation, which equals the domestic production of
capital goods plus their imports, and the exogenously given rate of
growth of labour, which is assumed to be the same for the two
countries. The rate of capital accumulation is then given by

K
* =D-
2 sr (w+k)
=-·-- (14.18)
K p2 k

because we have assumed for simplicity that there is no deprecia-


tion. Now the long-run equilibrium in each economy is defined by
the equality of K* and L *, and the capital/labour ratio prevailing
at that equilibrium is called the steady-state capital/labour ratio.
This steady-state capital/labour ratio is unique if, whenever there is
a divergence between K* and L *, K* comes back to the given L *,
such that the steady-state capital/labour ratio remains the same.
This means that whenever k rises above the steady-state value of k,
the adjustment of the economic system should be such as to lower
K*, and conversely. In other words, the stability of the long-run
equilibrium in each country requires that dK* /dk < 0 and
dKlfdk1 < 0.
International Trade in a Dynamic Economy 343

The rate of change in the capital/labour ratio in each economy


is given by

Q = k* = K*-L* = sr (w+k) -L* (14.19)


Pz k
and

(14.20)

In the steady-state equilibrium, both Q and Q1 are zero. In general,


both Q and Q1 depend on k and k1 , but the effects of a change ink
or k1 depend on the pattern of specialisation in the two countries.
At any moment of time, the pattern of specialisation in any
country will conform to any one of the three possible patterns. In
the case of the home country, for example, the country may be
(a) incompletely specialised, (b) specialised in X1 , or (c) specialised
in X 2 . As stated before, the reason for these varying patterns lies in
the fact that in a dynamic model, where k and k1 are changing over
time except when the economies are in the steady state, the patterns
of trade and specialisation are themselves subject to changes over
time. In the same way, the foreign country may have any of the
three possible patterns of specialisation. In the following we shall
consider three cases, namely, (a) both countries are incompletely
specialised, (b) the home country is incompletely specialised but
the foreign country is specialised in the capital goods, and (c) the
home country is specialised in the capital goods but the foreign
country is incompletely specialised. It turns out that it is sufficient
for the stability analysis to consider these three combinations out of
a possible seven.

(a) Both Countries Incompletely Specialised


As indicated before, the signs of dK* jdk and dKl jdkf determine
whether or not the long-run equilibrium in each economy is
stable. This evidently calls forth a differentiation of (14.19) and
(14.20). Let kf be constant. Then from (14.19)

w
Q* = (r*-p*)+--(w*-k*) (14.21)
2 (w+k) ·
344 Studies in the Pure Theory of International Trade

From equations (2.36) and (2.31) in Chapter 2 we know that with


incomplete specialisation

• •
r -p 2
(hzp*
=-~-()-I an
d
w• = w·
-p*

Substituting these in (14.21), we have

Q* w ]p*
K* =
[
() LZ - k +w rei - k +w w k*.
Now
w
-k-- = {)L = {)l{)Ll +{)2{)L2·
+w
Using this relationship and the fact that
1{)1 = {)Ll- {)L2

it is readily seen that


Q*
(14.22)
k*
For the attainment of the steady-state equilibrium in the home
country, we require that Q*/k* < 0. If the capital-intensity
condition is satisfied, then from (14.16) p* fk* > 0, which means
that Q*/k* < 0. However, if the capital-intensity condition is not
satisfied, p*/k* < 0, so that the sign of Q*/k* becomes uncertain.
Proceeding in exactly the same manner, keeping k constant, we
can show that
Q* p* K*
- ()1 k* Q" (14.23)
kJ f
Similarly, from (14.20),
QJ p* K*
f
-()If k* (14.24)
k* Q/
and
QJ
-= (14.25)
k*
f

These three equations confirm the fact that Q* /kJ, QJik* and
QJikJ are all negative under the capital-intensity condition.
International Trade in a Dynamic Economy 345

The discussion up to this point suggests that under the capital-


intensity condition the long-run equilibrium in each economy is
stable. Starting from any arbitrary level, the capital/labour ratio
in each country tends to approach the steady state, where both Q
and Q1 are equal to zero. However, there still remains the question
of what happens to the steady-state capital/labour ratio in one
country when there is a change in the capital/labour ratio of the
other country. This question is motivated by our interest not only
in the conditions ensuring the attainment of the steady-state
equilibrium in each country, but also in the simultaneous attain-
. ment of the steady state in both countries. Furthermore, how do
we determine the pattern of specialisation in each country when the
world economy rests in the steady-state equilibrium?
The answer to these questions is hidden in the behaviour of the
Q = 0 curve as k1 changes and the Q1 = 0 curve as k changes.
When both countries are incompletely specialised,
Q = Q(k,kf)
and
Qf = Q/k,kf ).
By differentiating the Q = 0 curve totally with respect to k, we
obtain
Q* dk Q* dk
k*T+k*J!
J f
= o.
Using (14.22) and (14.23) then gives us
k dk1 ) Q*/k* (0 1 p*jk*)+fh
(
k1 dk Q=o Q*/kJ elp*/kJ
(p*jk*)+(0Lf0t)
(14.26)
(p*fkJ)
Similarly, we can derive
p*jk*
(14.27)

It is clear from these two equations that

(~t)Q=O <(~)Q,=O <O. (14.28)

This last equation will be used in the subsequent analysis.


346 Studies in the Pure Theory of International Trade

(b) The Home Country Specialised in the Capital Good, but the
Foreign Country Incompletely Specialised
If the home country is specialised in the production of capital goods,
then p 2 D 2 = SP2X2 , so that

K* = D 2 sX2 sx 2
K K y·
Moreover, x 2 = l/CL2 and k = CK 2 /CL2. Therefore
s
K*=-
CK2

whence
s
Q = K*-L* = --L*. (14.29)
CK2
However, since the foreign country continues to be incompletely
specialised,

(14.20)

Q* Q*
K*
but k* = 0
J
because CK 2 does not change with a change in k 1 alone. From the
analysis in Chapter 2 we know that
c:.2 = ()L2(1'2(J). and c:.2- Ct2 = k* = (1'2(1).

so thatt
Q*
k* = -{}L2 < 0. (14.30)

Evidently in this case:j:

00. (14.31)

t The reader may be reminded here that in the absence of technical progress Ct.J
is the same as At, the notation used in Chapter 2 for a change in the input-output
coefficient.
t It can easily be seen that the foreign pattern of specialisation plays no role in
determining the sign of (dk1 jdk)a=o when the home country is specialised in the
capital good.
International Trade in a Dynamic Economy 347

(c) The Home Country Incompletely Specialised, but the Foreign


Country Specialised in the Capital Good
In this case
s
Qf= - -
CK2f

but Q is given by (14.19), so that Q*/k* < 0 as before, but Q*/k!


is positive because, with the foreign country specialised in the
capital good, p*fk! < 0. Clearly, then,

( !5___. dkf) > 0 (14.32)


kf dk Q=O •

From our discussion so far,t we can draw the Q = 0 curve in the


k1 and k space as shown in Fig. 14.3, where the Q = 0 curve is

0=0

0 A k

Figure 14.3

t Here again, one can see that the pattern of home specialisation plays no role in
determining the sign of (dk1 /dk)a~o· when the foreign country is specialised in the
cap1tal good. Thus we see that whenever a country is specialised in the capital good,
the level of specialisation in the other country is immaterial in determining the shape
of the Q = 0 and Q1 = 0 curves.
348 Studies in the Pure Theory of International Trade

composed of three segments, namely, (i) the negatively sloped


segment corresponding to the case of incomplete specialisation
and the capital-intensity condition, (ii) a vertical line parallel to the
k1 axis corresponding to the case of complete home specialisation
in the capital good (see equation (14.31)), and (iii) a positively
sloped segment corresponding to the foreign specialisation in the
capital good (see equation (14.32) ). Now for a country to be spe-
cialised in the production of the labour-intensive capital good,
its capitaljlabour ratio ought to be relatively low. For this reason,
the maximum level of the capitaljlabour ratio, below which there
will be complete specialisation in the capital good, is relatively low
for each country; for the home country it is given by k ~ OA
where A corresponds to point B, and for the foreign country it is
given by k1 ~ OC where C corresponds to point F.
The derivation of the Q1 = 0 curve proceeds in the analogous
way. The world equilibrium is achieved where these two curves
intersect, as in Fig. 14.4 at point E. The reason why the Q1 = 0
curve is drawn flatter than the Q = 0 curve in the segments

I
------a,=o

0 k
Figure 14.4
International Trade in a Dynamic Economy 349

relevant to incomplete specialisation is given by (14.28). The


steady-state capital/labour ratios in each country corresponding to
point E are such that both countries are incompletely specialised.
This is how the pattern of specialisation is determined when each
country is simultaneously in the steady-state equilibrium. Needless
to say, the steady-state world equilibrium depicted in Fig. 14.4
could have occurred at any pattern of specialisation for either
country, depending, of course, upon the position and elasticities of
the Q = 0 and Q1 = 0 curves.
The arrows determined by (14.22)-(14.25), (14.30), etc., indicate
the direction of capital accumulation in each country wherever the
initial capital/labour ratios of the countries are. If the countries
are incompletely specialised, then it is easily seen that under the
capital-intensity condition any initial capital/labour ratio in the
two countries will converge to those corresponding to point E as
time approaches infinity. Of greater interest is the case where both
countries are completely specialised in their respective exportable
goods, for then, as is evident from equations like (14.30), the long-

0=0

o,-o

0 k

Figure 14.5
350 Studies in the Pure Theory of International Trade

run equilibrium in the world economy is stable regardless of


whether or not the capital-intensity condition is satisfied. Such an
equilibrium is depicted in Fig. 14.5, and k and kf corresponding
to point E are such that the home country is specialised in the
capital good whereas the foreign country is specialised in the
consumption good.
Up to now we have assumed the capital-intensity condition and
shown that the world economy is globally stable. However, if the
consumption good is the labour-intensive sector, then, short of
complete specialisation in the two countries, our growth model may
not be stable, or there may be multiple equilibria in the world
economy. From the analysis presented above, the interested reader
can easily derive the conditions sufficient to preclude these results.

REFERENCES
[I] Bardhan, P. K., 'On Factor Accumulation and the Pattern of International
Specialisation', Review of Economic Studies, xxxm (Jan 1966) 39-44.
[2] - - , 'Equilibrium Growth in the International Economy', Quarterly Journal
of Economics, LXXXIX (Aug 1965) 455-64.
[3] Kemp, M. C., 'International Trade and Investment in a Context of Growth',
Economic Record, XLIV (June 1968) 211-23.
[ 4] Khang, C., 'A Dynamic Model of Trade between the Final and the Intermediate
Products', Journal of Economic Theory, I (Dec 1969) 416-37.
[5] Oniki, H., and Uzawa, H., 'Patterns of Trade and Investment in a Dynamic
Model of International Trade', Review of Economic Studies, XXXII (Jan 1965)
15-38.

SUPPLEMENTARY READINGS
[6] Atsumi, H., The Long-Run Offer Function and a Dynamic Theory of Inter-
national Trade', Journal of International Economics, I (Aug 1971) 267-99.
[7] Bardhan, P. K., and Lewis, S., 'Models of Growth with Imported Inputs',
Economica, xxxvii (Nov 1970) 373-85.
[8] Inada, K., 'International Trade, Capital Accumulation, and Factor Price
Equalization', Economic Record, XLIV (Sept 1968) 322-41.
[9] Johnson, H. G., 'Trade and Growth: A Geometrical Exposition', Journal of
International Economics, I (Feb I97I) 83-IOI.
(10] Stiglitz, J. E., 'Factor Price Equalization in a Dynamic Economy', Journal of
Political Economy, Lxxvm (May-June I970) 456-88.
[II] Vanek, J., 'Economic Growth and International Trade in Pure Theory',
Quarterly Journal of Economics, LXXXV (Aug 1971) 377-90.
Author Index

Allen, R. G. D. 175, 179 Harrod, R. F. 130, !53


Anderson, J. 239 Heckscher, E. F. 47, 58, 80
Atsumi, H. 350 Herberg, H. 250, 251, 280
Hicks, J. R. 7, 17, 130, !53
Balassa, B. 202, 204, 238 Hume, D. 129, 133
Balogh, T. 130, 153
Barber, C. L. 202, 203, 238 Inada, K. 350
Bardhan, P. K. 334, 350 Iverson. C. 129, !53
Basevi, G. 202, 238
Bastable, C. F. 129, !53 Johnson, H. G. 78, 80, 87, 109, 125,
Batra, R.N. 163, 179, 187, 201, 273, 126, 128, 145, 153, 203, 238, 244, 270,
274,279,280,290, 304,315,321 274,275,280,320, 321,350
Bhagwati, J. N. 51, 53, 61, 62, 79, 91, Jones, R. W. xi, xii, 27, 45, 110, 114,
94, 98, 99, 102, 106, 108, 138, 147, 115, 128, 220, 230, 238, 250, 280, 326,
153,250,270,279 333
Casas, F. R. 167, 179, 187, 188, 201,
235,238,280 Kemp, M. C. 63, 78, 80, 85, 89, 90, 94,
Chipman, J. S. 43, 45, 77, 78, 79, 169, 109, 117, 123, 128, 154, 179,250,280,
179 326,333,334,350
Clement, M.D. 130, 153 Keynes, J. M. 129, !53
Corden, W. M. 97, 108, 131, 134, 147, Khang, C. 201, 334, 350
153,203,230,238,2 39 Komiya, R. 312, 321
Krueger, A. 0. 99, 109
Dietzel, H. 129, 153 Kuo, C. 201

Edgeworth, F. Y. 129, 136, 153 Leith, J. c. 203, 209, 238


Ellsworth, P. T. 209, 238 Leontief, W. W. 58, 80
Ethier, W. 163, 179, 239 Lerner, A. P. 126, 128
Lewis, S. 350
Fanno, M. 129, 153 Lewis, W. A. 240, 280
Findlay, R. 147, 151, 153 Lipsey, R. 101, 102, 105, 107, 109
Lloyd, P. J. 202, 238
Gehrels, F. 102, 109, 326, 332
Graham, F. D. 110, 128 MacDougall, G. D. A. 326, 333
Grube!, H. G. 202, 238 Magee, S. P. 255, 280
Grubert, H. 147, 151, !53 Marshall, A. 73, 80, 110, 128
Guisinger, S. E. 163, 179, 209, 238 McKinnon, R.I. 169, 179
Melvin, J. R. 171-3, 179, 281, 283,290,
Haberler, G. 274, 280 304, 308, 321
Hagen, E. E. 280 Metzler, L.A. 110, 118, 128
Harberger, A. C. 240, 280, 320, 321 Mill, J. F. 47, 80, 110, 128, 129, 153
352 Studies in the Pure Theory of International Trade

Mundell, R. A. 323, 333 Scully, G. W. 273, 279


Singh, R. 201
Naya, S. 61, 62, 63, 80, 239 Smith, A. 81
Negishi, T. 326, 333 Sodersten, 8. 110, 128
Sonnenschien, H. 99, 109
Ohlin, B. 47, 58, 80, 129, 153 Srinivasan, T. N. 98, 99, 108, 230, 238,
Oniki, H. xii, 334, 350 250,280
Stiglitz, J. E. 350
Pattanaik, P. K. 163, 179, 274, 279,280 Stolper, W. F. 35, 45, 122, 128
Pearce, I. F. 280
Pfister, P. L. 153 Taussig, F. W. 129, 153
Prebisch, R. 131, 153, 273 Thornton, H. 120, 153
Travis, W. P. 202, 238
Rakowski, J. J. 333
Ramaswami, V. K. 230, 238, 280 Uzawa, H. xii, 334, 350
Ricardo, D. 47, 80
Robinson, R. 67, 80 Vanek, J. 154, 163, 179, 350
Rockwell, K. J. 153 Vind, K. 110, 128
Ruffin, J. R. 233, 235, 237-9 Viner, J. 101, 102, 109
Rybczynski, T. M. 37, 45, 322
Wagner, A. 129, 153
Samuelson, P. A. 35, 45, 82, 84, 109, Warne, R. D. 179,281, 283,290, 304
110, 122, 128 Wegge, L. L. 123
Savosnick, K. M. 7, 13, 17
Scitovsky, T. 41,45 Yates, P. L. 154, 179
Subject Index

Analysis: income 26, 32, 310


Comparative-statics 335 of foreign rate of return 329
dynamics 334-5 Equilibria:
autarky 40,63,290-3
Bolt Diagram 12-14, 37-8 free trade 50, 196-7,300-1,325-6
with three goods 308-9 international trade 68-9
with wage differentials 244-5 momentary 336-42
monopoly 284-93
Capital-intensity condition 341-2, multiple 44, 51, 53, 63, 350
344-5,348-50 self-sufficiency 51, 70, 87, 197
Commodity prices: stability 72-3, 342-50
factor endowments 32 steady state 342-5, 349
factor prices 27, 246, 289, 293
technical change 34 Factor:
Comparative: immobility of 274, 276
advantage 48 market imperfections 240-79,281
costs 46,67 price rigidity 275-9
factor productivity 49 unemployment 279
statics 22 Factor abundance 60, 64
Cone of diversification 77-8 physical definition of 58, 253, 293
Consumption gain 87, 266 price definition of 58, 251, 293
Consumption subsidy 318-19 Factor-intensities:
Consumption tax 98 in the physical sense 243
Cost minimisation 3, 24, 286, 310 in the value sense 243
Customs union 101 non-reversibility 18, 28, 75, 293
welfare implications 101-9 reversals 30-2,60-1, 66, 76, 78
trade-creating I02 under monopoly 287
trade-diverting 102 wage differential 243-6
with intermediate products 158,
Demand: 182-4, 215
pattern of trade 51, 61,66-7 with non-traded goods 312
reciprocal 69 Factor prices 27, 246, 289, 313
Dollar problem 130 Factor price equalisation 75-9, 252,
Duality relations 38-9, 45 293-4,306-7,324-5
Free trade 50. 75
Elasticities: and international investment 329
cross 310 non-optimality of 85, 263-70
demand substitution 26, 283 optimality of 85, 87, 231-4
factor substitution 3, 4, 23-4, 161, versus no trade 86, 263-5, 274-9,
174, 199, 219 296-301
import demand 72, 329 versus restricted trade 87
354 Studies in the Pure Theory of International Trade

Growth: of unions 256-7


Immiserising 136-9, 270-3 Non-traded goods 83, 90, 304-21
in a dynamic economy 334-50 and gains from trade 315-21
ultra-biased 134 as intermediate inputs 211, 217, 223
with factor accumulation 139-44, price output response 309-312
166, 272, 313-14 Rybczynski theorem 313-14
with technical change 145-53, 167-9 Stolper-Samuelson theorem 312
tariffs 315
Heckscher-Ohlin theory 58, 68, 193-6 wage-differentials 315-21
and wage differentials 251-4 Number of factors 78-9
under monopoly 293
Homothetic preferences 32-3, 43, 57, Offer-curves 69, 82, 120-1
63, 283-4 and economic expansion 133
and imported inputs 173-4
Indifference curves: and tariffs 118
community 40-2, 53 as an excess demand curve 338-9
social 87 in Ricardian economy 74
Inferior goods 91, 116
Intermediate products: Paretian optimum 81
factor intensities 158, 182-4, 215 Prebisch hypothesis 273-4
gains from trade 169-74, 196-8, Production coefficients 19, 22, 306
231-8 Production functions:
model with inter-industry flow homogeneous 1-3, 82, 225
155-8 international differences in 48, 68
model with pure intermediate international similarities of 47,
products 181-4 57-9,68,306-7,325
technical progress 167-9 neo-classical 2, 47
transformation curve 158-65, 186-7 technique 334
International investment: Production gain 87, 266-7
balance of payments 327 Production subsidy 88, 90, 267-8,
capital mobility 322-5 318-20
free trade 329
optimal tariffs 325, 328-9 Ricardian theory 48, 55, 68
optimal taxes 325, 330-2 Rybczynski theorem 37-8, 139, 166-7,
187-8,253,272,293,313-14,341
Lerner symmetry theorem 126
Simon-Hawkins condition 157
Magnification effect 38-9 Specilisation:
Metzler·s paradox 118-20 complete 22, 50, 77, 256, 324-5, 329,
Monopoly: 336,342-50
existence of equilibrium 285 incomplete 22, 36, 75, 77, 329, 336,
factor-price equalisation 293-4 342-9
free trade equilibrium 300-1 international 47
gains from trade 296-302 Stability conditions 72-3, 342-50
Heckscher-Ohlin theorem 293 Steady state equilibrium 342-5, 349
prices facing consumers 286-8 Stolper-Samuelson theorem 35-6, 43,
prices facing producers 285-6, 288 122, 165-7, 187-8, 217, 246, 295,
profits 282, 294 312, 329
Rybczynski theorem 293
Stopler-Samuelson theorem 295 Tariffs (effective):
Monopoly power: concept of 202-4
in domestic production 281-303 factor intensities 215
in international trade 86, 126, 297 gains from trade 231-8
Subject Index 355

in general equilibrium 210-38 Trade:


in partial equilibrium 204-10 absence of 47
negative value added 208-10 gains from 81-109, 169-74,315-21
real wages 215-17 intervention 87, 94
resource allocation 218-31 pattern of 47, 193-6, 335
second-best optimum 235-8 Transformation curve:
with fixed proportions 203-7 derivation 12-14
with variable proportions 208 elasticity of 26-7
Tariffs (nominal): linear 16, 49
changes 88,90,94,220-1,225 properties of 14, 15, 21, 66, 291, 316
discriminatory 101 under wage differentials 247-50
domestic prices 118 with imported inputs 231-4
export taxes 126 with inter-industry flows I 58
income distribution 122 with pure intermediate goods 186-7
inferior goods 91 Unique relations:
international investment 323-9, commodity prices and outputs
331-2 249-50,309-12
multiplier 115 factor and commodity prices 29, 36,
non-prohibitive 88 44,77-8,252,289
non-traded goods 315 factor endowments and prices 32, 44,
on many goods 93 63
optimum 123-6, 325, 328-9 factor prices and factor intensities 29
prohibitive 87, 325
terms of trade 113 Wage differentials:
versus consumption tax 98 box diagram 244-5
versus no trade 97 factor intensities 243-6
versus subsidies 95 factor-price equalisation 252
wage differential 264 gains from trade 263-70
Technical improvements 23, 25, 34, 55, Heckscher-Ohlin theorem 251-4
145 immiserisation 270-3
Hicksian classification of, 7 in underdeveloped countries 240,
intensive factor saving 145, 149 273-4
intensive factor using 145, !50 non-traded goods 315-21
neutral 34, 55, 146, 167 optimal policy 267-8
with inter-industry flows 167-9 price output response 249-50
with pure intermediate products productive efficiency 249, 266
188-92 Rybczynski theorem 253, 272
Terms of trade 50, 69, 87, 89, 129, Stolper-Samuelson theorem 246
338-9 terms of trade 259-63
and economic expansion 131 transformation curve 247-50
and tariffs 113 unions 240, 256-7
and welfare 99 welfare 257-8
of underdeveloped countries 131 Walras Law 20
wage differentials and 259-63 Welfare criterion 81-2

You might also like