Stolper Samuelson 1941

The Review of Economic Studies, Ltd.
Protection and Real Wages

Author(s): Wolfgang F. Stolper and Paul A. Samuelson
Source: The Review of Economic Studies, Vol. 9, No. 1 (Nov., 1941), pp. 58-73
Published by: Oxford University Press
Stable URL: http://www.jstor.org/stable/2967638 .
Accessed: 11/12/2014 00:11
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
Oxford University Press and The Review of Economic Studies, Ltd. are collaborating with JSTOR to digitize,
preserve and extend access to The Review of Economic Studies.
http://www.jstor.org
This content downloaded from 128.235.251.160 on Thu, 11 Dec 2014 00:11:36 AM

All use subject to JSTOR Terms and Conditions
Protection and Real Wages
INTRODUCTION
Second only in political appeal to the argument that tariffs increase
employment is the popular notion that the standard of living of the American
worker must be protected against the ruinous competition of cheap foreign
labour. Equally prevalent abroad is its counterpart that European industry
cannot compete with the technically superior American system of production.
Again and again economists have tried to show the falaciousness of this
argument. Professor Taussig, for example, stated that " perhaps most
familiar and most unfounded of all is the belief that complete freedom of trade
would bring about an equalisation of money wages the world over. . . . There
is no such tendency to equalisation."l And Professor Haberler classifies the
argument that wages might suffer from international trade among those " that
do not merit serious discussion. . . , An equalisation of wages comes about
only if labour is mobile [between countries]"2
More recently, however, the writings of Ohlin seem to suggest that a
re-examination of this accepted doctrine might be fruitful. It is the intention
of the present paper to show that definitive statements are possible concerning
the effects of international trade upon the relative remunerationsof productive
agencies, and more important, upon their absolute real incomes. That this is
possible is surprising since the voluminous literature appears to contain only
statements of possibilities and presumptionsrather than of necessities. Indeed,
in the beginning we expected to do no more than delineate factors which
would indicate a likelihood in one direction or another, and only in the course
of the investigation did we discover that unambiguousinferenceswere possible.
It may be illuminating, therefore, to follow in the exposition our original
sequence of thought ratherthan attempt the most direct derivation of theorems.
THE EFFECT OF TRADE UPON RELATIVE FACTOR PRICES

According to the train of thought associated with the name of Ohlin,
differences in the proportions of the various productive factors between
countries are important elements in explaining the course of international
trade. A country will export those commodities which are produced with its
relatively abundant factors of production, and will import those in the pro-
1 F. W. Taussig, International Trade, p. 38. The statement might have been made equally
well with respect to real wages, since in the classical formulation the prices of internationally
traded goods cannot diverge in different countries by more than the cost of transfer. In his
Principles there is a passage which might be interpreted in the opposite direction. " Under certain.
contingencies, it is conceivable that protective duties will affect the process of sharing and so will
influence wages otherwise than through their effect on the total product." 4th ed., p. 5I7. But
the phrasing is not quite clear and refers probably to the share in national income rather than to
the absolute size. We have not found any similar passage either in The Tariff History of the
United States, in Internationsal Trade, or in Free Trade, the Tariff, and Reciprocity.
2 G. Haberler, The Theory of International Trade, pp. 250-25i, bracketed expression ours. See
also the preceding sentence on p. 25 I where Haberler expressly denies that movement of goods will
lead to an equalisation of factor prices. However, as will be discussed below, he does in another
place introduce important qualifications to this denial.

PROTECTION AND REAL WAGES 59
duction of which its relatively scarce factors ate important.' And as a result of
the shift towards increased production of those goods in which the abundant
factors predominate,there will be a tendency-necessarily incomplete-towards
an equalisation of factor prices between the two or more trading countries.2
It is clear that the equalisation is only partial because otherwise we would
be involved in the contradiction that differences in comparative cost would
disappear,and there would be no trade. Although partial, the movement in.the
direction of equalisationis nevertheless real and can be substantial.
Assuming, as we shall throughout, that the total amounts of the factors
of production remain fixed, it is clear from the Heckscher-Ohlin theorem that
the introduction of trade must lower the relative share in the real or money
national income going to the scarce factor of production. For the total return
to a factor equals its price times the amount employed, and since we assume
full employment before and after trade, the total returns to the factors are
proportional to the rates per unit. This argument seems to have relevance to
the American discussion of protection versus free trade. If, as is generally
thought, labour is the relatively scarce factor in the American economy, it
would appear that trade would necessarily lower the relative position of the
labouring class as compared to owners of other factors of production. So far
we have dealt only with the relative shares of the various factors and have not
gone into the effect upon absolute shares. Before entering upon this latter
problem, it is of considerableinterest to mention the most important currently
held viewpoints.
SOME EXISTING VIEWS

Nobody, of course, ever denied that the workersemployed in the particular
industry which loses a tariff could be hurt in the short-run, but according to
the classical theory, in the long-run there would be an increased demand for
those commodities in which the country had a comparative advantage, i.e.
where labour is more productive.3 Although money wages might fall, the
removal of a tariff would result in a still larger reduction in price levels so that
the real wage must rise. In the words of Taussig, " The question of wages is at
bottom one of productivity. The greater the productivity of industry at large,
the higher will be the general level of Wwages."4
How can this argument be reconciled with the Ohlin type of discussion ?
If there were only one commodity produced, then indeed the marginal pro-
ductivity of labour would depend simply on the relative quantities of labour
1 Professor Viner has shown that this line of reasoning was not unknown to the classical
economists. See his Studies in the Theory of International Trade, pp. 500-507.
2 B. Ohlin, Interregional and Inter-national Trade, Chapter II and elsewhere. This appears to
be a novel theorem largely unknown to the classical economists, or at least completely unmentioned
in Viner's masterful review of doctrine. Perhaps the earliest clear enunciation of this doctrine is
that of E. Heckscher in a I9I9 article in the Ekonomisk Tidskrift, cited by Ohlin. Heckscher
apparently gives no prior references. Unfortunately, this important contribution is in Swedish,
and we are indebted to Mr. Svend Laursen for a paraphrasing of its contents. Because of its
extensive development at the hands of Ohlin, we shall refer to it as the*Heckscher-Ohlin theorem.
3 " The free-trader argues that if thie duties were given up and the protected industries
pushed out of the field by foreign competitors, the workmen engaged in them would find no less
well-paid employment elsewhere." F. W. Taussig, Principles of Economics, 4th ed., Vol. I, p. 5I6.
4 Ibid, p. 5I7.

6o THE REVIEW OF ECONOMIC STUDIES
and capital as a whole. And the same would be the case with more than one
commodity if labour and capital were combined in the same proportionsin the
production of each. A movement of the factors of productionfrom one employ-
ment to another would then leave the marginal productivities of labour and
capital unchanged.
Now, while it is true that under the assumptions of pure competition,
homogeneity, and perfect mobility of labour the value of the marginal product
of labour (expressed in terms of any commodity) must be the same in each
occupation,,it nevertheless does not follow that this will depend simply on the
proportionof labour and capital as a whole. For in so far as capital and labour
are combined in different proportionsin each occupation, any change from one
production to another will change the " value marginal productivity " of
labour (however expressed), even though it will, of course, still be equal in all
occupations. In this sense the value marginalproductivity of labour as a whole
may be consideredto depend upon a kind of weighted average of the effective
demands for the various producible commodities. It is the essence of the
argument of the previous section that international trade in accordance with
the principle of comparative advantage so shifts production and the relative
effective derived demands as to produce the Heckscher-Ohlin effect.
It is not surprising that the classical argument should not have touched
upon the problem of relative and absolute shares since for most purposes the
older economists implicitly assumed a one factor economy or an economy in
which different factors of production were applied in a dose whose proportions
never varied. It is to their credit as realists that again and again they relaxed
these assumptions, but they were not always able to weld into a synthesis
these excluded effects.'
Among more modern writers who are nevertheless in the classical tradition
it has long been recognisedthat a small factor of production specialised for the
production of a protected commodity might be harmed by the removal of
tariffs.2 This has received particular attention in connection with the problem
of non-competing groups in the labour market. Certain sub-groups of the
labouring class, e.g. highly skilled labourers, may benefit while others are
harmed. Thus, Ohlin holds that it is quite possible under certain circumstances
for free trade to reduce the standard of living of the manufacturinglabouring
class. " If manufacturing and agriculturallabourers form two non-competing
groups,highprotectionof manufacturingindustriesmay raisethe realwages of the
workers in these industries at the expense of the other factors."3 Similarly,
Haberler remarks that " . . . in the short-run, specialised and immobile
groups of workers, like the owners of specific material factors, may suffer
1 A good case can be made out that even Ricardo did not adhere narrowly to a labour theory
of value, but this is not the place to enter into controversy on this subject. See, however, John
Cassels, " A Re-interpretation of Ricardo on Value," Quarterly Journal of Economics, 'Vol. 49,
pp. 5I8 ff.
2 " It is perfectly clear that the imposition of a prohibitive tariff on the import of raw silk
into the United States would increase the rents of the owners of land suitable for the growth of
mulberry trees and the earnings of workers, if there be such, completely specialised in caring -for
silkworms." M. C. Samuelson, " The Australian Case for Protection Re-examined," Quarterly
Journal of Economics, November, I939, p. 149.
3 Ohlin, op. cit., p. 306.

PROTECTION AND REAL WAGES 6i
heavy reductions in income when for one reason or another they are faced with
more intense foreign competition."' Once the principle that no factor can
benefit from a tariff has been broken, one is tempted to ask whether similar
results are not possible for a large factor of production even if only two factors
are assumed. For the logic of the case seems the same whether two classes of
labour are considered to be non-competing or whether the " non-competing"
factors are labelled " capital" and " labour" respectively.
In treating this problem Haberler expresses doubt that a large and mobile
factor such as labour can be harmed by unrestrictedinternational trade. " We
may conclude that in the long run the working-class as a whole has nothing
to fear from international trade, since, in the long run, labour is the least specific
of all factors. It will gain by the general increase in productivity due to the
international division of labour, and is not likely to lose at all seriously by a
change in the functional distribution of the national income."2 This is not a
dogmatic necessity, but rather regarded as the most probable situation. For
lower on the same page Haberler recognises explicitly a possible qualification.
If labour enters more importantly in the protected industry, it might possibly
be harmed by free trade.3
Viner criticises Haberler's conclusion maintaining that there appears to be
" no a priori or empirical grounds for holding this to be an improbable case."4
In this connection Viner is concernedprimarilywith the relativeshare of labour
in the national money income. In his discussion he introduces as an element in
the problem the prices which consumersmust pay for commodities, particularly
imports and exports with and without protection. Thus, he says, " But even
if labour on the average had low occupational mobility and were employed
relatively heavily in the protected industries, its real income might still rise
with the removal of tariff protection . . . if it was an important consumer of
the hitherto protected commodities, and if the price of these commodities fell
sufficiently as a result to offset the reduction in money wages in the new situa-
tion."5 Ohlin and other modern writers raise this'problem, but it can also be
found in the older literature. Bastable, for example, in good classical fashion
points out that free trade may force a food exporting country " to bring worse
soils into cultivation, and to raise the value of food, thus permitting of an
increase in the amount of agricultural rent. In this instance, the labourers,
and possibly the capitalists, may suffer while the landlords gain."6
We may sum up as follows: (i) In the narrowest classical version the
problem of the effect of trade upon the relative and absolute shares of various
productive factors could hardly arise since only one factor is assumed. (2) Out-
side the confines of this rigid system it has long been recognisedthat the relative
and possibly even the absolute share of a small specific factor of production
might be increased by protection. This received particular attention in connec-
tion with the problem of non-competing groups. (3) With reference to large
1 Haberler, op. cit., p. I95.
2 Haberler, ibid, p. I95.
8 Similar views are attributed to Wicksell, Carver, Nicholson, and others.
4 Viner, op. cit., p. 533.
5 Viner, ibid, p. 533.
6 C, F, Bastable, The Theory of International Trade, 4th ed., p. Io.5,

62 THE REVIEW OF ECONOMIC STUDIES
categories opinion is more divided. Almost all admit the possibility of a decline
in the relative share of a large factor of production such as labour as a result of
free trade; many even admit the possibility of a decline in the real income of a
large factor of production. But all writers consider highly improbablea decline
in the absolute shares, and many believe the same with respect to the relative
shares. Some take the position that no a priori presumption is possible in
connection with the last problem. (4) The vast majority of writers take it as
axiomatic that a calculation of effects upon real income must take into con-
sideration the behaviour of prices of commodities entering into the consumer's
budget. Thus, if the owners of a factor of production consume only the
exported good (in Professor Pigou's terminology this is the wage good), a
different result will be reached than if the wage good were imported. And
since in the real world consumption is diversified so that the concept of a wage
good is an oversimplification,a difficultindex numberproblemwould appear to
be involved.
It is the purpose of the present investigation to show that under rather
general assumptions definite conclusionscan be derived concerningthe absolute
share of a factor (a) even when there is perfect physical mobility of factors of
production and a complete absence of specificity, (b) even if we are dealing with
as few as two large factors of production, and (c) without any recourse to the
index number problem or to the concept of a wage good.
ASSUMPTIONS OF THE ANALYSIS

For purposes of the analysis we shall start out with rather simplified
assumptions, consideringsubsequently the effect of more realistic modifications.
In order to keep the number of variables down to manageable proportionswe
assume only two countries. This involves no loss of generality since the " rest
of the world " may always be lumped together as Country II. For the sake of
exposition and diagrammatic convenience, only two commodities are con-
sidered, labelled respectively " wheat," A, and " watches," B. To accord with
the Ohlin assumptions the production functions of each commodity are made
the same in both countries and involve only two factors of productionidentified
for convenience as labour (L) and capital (C).1
Moreover, by means of a simple device it is possible to avoid detailed
considerationof the second country since all of its effects upon the first operate
via changes in the price ratio of the two traded commodities.2 We shall call
this price ratio of wheat to watches Pa/Pb. It is irrelevant for our argument
just why the exchange ratio of the two commodities is different after inter-
national trade is established; it is sufficient that it does change.3
1 It might possibly give rise to less confusion if instead of capital the second factor were called
land because of the ambiguities involved in the definition of capital. The reader who is bothered by
this fact is invited to substitute mentally land for capital in all that follows.
2 For an example of the use of this device see P. A. Samuelson, " The Gains from International
Trade," Canadian Journal of Economics and Political Science, May, I939.
3 In the limiting Pa/Pb would be unchanged. Also, in the classical constant cost case of a
large country facing a smaller one trade may take place, but to an extent insufficient to result
in complete specialisation on the part of the large country, and hence Pa/Pb may be unchanged.
This exception is touched upon later.

The effect of international trade upon the shares of the productive factors
can now be analysed by varying Pa/Pb as a parameter from its value as deter-
mined in the absence of trade, or with a given amount of protection, to its new
value after free trade is opened up. Throughout we follow the conventional
method of comparative statics, disregardingthe process of transition from the
old to the new equilibrium. Full employment of both factors is assumed to be
realised before and after the change, and each factor is assumed to have
perfectly complete physical mobility.' Throughout pure competition is
assumed. The following symbols are used:
The amount of labour used in producing A .. .. .. La
The amount of labour used in producing B .. .. .. Lb
The amount of capital used in producing A .. .. .. Ca
The amount of capital used in producing B .. .. .. Cb
The total amount of labour used in producing both A and B L
The total amount of capital used in producing both A and B C
It is assumed that regardlessof trade the total amounts of each factor of
production remain unchanged. Therefore, we have the following obvious
identities:
La + Lb L ..... ..(.)......
Ca + Cb C ......................(2)
The production functions relating each good to the inputs of the factors
allocated to its production can be written respectively as:
A A (La, Ca) .............. * (3)
B B (Lb, Cb) ............ (4)
Because we are concerned with proportions and not with the scale of the
process, these functions are assumed to be homogeneousof the first order.
It is a well-known condition of equilibriumthat the ratio of the marginal
productivities of the two factors must be the same in each occupation, because
otherwise there would be a transfer from lower to higher levels. Symbolically
this can be expressed as follows :2
aA(La, Ca) aB(Lb, Cb)
VLa 9Lb
- , ~~.................................... (5)
aA (La, Ca) aB(Lb, Cb)
aCa 9Cb
where the partial derivatives stand respectively for the marginal productivities
of given factors in the production of the indicated commodity.
We are still lacking one condition to make our equilibrium complete. If
we add as a known parameterthe value of Pa/Pb, that is, the price ratio between
the two goods, wheat and watches, all our unknowns will be completely deter-
I We should like to emphasise that in our argument there is no dependence upon imperfections
in the labour market such as form the basis for the Manoilesco type for defense of a tariff. See
M. Manoilesco, The Theory of Protection and International Trade (I93I).
2 Of course, this holds only if something of both commodities is produced, that is, if trade
does not result in complete specialisation. The effect of this qualification is treated below

mined: the amounts of each factor of production allocated to the various
commodities (La, Lb, Ca, Cb), the amounts produced of each good (A, B), and
most important for the present investigation, the marginalphysical productivi-
ties of each factor in terms of each good ( aL ' aCe' aCb)
But what is the meaning in terms of all of the above magnitudes of labour's
real wage ? This is not an easy question to answer if, as is usually true, labour
consumes something of both commodities. In principle it is of course possible
to determine whether a given individual's real income has gone up or down if
one has detailed knowledge of his (ordinal) preference field. But we cannot
gather such knowledge simply from observation of the price changes which
take place. Possibly an index number comparisonof the type associated with
the names of Pigou, Haberler, K6nus, Staehle, Leontief, and others could serve
to identify changes in real income. But we shall later show that this is unneces-
sary. At this point, purely for reasons of exposition, we shall consider the
highly restrictive case where labour consumes only one of the commodities,
that is, where there is a single wage good. In this case the real wage in terms of
that good is an unambiguous indicator of real income' because of the propor-
tionality between occupations indicated in condition (5). It is the marginal
physical productivity of labour in the production of the wage good.
The effect of international trade upon the real wage (thus defined) could
now be determinedmathematically by varying Pa/Pb, the price ratio of the two
goods, and observing how the marginal physical productivity of labour in the
wage good industry is affected. One could perform this purely mathematical
computation by differentiating our equilibrium equations with respect to
Pa/Pb, treating as variables all the unknowns listed above. The result of this
procedure, not shown here because of its purely technical character, would be
found to involve a sum of terms of necessarily different sign, and without
introducing further economic content into the problem, we would not be able
to achieve a definite result, but would be forced, like the older writers, simply to
indicate that all things are possible. However, by introducingfurther economic
content of no less generality than theirs, we shall find that definite results can
be derived.
THE ELIMINATION OF THE INDEX NUMBER PROBLEM
With the assumptions made so far it is hardly surprising that no more
definite results have been reached. For no assumption has as yet been made as
to which country is relatively well supplied with capital or with labour. To
begin with we make two assumptions. The first is that the country in question
is relatively small and has no influence on the terms of trade. Thus, any gain
to the country through monopolistic or monopsonistic behaviour is excluded.
Secondly, it is assumed that the removal of the duty will not destroy the
formerly protected industry, but only force it to contract.
1 It is true that we have been talking about the real wage rate and not about the total amount
of real wages, but as we have assumed full employment before and after any change and unvarying
total amounts of the factors of production, it follows that the real wage sum will always be pro-
portional to the real wage rate.

Now in equilibrium the value marginal productivity (expressed in terms

of any numeraire)must be the same in all occupations, and so must be the wage.
Therefore, whatever wage labour receives in the wage good industry it must
also receive in any other employment. Moreover, any change in the value
marginal productivity and, therefore, the wage rate of labour in the wage good
industry must mean a corresponding change in the wage rate in all other
employments. It follows that we can tell what will happen to real wages (rates
as well as sums) of labour as a whole by investigating what will happen to wages
in the wage good industry. Since the relevant value marginalproductivity, and
hence the wage of labour in the wage good industry, is in terms of the wage good,
and since labour gets the same wage in all occupations, a decline of the marginal
productivity of labour in the wage good industry means a fall in the real wage
rate and the real wage sum of labour as a whole.
In other words, whatever will happen to wages in the wage good industry
will happen to labour as a whole. And this answer is independent of whether
the wage good will be imported or exported, and can be reached without any
discussion of what will happen to prices of the commodities as a consequenceof
international trade.'
Assume, for example, (a) that the country in question is relatively well
supplied with capital, and (b) that the proportion of labour to capital is lower
in the production of wheat than in the production of watches. There is nothing
restrictive about these assumptions because in terms of our previous assump-
tions one of the countries must be relatively well supplied with a given factor,
and through our postponement of the constant cost case for later discussion the
importance of labour must be greater in the production of one of the commodi-
ties. And since the names " wheat " and " watches " are arbitrary, by
re-namingthe variables all possible cases could be expressed in the formulation
given above.
Two alternative cases must now be considered. (I) The good in whose
production capital is relatively important (wheat) is also the wage good.
(2) The good in whose production labour is relatively important (watches) is
the wage good. Each of these possibilities must be considered in turn.
(i) The introduction of trade will shift production in the direction of the
good with " comparative advantage." According to the Ohlin analysis-even
though he would not employ the previous term-this will be wheat which uses
much of the abundant factor. Its production will expand, and part of it will
be exported, while watch production will contract, and part of the watch con-
sumption will be satisfied by imports. This shift in production will be accom-
panied by a transfer of bothlabour and capital from the watch industry to the
wheat industry. But by a reduction in the production of watches more labour
will be set free than can be re-employed at the same rates in the production of
wheat. This is because the amount of capital released,while sufficientto employ
a worker in watch production, is insufficient to employ him in wheat growing
at the old wage rate. Hence wage rates have to go down in wheat growing, and
1 In connection with a slightly different problem the same point is made by F. Benham,
" Taxation and the Relative Prices of Factors of Production," Economica, N. S. Vol. 2, 1935,
pp. I98-203.
E

it follows from the changed factor proportions that the real wage must also
decline. It would be clearly incorrect to argue-as one familiar with the
orthodox theory of international trade would be tempted to do-that in
addition to this decline in productivity due solely to changed factor proportions,
there must be added a further loss to the worker qua consumer resulting from
the inevitable price rise of the exported wage good.
(2) We turn now to the case where watches are the wage good. On the
face of it this case would seem to admit only of an ambiguousanswer, since any
definite conclusion in the productivity sphere would have to confront a neces-
sary fall in the (relative) price of the wage good. Fortunately, that is not so.
This case admits of no less definite an answer than the previous one.
The introductionof trade will increasethe productionof wheat and decrease
that of watches. As shown in the previous case, this will entail a movement of
both labour and capital. But just as labour has less capital to work wvithin
wheat production than formerly, so does labour have less capital to work with
in the production of watches. This is brought about by the change in relative
remunerations of the factors necessary to result in the reabsorption of the
otherwise redundant labour supply. Therefore, regardless of the behaviour of
consumer's good prices, the lowering of the proportion of capital to labour in
the production of watches must adversely affect the marginal physical pro-
ductivity of labour there, and hence, along now familiar lines, the real wage.
We see, therefore, that the seemingly opposite cases lead to exactly the
same result. International trade necessarily lowers the real wage of the scarce
factor expressedin termsof any good: It follows that we are now in a position to
drop the assumption of a single wage good. For if the real wage declines in
terms of every good, real income must suffer regardless of the tastes and
expenditure patterns of the labourers as consumers. Not only can we avoid
making index number comparisons, but it is also unnecessary to make the
assumption of uniform tastes of all workers which such comparisonsimplicitly
presuppose.
DIAGRAMMATICALTREATMENT
It may be useful to illustrate the above arguments graphically. In Fig. i
we plot the familiar substitution curve (productionindifferenceor transforma-
tion curve) between the two commodities in the given country. Before trade,
equilibriumwill have taken place at M with a price ratio correspondingto the
slope of the tangent there. International trade will change the price ratio of the
two goods, and a new equilibriumpoint may be taken as N with more wheat
production, less watch production, and a higher price ratio between wheat and
watches. This diagram represents the result of a fairly complicated economic
process by which the given fixed amounts of productive factors are optimally
allocated between the two commodities in accordance with marginal produc-
tivity conditions which guarantee a maximum amount of one commodity for
preassignedgiven amounts of the other. For many internationaltrade problems
this " short-circuiting" is an advantage; but it omits the essential features of
the present problem, and so we must go back of the substitution curve to the
underlying production relations,

X This is done in Fig. 2 which consists
ta"s
of a modified box diagram long utilised by
Edgeworth and Bowley in the study of con-
sumers' behaviour. This rather remarkable
diagramenables us to represent the relations
between six variables on a two dimensional
figure. On the lower horizontal axis is indi-
cated the amount of capital used in the pro-
duction of wheat. On the left-hand vertical
axis is indicated the amount of labour used in
0 R the production of wheat. Because the amount
of each factor which is not used in the
A
Fig. I productionof wheat must be employed in the
production of watches, the upper horizontal
axis gives us, reading from right to left, the amount of capital used in the pro-
duction of watches. Similarly, the right-handvertical axis, readingdownwards,
gives us the amount of labour used in the production of watches. The dimen-
sions of the box are, of course, simply those of the unchanging given total
amounts of the two productive factors. Any point in the box represents four
and capital used to things: measuring from the lower left-hand corner the
amounts of labour produce wheat, and measuring from the upper right-hand
corner the amounts of labour and capital used in the production of watches.
Disregarding for fhe moment the other commodity, watches, it is clear
that every point in the box correspondsto a given production of wheat, and
\~~~~~~~o
F. svm
Wkt4t~~~~~~~~~~O
Fig.2

hence lies on a uniquely determinableisoquant or contour line of the production
surface. There is a one-parameter family of such curves with the shape as
indicated by the light lines, convex to the lower left-hand corner. Turningnow
to the production of watches, there also exists a one-parameter family of
isoquants convex to the upper right-hand corner, and indicated in the diagram
by a second family of curves.
We are now in a position to derive the substitution curve. Any point in
the box taken at random correspondsto given amounts of watch and wheat
production,but not necessarilyto a point on the substitution curve. Only those
points which reflect an optimal allocation of resourcesaccordingto the marginal
productivity relations stated earlier correspondto points on the substitution or
opportunity cost curve. The locus of points representing optimal positions is
clearly given by joining all the points of tangency of the two sets of contour
lines. It correspondsgeometrically to Edgeworth's contractcurve,and although
the present study does not deal with bargains between contracting parties, we
shall retain this descriptive title. If we hold the productionof one good constant
and thus move along a given isoquant, we will only stop when there is the
maximum possible amount of the other good, or when we have reached the
highest possible isoquant of the other family. This will be so only at a position
of tangency where the ratios of the marginal productivities of the two factors
are the same in each line of production.
Under the assumption of homogeneousproduction functions in two inputs,
the contract curve must have the shape indicated in our figure. On the contract
curve we have indicated points M and N correspondingto the situation before
and after trade. It can now be shown graphically how the following somewhat
paradoxical statement can be true: even though the proportion of total
capital to total labour remains the same in both lines together, nevertheless the
introduction of trade lowers the proportionof capital to labour in each line, and
the prohibition of trade, as by a tariff, necessarily raises the proportion of
capital to labour in each industry. Although it seems intuitively anomalous, it
is graphically clear from the diagram that a movement from N to M raises the
proportion of capital to labour in watches, the total proportions remaining
unchanged as indicated by the box. The proportion of labour to capital in the
production of wheat with trade is indicated by the slope of the angle of the
dotted line going between N and the wheat origin. A similar dotted line between
the same origin and M shows the proportion of labour to capital in the pro-
duction of wheat after trade. Its being less steep than the other makes it ciear
that the ratio of capital to labour has increased. Utilising similar dotted lines
between the watch origin and the points M and N, it is likewise seen that the
abolition of trade increasesthe proportionof capital to labour in the production
of watches.
How can we reconcile the graphical result with our numerical intuition
which tells us that when each of two quantities goes up, an average of them
cannot remain constant ? An examination of the exact relationship between
the proportionsof capital to labour in each line and the proportionsin both at
once dispels the paradox. The proportion in both is found to be not a simple

average but a weighted arithmetic mean of the proportionsin each as indicated
by the following identity:
La Ca Lb Cb C
L La L Lb L ...................(6)
The weights are simply the proportions of the total labour supply used in the
respective industries. The abolition of trade raises the proportionof capital to
labour in each line, but at the same time through the reverse operation of the
principle of comparative advantage automatically gives more weight to the
industry which uses the lesser amount of capital to labour.
Thus, we have shown conclusively that a restriction of trade will increase
the proportionof capital to labour in both lines. It follows necessarily that the
real wage in terms of each commodity must increase regardless of any move-
ments of prices of the consumer's goods. For within each industry increasing
the capital which co-operates with labour raises the marginal productivity of
labour expressed in physical units of that good. Not only are the labourersof
that industry better off with respect to that good, but by the equivalence of
real wages everywhere (expressed in terms of any good) labour in general is
better off in terms of that good. If the real wage in terms of every good
increases, we can definitely state that real income has increased. This is one
of the few cases in economic analysis where a given change moves all relevant
magnitudes in the same direction and obviates the necessity of a difficult, and
often indefinite, index number comparison.
Under the assumed conditions-(a) two commodities, (b) producedby two
factors of production,and (c) where trade leaves something of both commodities
produced but at a new margin-it has been unequivocally demonstrated that
the scarce factor must be harmed absolutely. This is in contrast to the accepted
doctrine which may be fairly representedas saying that trade mightconceivably
affect adversely the relative share of a factor, but cannot be expected to harm
absolutely an important factor of production. Not only is the latter possible,
but under the posited conditions it follows necessarily.
THREE OR MORE COMMODITIES

If the above conclusion held only for two commodities, its interest even
for theory would be limited. It is of interest to show, therefore,that the intro-
duction of any number of commodities in no way detracts from the validity of
our conclusions. Of course, no simple graphical device can be used to portray
this because of the increased number of variables.
One method of approaching the problem might be to arrange the com-
modities in a sequence according to the relative importance of labour in each.
This is not unlike the ordering of commodities long used by Mangoldt,l Edge-
worth, and others to explain which commodities will be imported and which
exported when more than two commodities are introduced into the classical
theory of comparative advantage. In our case, however, costs are not constant
and are not expressible in a single homogeneous unit of a factor or in a given
composite factor.
For the present purpose one need not rely upon such a construction, but

need only realise that the introduction of trade will increase the production of
those commodities which use relatively much of the abundant factor, and will
lower the production of the commodities using relatively little of the abundant
factor. Accompanyingthis, there will be the familiarHeckscher-Ohlintendency
towards partialequalisationof factor prices in the two countries, the price of the
scarce factor falling in relationship to the price of the abundant factor. By
itself this tells us nothing concerningthe absolute burden or benefit from trade,
but deals only with the effect upon relative shares. We cannot simply infer
from this anything concerning the behaviour of absolute shares. For it is not
as if international trade leaves the total amount of real national income un-
changed so that the more one factor receives, the less there will be left for the
other. On the contrary, it has been shown elsewhere that trade must increase
the national income under the conditions here postulated.
It is nevertheless true that the introduction of trade will harm absolutely
the scarce factor of production. To demonstrate this we must recall the fact
that at the new higher relative price of capital to labour there will inevitably
be a relativesubstitution of labour for capital in each line of production. In
exactly the same way a restriction upon trade will raise the price of the scarce
factor, labour, relative to the abundant factor, capital. There is nothing
paradoxical in the fact that the ratio of capital to labour can increase in every
line, while the ratio of total capital to total labour remains constant. The
explanation given in the two commodity case whereby the weights in the
arithmetic mean change in an appropriate fashion holds without modification
when there are any number of commodities.
It is now a simple matter to show that the physical marginal productivity
of labour in each line must increase, and because of the equalisation of wages in
all lines, expressed in terms of any commodity, it immediately follows that
restrictionof trade increasesthe real wage of workersexpressedin terms of each
and every commodity. This obviates the necessity for any index number
comparisonor for any consideration of the worsening of the terms of trade.
THE CASE OF COMPLETE SPECIALISATION
The reader of the above argument will have realised that its remarkable
simplicity springs from the fact that we may infer the real wage of workers in
terms of a given good from the real marginal physical productivity of those
workerswho produce that good. This requiresthat before and after trade some
finite amount, however small, be produced of every good. In a world where
technological conditions are conducive towards the maintenance of the state of
pure competition implicit in all our previous argument, this is perhaps not too
unrealistic an assumption. However, it is still desirable to see what remains of
the argument when this assumption is dropped. This-is even more so because
in the course of the argument it will be shown that the classical theory was not
so much incorrect as limited in scope.
Provided that costs are not constant, and that something of both goods was
previously consumed, at first price changes brought about by international
trade will shift the margin of production,but will still leave some productionof
1 J. Viner, op. cit., p. 458; G. Haberler, op. cit., pp. I36-I40.

PROTECTION AND REAL WAGES 7I
both commodities. At one crucial price ratio correspondingto the slope of the
tangent at R in Fig. i the production of one of the commodities will cease
completely, and further changes will not alter the specialisation. Up until the
critical price ratio is reached, the introduction of trade worsens the position of
labour according to the previous arguments. But what happens after this
critical price ratio ?
There is no essential loss of generality in considering the two commodity
case. For the commodity which is still producedthe real wage is determinedas
before by the physical productivity of the workers in that line. Up until the
critical price ratio at which complete specialisation takes place, the scarce
labour factors have been shown to lose. Beyond this critical price ratio their
physical productivities remain unchanged. It is clear, therefore, that the real
wage in terms of the good using little labour is necessarily harmed by the
introduction of trade.
With respect to the other commodity the matter is more complicated, and
the final result is indeterminate. Up to the critical price ratio we know that
the real wage in terms of this commodity must fall. But after specialisa-tion,
the level of real wages cannot be determined by the productivity of workersin
this line since there are no such workers. One cannot avoid bringing into the
analysis the price ratio between the two consumers'goods, that is, the terms of
trade. Given this price ratio, it is possible to convert real wages in terms of one
commodity into real wages in terms of the other. It becomes apparent that
beyond the critical point the real wage in terms of the non-produced,imported
good must begin to increase. This is to be balanced against the loss of real
wages in terms of this good wvhichtook place before the critical point was
reached. Whether the result will be on balance favourable or unfavourable
cannot possibly be determinedon a priori grounds,but rests,upon the technolo-
gical and economic features of the countries in question. Even if in a limited
number of cases we could determinethat the real wage in terms-ofthe imported
good would increase, there would still be involved a problem of weighing
against this the demonstratedloss in real wages expressed in terms of the good
in which the country has a comparative advantage. Here again the final result
would be indeterminate, although in favourable cases an index number com-
parison might be decisive.
Applying this same line of reasoning to the constant cost case of the
classical theory of international trade, it is seen that theirs is one of the special
unambiguous cases. Either a single factor of production or a never varying
composite dose of factors is assumed. Because of constant costs the slightest
change in the price ratio of the goods will lead instantaneously to complete
specialisation. There results no shifting of the proportions of the factors, and
hence no deterioration of wages in terms of either good. On the contrary, in
terms of the imported good there must be an improvementin real wages with a
consequent increase in real income. This is made intuitively obvious from the
considerationthat trade necessarily increases the real income of a country, and
in the classical case the proportion of income going to the respective factors
cannot be changed by trade. It is the latter feature of the classical theory which
constitutes one of its important short-comings.

MORE THAN TWO FACTORS
One by one we have been able to drop our various restrictive assumptions
with only slight modifications of results. Still there remains the problem of
introducinginto the analysis more than two productive factors. Unfortunately,
this entails more serious consequences.
In the first place, the definiteness of the Heckscher-Ohlintheorem begins
to fade. With three or more factors of production it is certainly not necessary
that the result of trade is to make the ratios of factor prices in the respective
countries more closely approach unity. Some may do so, but others may
diverge depending upon complicated patterns of complementarity and com-
petitiveness.' Whether on balance the movement towards equalisation exceeds
the tendency towards diversification is not a meaningful question until a non-
arbitrary method of weighting these changes is specified. Furthermore, even
the concepts of scarce and abundant factors lose their sharpnessof definition.
The fact that the Heckscher-Ohlin theorem breaks down when many
factors of productionare involved affordsan explanation of its failureto account
for the facts if the production functions in the two countries differ, or if the factors
of production of different countries are not identical. By appropriate termino-
logical conventions it is always possible to attribute differences in the pro-
duction functions to differences in amounts of some factors of production
(knowledge, available free factors, etc.). Similarly, if the factors of production
of different countries are regarded as non-comparableand incommensurable,2
this can be classified as an extreme case of factor disproportionality,but there
must be more than two factors. We conclude, therefore, that the Heckscher-
Ohlin theorem does not necessarelyhold in the case of constant costs or multiple
factors of production.
It does not follow that our results stand and fall with the Heckscher-
Ohlin theorem. Our analysis neglected the other country completely. If
factors of production are not comparable between countries, or if production
functions differ, nevertheless, so long as the country has only two factors,
international trade would necessarily affect the real wage of a factor in the
same direction as its relative remuneration.3 The only loss to our analysis
would be the possibility of labelling the factor which is harmed as the " scarce "
(relative to the other country) one.
However, we must admit that three or more factors of production within
a single country do seriously modify the inevitability of our conclusions. It is
not only that the relatively scarce factor can be defined only circularly as the
one whose price falls most after trade, but even if we do know the behaviour
See Ohlin, op. cit., pp. 96-105 and parsim.
2 If the extreme classical assumption of immobility of labour between countries were valid,
then over time the working populations of the various countries would become differentiated
culturally, genetically, and in the limit cease to be of the same species. But those in the narrower
classical tradition are least in a position to bring this up as an argument against the Heckscher-
Ohlin theory, for in expositing the comparative cost doctrine they repeatedly (and sometimes
unnecessarily) compare labour (costs, productivities, hierarchies, etc.) in various countries.
3 This is in contrast to the problem of the effect of a technological innovation to which
Professor Haberler (op. cit., p. I95) has compared the effects of trade. Technological change
shifts the production function, and no inferences concerning the new marginal productivity
relationships are possible. As we have shown, trade leads to definite effects.

of relative factor prices, i.e. relative shares in the national income, it seems that
we cannot infer unambiguously that the physical marginal productivities move
in the same direction. Even though these continue to depend only upon the
proportionsof the factors in the respective industry, diverse patterns of comple-
mentarity and competitiveness emerge as possibilities. It is outside the scope of
the present paper to attempt a catalogue of the various conceivable permuta-
tions and combinations.
This lack of definiteness in the more complex case is typical of attempts to
go beyond the level of abstraction current in economic theory. We have
resisted the temptation to lump together diverse factors into two composite
factors and thereby achieve the appearance of versimilitude, although others
may care to do so for some purpose.
CONCLUSION
We have shown that there is a grain of truth in the pauper labour type of
argument for protection. Thus, in Australia, where land may perhaps be said
to be abundant relative to labour, protection might possibly raise the real
income of labour.1 The same may have been true in colonial America. It does
not follow that the American working man to-day would be better off if trade
with, say, the tropics were cut off, because land suitable for growing coffee,
rubber, and bananas is ever scarcer in America than is labour. The bearing of
the many factor case will be obvious.
We are anxious to point out that even in the two factor case our argument
provides no political ammunition for the protectionist. For if effects on the
terms of trade can be disregarded,it has been shown that the harm which free
trade inflicts upon-one factor of production is necessarily less than the gain to
the other. Hence, it is always possible to bribe the suffering factor by subsidy
or other redistributive devices so as to leave all factors better off as a result of
trade.2
WOLFGANG F. STOLPER.
Swarthmore,Penna. PAUL A. SAMUELSON.
Cambridge,Mass.
I See D. B. Copland, "A Neglected Phase of Tariff Controversy," Quarterly Journal of

Economics, 1931, pp. 289-308; K. L. Anderson, "Protection and the Historical Situation:
Australia," Quarterly Journal of Economics, November, I938, pp. 86-I04; M. C. Samuelson,
op. cit., pp. I43-I49.
2 Viner, op. cit., p. 534; P. A. Samuelson, op. cit., p. 204.


Stolper Samuelson 1941

Uploaded by

Copyright:

Available Formats

Stolper Samuelson 1941

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Stolper Samuelson 1941

Uploaded by

Copyright:

Available Formats

The Review of Economic Studies, Ltd.

Protection and Real Wages

This content downloaded from 128.235.251.160 on Thu, 11 Dec 2014 00:11:36 AM

THE EFFECT OF TRADE UPON RELATIVE FACTOR PRICES

This content downloaded from 128.235.251.160 on Thu, 11 Dec 2014 00:11:36 AM

SOME EXISTING VIEWS

This content downloaded from 128.235.251.160 on Thu, 11 Dec 2014 00:11:36 AM

This content downloaded from 128.235.251.160 on Thu, 11 Dec 2014 00:11:36 AM

This content downloaded from 128.235.251.160 on Thu, 11 Dec 2014 00:11:36 AM

ASSUMPTIONS OF THE ANALYSIS

This content downloaded from 128.235.251.160 on Thu, 11 Dec 2014 00:11:36 AM

This content downloaded from 128.235.251.160 on Thu, 11 Dec 2014 00:11:36 AM

This content downloaded from 128.235.251.160 on Thu, 11 Dec 2014 00:11:36 AM

Now in equilibrium the value marginal productivity (expressed in terms

This content downloaded from 128.235.251.160 on Thu, 11 Dec 2014 00:11:36 AM

This content downloaded from 128.235.251.160 on Thu, 11 Dec 2014 00:11:36 AM

This content downloaded from 128.235.251.160 on Thu, 11 Dec 2014 00:11:36 AM

This content downloaded from 128.235.251.160 on Thu, 11 Dec 2014 00:11:36 AM

THREE OR MORE COMMODITIES

This content downloaded from 128.235.251.160 on Thu, 11 Dec 2014 00:11:36 AM

This content downloaded from 128.235.251.160 on Thu, 11 Dec 2014 00:11:36 AM

This content downloaded from 128.235.251.160 on Thu, 11 Dec 2014 00:11:36 AM

This content downloaded from 128.235.251.160 on Thu, 11 Dec 2014 00:11:36 AM

I See D. B. Copland, "A Neglected Phase of Tariff Controversy," Quarterly Journal of

This content downloaded from 128.235.251.160 on Thu, 11 Dec 2014 00:11:36 AM

You might also like