/4azfl

Public Disclosure Authorized

THE WORLD BANK ECONOMIC REVIEW, VOL. 4, NO. 1: 55-79

FILE CO!PY
""/J
Effects of Macroeconomic Policies
on Sectoral Prices
Yair Mundlak, Domingo Cavallo, and Roberto Domenech

The effectof macroeconomic policieson the relativepricesof internationallytraded

Public Disclosure Authorized

and domesticgoods has beetnthe subjectof extensivestudy. Analysisof the way in

which these policiesthen affect pricesat the sectorallevelis complicatedby the
heterogeneityof sectoralproduction:even the pricesof singleproductsusuallyare
determinedby both domesticand tradedcomponents.Wepresenta frameworkwhich
first tracesthe influenceof macropolicyon the relativepricesof exports,imports,and
homegoods.It thenaccountsfor eachsector'sdegreeof "tradability,"whichis based
on the importanceof tradein sectoralincome,and the influenceof macroeconomic
policyon sectoralprices. To illustratethe use of this approach,it is appliedto a
simulationof tradeliberalizationin Argentina.Our resultssuggestthat economywide
policieshad substantial negativeeffects on both the real exchangerate and the
incentivesto agriculturalexports.
Public Disclosure Authorized

In an open economy, the prices of tradable products are determined by world

prices, nominal exchange rates, and taxes. The prices of products which are
not tradable are determined by domestic supply and demand, which are them-
selves influenced by the actions and policies of the government. Some very
important economic decisions depend on the price of tradables relative to that
of nontradables-the real exchange rate. We examine here the way in which
broader government policies affect the real exchange rate, and through it,
prices at the sectoral level.
Analyses of the real exchange rate generally aggregate all production into
two sectors, tradables and nontradables. This aggregation simplifies the discus-
sion and helps illuminate some important issues, but it has limited empirical
relevance: there are no products which can be classified as purely tradables or
nontradables. To illustrate, a television set is a tradable product, but the price
of a television set quoted in a department store in the Ginza district of Tokyo
Public Disclosure Authorized

reflects inputs, such as location, which are not tradable. Thus, if we are to

Yair Mundlak is a professor at the Universityof Chicago and a researchfellow at the International
Food Policy Research Institute. Domingo Cavallo and Roberto Domenech are economists at Instituto
de Estudios Economicos sobre la Realidad Argentina y Latinoamericana, Fundacion Mediteranea. In
revisingthe paper, the authors benefited from commentsby Maurice Schiffand the referees.
© 1990 The International Bank for Reconstruction and Development/ THE WORLD BANK.

55
56 THE WORLD BANK ECONOMIC REVIEW, VOL. 4, NO. I

understandl price differentials over time, across sectors or countries, we require

a measure of the share of the tradable component in the price of a product.,
This measure is useful in evaluating the response of sectoral prices to policies
which are not sector-specific.
We apply this approach in an evaluation of the relative effects of Argentina";
currency overvaluation on agriculture and nonagriculture, first analyzing the
determinants of the real exchange rate and then relating the real exchange rate
to sectoral prices. The structural relationships depend on the degree of open
ness of the economy, which is taken into account.
Time series data for 1913-84 are used to estimate real agricultural and
nonagricultural prices, the real exchange rate, and proxies for the degree oil
openness of the economy. On the basis of these estimates, we then simulate the
effect of policy changes that would make the economy more stable and more
competitive in world markets.

I. THE REAL RATE OF EXCHANGE

Modeling Commercial Policy

Much of the empirical work on the effects of tariffs on the real exchange rate
has followed the framework of Dornbusch (1974), which serves as a point of
departure for this analysis. The economy is divided into three sectors: export-
ables (x), importables (m), and home goods (h). It is assumed that Argentina
can be treated as a small open economy in the sense that it is a price taker in
world markets. In this case, the prices of the two traded goods, Pj, are deter-
mined by the world price, P,;, the nominal exchange rate E (expressed in units
of domestic currency per unit of foreign currency), and the trade tax, Tj = (1
+ tj), where tj is the tax rate, which is positive for imports and negative for
exports.
(1) Pj = P*ETJ, j=x,m
The domestic supply and demand of each of the two traded goods need not be
equal because the gap is closed by trade. But for the home good, domestic
supply and demand are equalized through the adjustment of P.. Thus reflecting
market clearing in the home goods sector, we set its excess supply function to
zero to obtain the following implicit function:

(2) I (t P.) =

Under weak conditions equation 2 can be differentiated logarithmically to

yield:

(3) d ln (p) = m (d In Px-d InPm)

Amh + A~

where Ajis the elasticity of excess supply of the home good with respect to the
price of thejth tradable good. We integrateequation 3, write c = Am / ( Am +
 Mundlak, Cavallo, and Domenech 57

A.), decompose the price of the tradables into world price (P*) and taxes, T,
and label T = Tx / Tm to obtain:

(4) In (p)= a + coIn (p)+ X In T

where X is the elasticity of the price of exportables (measured in terms of the

domestic product) with respect to the terms of trade (the price of exportables
in terms of importables). It should be noted that cocan vary: it is not necessarily
a constant as assumed in empirical studies. Assumed constancy may produce a
good local approximation for marginal changes but may be too restrictive when
the data reflect big changes.
Equation 4 expresses the determination of the price of exportables in terms
of the home good. It is positively related to the terms of trade, PI/P,, and
negatively related to the two trade taxes, tmand t,. The converse is true for the
price of importables, which is obtained by rearranging terms:

(4') In(PE) -(1(a ) Iln (p) - (1 - w) ln T

Thus, both PIP h and PmIPh constitute measures of the real exchange rate,
but they behave differently in response to foreign terms of trade or taxes. A
more conventional measure of the real exchange rate, e, is obtained by aggre-
gating these two measures using the geometric averages of the foreign prices,
P-, and the taxes, T':
(5) e = P*T*E/Ph
The behavior of e in response to changes in the foreign terms of trade and taxes
depends on the weights used in the aggregation. To demonstrate, let PTF' -
(P:Tj)b(P.T.)I-b, insert this term in equation 5, combine with equation 4, and
rearrange to get:

In e = a - (1 - b - c) In (x)+ (1 - b - c) In Tm - (1 - b - c) In T.

When b = 1 (that is, the foreign price is measured by the export price), e varies
positively with the foreign terms of trade. The opposite is true for the case
where b = 0, (the foreign price is measured by the import price).

Previous Estimations of the Real Exchange Rate Equation for Argentina

Earlier estimates of the real exchange rate equation for Argentina were
obtained by Rodriguez and Sjaastad (1979); Cavallo and Garcia (1985); and
Mundlak, Cavallo, and Domenech (1987). They differ somewhat in the vari-
ables used and the periods of analysis. On the whole the estimated values for
the degree of substitution between imports and home goods, co,were relatively
low (see Sjaastad and Clements 1981 to compare results for some other coun-
tries).
58 THE WORLD BANK ECONOMIC REVIEW, VOL. 4, NO. 1

To see the implications of low values for ci, we rewrite equation 3:

(6) dlnPh = (1 - w)dlnPx + codlnPm
The smaller is w, the closer is the comovement of Ph and Px. This implies that
in Argentina the price of the home good moved more closely with P. than with
Pm. Therefore, changes in t,xhave a dominant influence on home goods prices
when compared with changes in the import tax (as shown in Cavallo and
Mundlak 1982). Furthermore, calculations of the aggregate real exchange rate,
assuming no trade taxes and using the various estimates of X from these studies,
show that the market exchange rate was lower than its actual level. This is in
contrast to the common belief that trade liberalization should increase the real
exchange rate. Our result is a consequence of the low value of w.
It is important to note that in the present model the price of the home good
changes only as a result of changes in the domestic prices of the tradables, but
such changes displace the system from its equilibrium, which can only be
restored by a change in P,. Therefore there is only one way to eliminate
overvaluation of the real exchange rate, and this is by changing taxes on trade.
This follows directly from equation 6, which helps to focus on the role of taxes
but abstracts from other considerations which are important in interpreting the
data. These factors are taken up in the next section.
Extensions: The Role of Macro Policy
The foregoing model is basically a derivative of the Hecksher-Ohlin-Samuel-
son model with a nontradable sector added. As such it assumes a constant-
returns-to-scale technology, full employment, and perfectly competitive factor
markets. Demand is derived from utility maximization of the private sector,
which has a one-period time horizon. Deviations from these assumptions affect
the results. Various aspects of a more general framework are reviewed and
discussed in Dornbusch (1987), Edwards (1988), and Snape (1988).
With taxes and the foreign price of tradables given, anything that affects
domestic prices affects the real exchange rate directly, and also indirectly
through the effect on E. These factors are generated by macro policies and are
related to the relative size of the public sector, fluctuations in its expenditures,
and the methods of financing those expenditures. Trade policies also determine
the openness of the economy. And finally, long-term factors affect the supply
or demand for the various products.
Capital inflows increase the supply of tradables and the level of expenditures;
because all goods are normal (that is, have positive income elasticities), the
demand for nontradables thus increases. Because prices must increase in re-
sponse to the increased demand, the real exchange rate should decline.
While our previous discussion has implicitly assumed that demand consists
only of private consumption, the analysis can be generalized to include invest-
ment. If investment constitutes a different share of the home good, this change
in the composition of expenditures also changes the real exchange rate. The
 Mundlak, Cavallo, and Domenech 59

composition of expenditures becomes more important when the analysis is

extended to include government, which has a different composition of con-
straints on demand and budget than does the private sector. In general, home
goods are a larger share of government than private expenditures.
The effect of government on the real exchange rate is stronger when the
government runs a deficit because of the macroeconomic effects of its means of
financing. When the government borrows to finance a deficit and the economy
is financially open, this results in a capital inflow, which causes a decline in e.
When the economy is financially closed, the borrowing will drive up the rate
of interest and thus reduce private sector expenditures. This change of expen-
diture composition causes a decline in e.
When the deficit is financed by an expansion of the money supply and the
economy is financially closed, the monetary expansion causes an increase in
prices and the expenditure of the private sector is reduced by the inflationary
"tax." Again, because of the change in composition of expenditure in favor of
government, e declines. If the economy is financially open, and the nominal
exchange rate is fixed, however, the monetary expansion will raise private
demand. This is matched in part by a rise in net imports or by an increase in
the capital inflows, causing e to decline. This effect on e is reinforced by the
increase in demand for the home good. The mechanism will change when the
nominal exchange rate is flexible, but nevertheless e declines.
The real exchange rate is also affected by the relative income elasticity of
demand for the home good and tradables. If the demand for the home good is
income elastic, this means that as income increases the demand for home goods
rises relative to the demand for tradables, and therefore Ph rises so that e
declines.
Restrictions on trade modify the adjustment mechanisms of the economy and
therefore the determination of the real exchange rate. Limits on imports, for
example, tend to lower e. In order to incorporate trade restrictions in the
empirical analysis, there must be a way to measure the degree of openness of
the economy (this is discussed below).
Much of the discussion of the determinants and effects of the real exchange
rate is related to short-term variations with resources and technology held
constant, whereas empirical analyses commonly use data that reflect changes
over time. Changes in resources and technology affect the supply of the various
goods differentially. Home goods production is generally thought to be more
labor-intensive, so that capital accumulation reduces the price of the more
capital-intensive tradable sectors, which implies a decrease in e. Changes in
technology may take different forms, which we shall not detail here. The net
effect of such changes can be determined empirically.
Introducing the Macro Variables
Previous estimates of the real exchange rate (equation 4) for Argentina, with
macropolicy variables added, indicate that macroeconomic policy has had an
60 THE WORLD BANK ECONOMIC REVIEW, VOL. 4, NO. I

important effect on the real rate of exchange (see Cavallo and Mundlak 1982,
Cavallo and Garcia 1985, Cavallo 1986, and Mundlak, Cavallo, and Dome-
nech 1987). The main conclusion derived from these studies is that overvalua-
tion of the Argentine currency arose not just from commercial policy but also
from macro and incomes policies. Moreover, these effects were shown to
depend on the structural features of the economy. That has led us to a more
detailed specification of exchange rate determinants, including government con-
sumption (g) and borrowing (f), money growth (,a), and income (Y):

(7) (2)= cJ-Em/I+ EEY

E(Px-Pm)-egg-

where x = d In x. The variable g measures the share of government consump-

tion in total income, fi is the share in total income of the fiscal deficit financed
by borrowing, and t is the proportion of money in nominal income evaluated
in terms of foreign prices and converted to local prices by the nominal exchange
rate: y = M/EP*Y. Thus, 4 measures the rate of growth of the money supply
over and above real growth in gross domestic product (GDP), foreign inflation,
and nominal devaluation. A positive sign for A implies that the monetary
expansion is inflationary. The effect of this variable depends on the velocity of
money, but we have not accounted for this in our analysis.
Total real income, Y, is introduced to reflect changes in the composition of
demand, and variations in resources and technology in production. A further
refinement would eliminate the transitory variations in this variable and allow
us to analyze only the longer-term sources of growth. Relatedly, changes in
sectoral incentives affect the pace of capital accumulation and technical change
(Mundlak, Cavallo, and Domenech 1989a). However, at any point in time the
capital stock and technology are predetermined, and their long-term variations
can be approximated by Y.
We are not interested here in separating out the long-term supply and demand
effects. It should be noted, however, that such structural changes have an effect
on the importance of trade in the economy and therefore on the impact that
various shocks have on the real rate of exchange. For this reason a measure of
the importance of trade is introduced.
Changes in the importance of trade also reflect restrictions on trade and
capital mobility. Such restrictions have an important effect on the prices of the
home good and therefore on the real exchange rate. To allow for such effects,
the coefficients in equation 7 are formulated as linear functions of our proxies
for the degrees of commercial and financial openness. We use a ratio of the
value of trade to total income to measure the openness to trade. Financial
openness is measured as the ratio of the official exchange rate to the black
market rate, e/Eb. Restrictions in commercial and financial markets are inter-
connected, most directly here because import restrictions encourage the growth
of the black market to meet excess demand for imports and for the foreign
 Mundlak, Cavallo, and Domenech 61

exchange they require. It is not hard to think of more ideal measures, but the
problem is the lack of appropriate available data.
We assume that o, and the elasticities of real income (Y) and government
consumption (g), depend only on the share of trade in total income. The
elasticities of the fiscal deficit financed by borrowing (f) and of the money
supply (,) are assumed to depend on both of these openness variables. A
summary of the results appears in table 1.
The elasticity of the real exchange rate with respect to the terms of trade as
reflected in the values of X computed from the regression are plotted in figure
1. The value ranged between 0.6 and 0.7 before 1925, when the economy was
very open to the rest of the world. In that period, the price of the home good
was more closely related to the price of imported goods than to the price of
exports. This reflected a high degree of substitution in production and demand
between the domestic and the imported good. As the restrictions imposed on
imports increased over the following two decades, co declined. The lowest
values are observed in the early 1950s, when the economy was very closed.
Recall that lower values of c mean that the prices of home goods are more
closely related to the domestic price of exports than to prices of imports.
Since the late 1950s, X has oscillated around 0.25. This low value of X
explains why changes in export taxes produce only a small change in the

Table 1. Determinants of the Real Exchange Rate, 1916-84, Argentina

Macroeconomic variable Average value of
(change in) Coefficient the coefficient
Terms of trade: 0.72 + 0.29 log DO, 0.37
dlnPx-dlnfPm (5.1) (2.5)
Real income: d In Y 0.24 0.24
(1.6)
Government 0.43 log DOc -0.52
consumption: d In g (6.7)
Borrowing for fiscal -1.69 - 2.04 log DOf -1.13
deficit financing: ft (3.7) (2.3)
Monetary expansion: d In u -0.44 + 0.2 log DOf -0.45
(5.1) (2.1)
Note: P., P_, and P, are prices of exports, imports, and home goods respectively,with P. and P.
valued inclusive of taxes, at the nominal exchange rate; g is the share of government consumption in
real income;f is borrowing to finance the fiscal deficit, as a share of total income; IL is the ratio of the
money supply to total income in foreign prices valued at the nominal exchange rate-I = MIEP*Y;
DO, is the share of trade in total income; DOf is the ratio of officialto black market exchange rates.
The equation was estimated by ordinary least squares; the dependent variable is (d In P. - d In Pj).
The intercept of the equation is 0.02 with a t-ratio of 1.6; the coefficientof DO, is 1.39 with a t-ratio
of 8.1; R' is 0.87; and the Durbin-Watsonstatistic (DW) is 1.65. Absolute values of the t-ratios are in
parentheses.
a. Definedin absolute terms (as share of total income),not as change.
Source:Mundlak, Cavallo, and Domenech(1989b).
 62 THE WORLD BANK ECONOMIC REVIEW, VOL. 4, NO. 1

Figure 1. Elasticity of the Real Rate of Exchange with Respect to /Pm,

P
1913-84, Argentina

0.80

0.70

0.60

0.50

0.40

0.30

0.20

0.10

0.00
1915 1925 1935 1945 1955 1965 1975

Note: Verticalaxis shows elasticity of the price of exportables with respect to the terms of trade (0)).
Px = price of exports; Pm = price of imports.
Source: Mundlak, Cavallo, and Domenech (1989a).

effective real exchange rate for exports. When t. goes down, the domestic
producer price of exportables, equation 1, increases accordingly. With other
variables held constant, equation 4 indicates that 1 - X of the increase in P. is
transmitted to P,. Thus, with w = 0.25, the price of the home good increases
by 75 percent of the increase in P,. This in turn implies that the real rate of
exchange for exportables, measured as the difference between the rates of
change of the two prices, increases only by 25 percent of the initial increase in
P.. In other words, a 20 percent reduction in the export tax produces only a 5
percent increase in the price of the exported good relative to the price of the
home good.
The intuitive explanation is as follows. When the tax on exports is reduced,
the increased incentive to produce exportable goods induces an increase in
exports and thereby an increase in income. As all goods are assumed to have
positive income elasticities, their demand increases accordingly. That generates
excess demand for the home good and forces its price to increase. Restrictions
on imports cause some of the augmented demand for imports to be diverted to
 Mundlak, Cavallo, and Domenech 63

the home good and thereby generate a further increase in the price of the home
good. The increased price of exportables also reduces the demand for them and
further increases the demand for the home good. As a consequence, domestic
prices increase and the real exchange rate decreases to absorb much of the
initial increase in export prices. It is in this sense that domestic prices move in
line with export prices. Of course, the outcome would be different if imports
were allowed to increase without restriction, that is, if the economy were open.
This suggests that reducing import restrictions would allow more of the income
increase to be absorbed by imports, reducing the pressure on home goods
prices. Therefore a given change in t. would have a stronger effect on the
relative price of exportables vis-a-vis the home good.
Government consumption has a negative effect on the real exchange rate.
This is so because government expenditures have a larger share of nontraded
goods than do private expenditures taxed away and because home goods prices
rise when the substitution between imports and domestic goods is low due to
import restrictions.
The effect of the fiscal deficit financed by borrowing is more pronounced
when the economy is financially open, that is, when there is no black market
premium on foreign exchange. The increase in the deficit requires increased
foreign financing and produces either a decline in the nominal rate of exchange
or an increase in domestic prices, or a combination of both. When domestic
financial markets are completely closed, that is, when the black market pre-
mium is very large, financing the deficit by domestic borrowing produces a
very strong crowding-out effect on private expenditures.
The effect of money supply and nominal exchange rate management also
depends on the openness of the economy. When the economy is financially
open, monetary expansion over and above the value of income valued at
foreign prices affects the real exchange rate with an elasticity of -0.44. This
means that a 10 percent increase in ja produces a 4.4 percent reduction in the
real rate of exchange. The elasticity becomes larger in absolute value when the
economy is more closed to financial transactions with the rest of the world.
This is because financial openness will dampen the real effect of nominal shocks
in the money supply or in the exchange rate since capital inflows or outflows
will respond quickly to such shocks. This dampening effect does not operate
when the flows are obstructed, and a large black market premium is created.

II. SECTORALTRADABILITYAND SECTORALPRICES

The analysis outlined above provides the basis here for evaluating the effect
of macro and trade policies, through the real exchange rate, on sectoral prices.
The various shocks considered above affect the sectoral prices largely because
they affect the relative prices of the tradables, and it is therefore important to
examine this effect first. Having done this, we can now move to the analysis of
the sectoral prices.
64 THE WORLD BANK ECONOMIC REVIEW, VOL. 4, NO. 1

In dealing with sectoral analysis, it should be kept in mind that a sector is

often heterogeneousin that it is importing and exporting at the same time. To
deal with this problem, it is assumed that each sector can be subdividedinto
three subsectors: (i) domestic production of goods actually exported, (ii) do-
mestic production of goods actually imported, and (iii) domesticproduction of
nontraded goods. Thus the aggregateprice for sector i, Pi, can be represented
as a geometricaverageof Px,P, and P,:
(8) Pi = PjxP 'hal-a2 i = 1,2
where sector 1 is agriculture, 2 is nonagriculture, and (x, and (X2 are some
functionsof the quantities in question.
In the case of Argentina, almost no domesticallyproduced agriculturalprod-
ucts are also imported, and nonagriculturalexports are negligible.Incorporat-
ing this into equation 8, the two sectoral pricesare:
Pl (Px)a P2h (Pm)

where a1 indicatesthe share of the traded component and as such constitutes a

measure of the degree of tradability of sector i. Equation 9 relates the two
measures of the real rate 'of exchange to the sectoral prices relative to the price
of the home good.
The degree of tradability depends on economic variables which generate
changes in supply and demand, but in the first place they should reflect the
degree of openness of each sector. We accomplish this by allowing ai to depend
on the share of total trade in sectoral income (DO,):
(i = a,' + fi In (DO,)
The prices P,, P2, P., and Pm are observed, but by the very fact that the home
sector is not well defined, there are no direct observations on Pih. There are
data on the price index of government services, P3 . The empirical analysis is
carried out under the assumption that whatever the "correct" Pih is, it is related
to P3 and that this relation depends on the aforementioned macropolicies which
affect the demand for and supply of domestic goods. The following specifica-
tion is used:

(10) In (pib = hi In (MPi)

where hi is a vector of coefficients to be estimated and MPI denotes a vector of

macropolicy variables. Combining equations 9 and 10, an estimable function
is obtained for the relation of the price of sector i to our proxy for home goods
prices:

(11) In (p) (X
a In (p'T) + (1 - ii)hiIn (MPI)
 Mundlak, Cavallo, and Domenech 65

where P, is equal to P. for i = 1 and P,, for i = 2.

Equation 11 was estimated for sectors 1 and 2, using ordinary least squares
(OLS) on first differences. The results of this analysis are summarized by plotting
the estimates for the shares of the traded component, ao in figure 2.
In Argentina before 1930 the traded component of agriculture oscillated
around 75 percent, while that of nonagriculture was about 55 percent. These
were the highest values of a in both sectors and reflected the existence of an
open trade regime. From that year until the beginning of the 1950s, the share
of the traded component declined as the trade restrictions grew. This trend was
briefly interrupted in the years immediately following World War II, mainly as
a result of the extraordinary boom in world trade at a time when Argentina
had exceptionally high levels of grain stocks. From 1947 to 1954 the as reached
their lowest values. After 1955, the share of exports in agricultural output
grew, and by the 1980s the composition was similar to that which had pre-

Figure 2. Sectoral Degree of Tradability, 1913-84, Argentina

Share
0.90

0.80

0.70

0.60

0.50

o.4o~~
0.40 . . . . . . . . ' :
1915 1925 1935 1945 1955 1965 1975

-... nonagriculture (excluding government); agriculture.

Note: This is the share of trade in sectoral output.
Source: Mundlak, Cavallo, and Domenech (1989a).
66 THE WORLD BANK ECONOMIC REVIEW, VOL. 4, NO. I

vailed before 1930. However, traded nonagriculture output remained low:

since 1955 it has been about 42 percent.

III. THE DEGREEOFCOMMERCIAL

OPENNESS

The degree of openness reflects government decisions and world market:

conditions and as such it is exogenous in this framework. However, our meas-
ure of openness depends on endogenous variables and our empirical analysis
accounts for this.
Commercial openness is measured here as the share of total trade in total
income (plotted in figure 3). Note the significant reduction in the relative
importance of trade that took place after the Great.Depression. Government
policies were implemented to attenuate the effects of the world depression and
were similar to policies adopted by most other countries. They included high
taxes on foreign trade, quantitative restrictions on imports and controls on

Figure 3. Indicator of the Degree of Commercial Openness,

1913-84, Argentina
Ratio
1.25

1.00

0.75

0.50

0.25 . . : 4. X

0.00
1915 1925 1935 1945 1955 1965 1975

Key .. fitted; actual.

Note: This is the ratio of total trade to total income.
Source: MundL*, Cavallo, and Domenech (1989a).
 Mundlak, Cavallo,and Domenech 67

foreign exchange, and increasing government expenditures and fiscal deficits.

In Argentina, however, this declining trend in trade continued up to 1955,
except during 1946-47, when high world demand for Argentine exports in-
creased the value of trade to about 40 percent of total income. Despite the
postwar revival of world trade, Argentina increased its restrictions and the
value of trade reached its nadir at about 20 percent during 1952-55. Since
1956 this value has oscillated between 20 and 25 percent.
During the postwar period macroeconomic policy was characterized by higher
government expenditures, higher fiscal deficits, and increased volatility in the
rate of monetary expansion. Stricter restrictions on financial transactions with
the rest of the world were imposed, and commercial policy relied more heavily
on quantitative restrictions than on taxation.
This review of the historical experience suggests that the degree of commer-
cial openness (DO,) may depend on commercial policy, the degree of financial
openness, DOf, the foreign terms of trade, and perhaps other determinants.
More formally:

(12) DO, = f(commercial policy, DOf, ..

The lagged value of DO, is included to represent the more permanent struc-
tural changes that affect trade. To estimate equation 12, it is necessary to
distinguish between trade taxes and quantitative restrictions. Because no annual
data are available for the quantitative restrictions, however, macropolicy indi-
cators are introduced in the empirical equation to capture their effects. Foreign
terms of trade were not significant and were eliminated from the equation.

IV. SIMULTANEOUSESTIMATION

It is now possible to assemble the equations for the degree of commercial

openness, the real exchange rate, the relative prices for agriculture and non-
agriculture (excluding government), and to build a system that is estimated
simultaneously using three-stage least squares. The results are reported in the
appendix, and in general they are very similar to the OLS estimates. The values
based on the static simulations of relative prices fit the data very closely (figures
4-6). Because policy shocks change the dynamic paths of prices, however, in
evaluating policy changes, dynamic simulations are used. Those are shown as
the base run in figures 10-13 below.

V. SIMULATIONOF A TRADE LIBERALIZATIONPROGRAM

The system is now used to simulate the response of sectoral prices to a

program of trade liberalization that is implemented with consistent macroeco-
nomic policies. The attempt to open the Argentine economy in the late 1970s
68 THE WORLD BANK ECONOMIC REVIEW, VOL. 4, NO. 1

Figure 4. TheReal Excbange Rate for Exports, 1913-84, Argentina

Index

1.50

1.25

1.00

0.75

0.50

0.251
1915 1925 1935 1945 1955 1965 1975

K-y .......... fitted; ~ ~actual.

Source: MundIak Cavallo,and Domenech (1989a).

failed mostly because of the inconsistent and inappropriate policies that were
followed (Cavallo and Cottani 1986).
The trade liberalization exercise is carried out for a limited set of commercial
and macroeconomic policies. Modifications in commercial policy are intro-
duced into the system in the year 1930. They consist of complete elimination
of export taxes (T. = 1) and imposition of a 10 percent import tariff (Tm
1.1); the actual values are plotted in figure 7. For fiscal policy, it is assumed
that public expenditures followed their historical levels except for two actual
nonsustainable jumps: a smooth increase in the growth of expenditures between
1946 and 19S3, and a jump to a constant level from 1973 on (figure 8).
Eliminating these two sharp rises in public expenditures reduces the simu-
lated deficit; we assume by the amount of the expenditure cuts. We then allow
borrowing to decline by an equal amount so that the level financed by moneti-
zation remains unchanged (figure 9).
 Mundlak, Cavallo, and Domenech 69

Figure 5. 7he Relative Price of Agriculture, 1913-84, Argentina

index
1.25

1.05

0.85

0.65

0.45

0.25
1915 1925 1935 1945 1955 1965 1975

Ky ............
fitted; actual.
Source: Mundlak,Cavallo,and Domenech (1989a).

We hold the rate of change of It at its average level for the 1930-84 period:
-0.008. We assume that the system is financially open so that there is no black
market premium on the exchange rate.
We compare the simulated values of our measures of commercial openness,
the real exchange rate, and sectoral prices with the base run values (figures 10-
13). As can be seen, all the relative prices respond strongly to trade liberaliza-
tion. This response is quantified in table 2, where the increases in the "free-
trade" values relative to the actual values are reported.
These results imply that if the Argentine economy had been more integrated
with the world economy after 1929, the relative volume of trade would have
been almost 70 percent higher than its actual level. Moreover, domestic relative
70 THE WORLD BANK ECONOMIC REVIEW, VOL. 4, NO. I

Figure 6. 7he Relative Price of Nonagnculture, 1913-84, Argentina

index
1.30

1.20

1.10

1.00

0.90 *

0.80

0.70

0.60

0.50

0.40
1915 1925 1935 1945 1955 1965 1975

Key:. fitted; actual.

Source: Mundlak,Cavallo,and Domenech (1989a)

prices would have been more in line with international prices, implying much
greater price incentives for both agriculture and nonagriculture. For the period
1930-84, the price of agriculture would have been, on average, 40 percent
higher, and the price of private nonagriculture would have been almost 20
percent higher relative to our measure of home goods prices, P3 A greater
supply of agricultural and nonagricultural goods might have dampened some-
what the changes in relative prices, but this would not change the general
pattern. Finally, as it is shown elsewhere, these changes in sectoral prices have
a very substantive positive effect on sectoral and overall growth (Mundlak,
Cavallo, and Domenech 1989a).
 Mundlak, Cavallo, and Domenech 71

Figure 7. Index of Trade Taxes, 1913-84, Argentina

Index
1.40

1.30 *,

1.20

1.10

1.00

0.90

0.80 . . . . ,, . . . _ . . . . . . .
1915 1925 1935 1945 1955 1965 1975

Key. -...........-- + tm; tm =ux on imports. 1 - tx; t, = tax on exports.

Source: Mundlak, Cavallo, and Domenech (1989a).

VI. CONCLUSIONS

A framework has been developed for evaluating the effect of macroeconomic

and trade policies on sectoral incentives. Variations in the prices of home goods
affect the real exchange rate, and through it, sectoral prices, according to their
relative importance in sectoral output, or simply the degree of tradability.
We extend the standard model of the effect of tariffs on the real exchange
rate to include the effects of government consumption, borrowing to finance
the fiscal deficit, changes in the money supply, and income growth, which
reflects capital accumulation and technical changes on the supply side, and
72 THE WORLD BANK ECONOMIC REVIEW, VOL. 4, NO. 1

Figure 8. Government Expenditures, Actual and Imposed Values,

1913-84, Argentina

Share
0.30

0.25

0.20

0.15

0.10

0.05

0.00
1915 1925 1935 1945 1955 1965 1975

Key ... imposed; actual.

Note: This is government consumption as a proportion of total income.
Source: Mundlak, Cavallo, and Domenech (1989a).

changes in demand composition. The effects of these variables depend on the

restrictions on commercial and financial transactions. To reflect these elements,
we include a measure of the value of trade in total income, and the ratio of the
official to black market exchange rates.
Under this structure the elasticity of the real exchange rate with respect to
the terms of trade is higher under a more open regime and lower when the
possibilities for substitution between home and traded goods are limited.
While this framework provides insights into the relations between some
macroeconomic policies and the real exchange rate, their influence on sectoral
prices is obscured by the heterogeneity of production even within relatively
 Mundlak, Cavallo, and Domenech 73

Figure 9. Debt-Financed Fiscal Deficits, Actual and Imposed Values,

1913-84, Argentina

Share
0.15

0.10

0.05

0.00

-0.05
1915 1925 1935 1945 1955 1965 1975

Key .. iaposed; actual.

Note: This is the fiscal deficit financedby borrowing as a proportion of income.
Source: Mundlak,Cavallo,and Domenech (1989a).

disaggregated product groups. Most product groups have both traded and
nontraded components and therefore are affected by changes in the real ex-
change rate. This allows us to measure the degree of tradability from the
relation of sectoral prices and the real exchange rate. This relation depends on
the openness of the sector to trade, indicated here by the share of trade in
sectoral income.
We applied this approach to an evaluation of the consequences of macroeco-
nomic policy in Argentina from 1913 to 1984. To assess the extent to which a
more open trade regime and restrained macropolicies would affect sectoral
prices, we simulated a policy of low uniform tariffs on imports and elimination
74 THE WORLD BANK ECONOMIC REVIEW, VOL. 4, NO. 1

Figure 10. Degree of Commercial Openness under Simulated Wwade

Liberalization, 1913-84, Argentina
Index
1.00

0.80

0.60

o.40

.. . .~~~~~............

0.20

0.00
1915 1925 1935 1945 1955 1965 1975

ey .. simulated; base run.

Note: This is the share of total trade in total income.
Source: Mundlak,Cavallo,and Domenech (1989a).

of export taxes from 1930 on, combined with changes in the macro variables.
The counterfactual analysis suggests that such policies would have increased
incentives to agricultural and nonagricultural production by nearly 40 and 20
percent, respectively. As a result, the volume of trade would have been almost
70 percent higher.
Such changes in incentives are of importance because of their powerful effect
on production and growth. Increased sectoral incentives encourage capital ac-
cumulation, intersectoral resource transfers, and the implementation of new
techniques and adoption of new technology. These relations have been exten-
sively studied, and we have evaluated them in detail using the Argentinian
example (see Mundlak, Cavallo, and Domenech 1989a).
The main message is clear. There is, however, a danger that these results will
 Mundlak, Cavallo, and Domenech 7S

Figure 11. TheReal Exchange Rate under Simulated ThadeLiberalization,

1913-84, Argentina
Index
1.25

1.00 V

0.75 .

0.50

0.25
1915 1925 1935 1945 1955 1965 1975

Key. *-..-------wtrade liberalizationscenario; base run.

Source: Mundlak,Cavallo,and Domenech (1989a).

be attributed to some specific conditions which are not widely applicable. The
purpose of the analysis is to derive the results within a framework which is
universally applicable. If there is something which is specific to Argentina it is
that it has had very favorable initial conditions and that its relatively poor
performance can be attributed to its policies. This shows the cost of wrong
policies but at the same time also indicates what are the potential gains from
alternatives which take the long-run consequences into account.
The four equations were estimated by nonlinear three-stage least squares.
The exogenous variables are g, g, DOf, PIlPm, Y', and f. Note that the system
has a recursive structure. DO, is determined only by predetermined variables;
P,/1P3is determined by DO, and the predetermined variables. Finally, sectoral
prices are determined by DOE, P, P3, and predetermined variables. The two
symbols, x and d log x, are used interchangeably.
76 THE WORLD BANK ECONOMIC REVIEW, VOL. 4, NO. 1

Figure 12. The Relative Price of Agriculture under Simulated Trade

Liberalization, 1913-84, Argentina
Index
1.30

1.20

1.10

1.00

0.90

0.80

0.70

o.60

0.50

0.40
1915 1925 1935 1945 1955 1965 1975

.trade.y: liberalization scenario, base run.

Source: Mundlak, Cavallo, and Domenech (1989a).
 Mundlak, Cavallo, and Domenech 77

Figure 13. The Relative Price of Nonagriculture under Simulated Trade

Liberalization, 1913-84, Argentina
index
1.10

0.80

0.50

0.40
1915 1925 1935 1945 1955 1965 1975

Key .. trade liberalizationscenario; base run.

Source: Mundlak,Cavallo,and Domenech (1989a).

Table 2. Simulations of the Response of Relative Prices to Trade

Liberalization, Averages, 1930-84, Argentina
Simulated
Simulated trade liberalization
Variable base run scenario Increase'
Share of trade in total income (DO,) 0.24 0.40 0.67
Real rate of exchange (e) 0.54 0.82 0.52
Relative price of agriculture (P,IP3 ) 0.68 0.95 0.40
Relative price of nonagriculture (P2 IP3) 0.77 0.91 0.18
a. Ratio of trade liberalization to base run values minus 1.
Source: Mundlak, Cavallo, and Domenech(1989b).
78 THE WORLD BANK ECONOMIC REVIEW, VOL. 4, NO. 1

APPENDIX: SIMULTANEOUS ESTIMATES OF THE PRICE SYSTEM

(A-1) log DO, = -0.516 + 0.648 log T - 0.170 log g - 0.590k

(4.2) (4.0) (4.2) (8.3)

+ 0.146 log DOf + 0.770 log DO,(t - 1)

(4.0) (18.1)
R2 = 0.97; D.W. = 1.93

(A-2) log (P.IP3 ) = 0.026 + 0.744 log (PJP,) + 0.349 [log (PIP_) log DOJ]
(1.9) (5.0) (2.7)
+ 0.194k + 0.428 [(logg)(logDO,)] - 1.12f - 1.31f(logDOf) - 0.130A
(1.6) (6.7) (2.5) (1.4) (1.2)
- 0.022D[(logA)(logDOf)]+ 1.88 tOc
(2.1) (.95)
R2 = 0.89; D.W. = 1.59

(A-3) log (P,/P3) = 0.029 + 0.596 log (PX/ P3 ) - 0.756k - 0.360f

(2.5) (6.0) (5.5) (1.2)
+ 0.219 {log(P./P3 ) [logDO, + log (PY/P,Y1 )]} + 0.174k
(2.6) (1.8)
2
R = 0.88; D.W. = 1.97

(A-4) log (P2/P3 ) = 0.023 + 0.355 log (PI/P3) - 0.630k- 0.499f

(2.7) (3.9) (7.2) (2.2)
+ 0.052 {log(P_/P3) [logDO, + log (PY/P2Y2)]} + 0.0804
(1.9) (1.2)
2
R = 0.85; D.W. = 2.13
 Mundlak, Cavallo,and Domenech 79

REFERENCES

Cavallo,Domingo. 1986. "ExchangeRate Overvaluationand Agriculture:The Case of

Argentina." World Bank Latin Americaand the Caribbean Country Programs De-
partment II. WashingtonD.C. Processed.
Cavallo, Domingo,and Joaquin Cottani. 1986. "The Timingand Sequencingof Trade
LiberalizationPolicies:The Case of Argentina."WorldBank CountryPolicyDepart-
ment. Washington,D.C. Processed.
Cavallo, Domingo, and Raul Garcia. 1985. "PoliticasMacroeconomicasy Tipo de
CambioReal." Paperpresentedat the Seminaron MonetaryPolicyand the External
Sector,Central Bank of Argentina,BuenosAires.
Cavallo,Domingo, and YairMundlak. 1982. Agricultureand Economic Growth: The
Caseof Argentina.ResearchReport36. Washington,D.C.: InternationalFood Policy
ResearchInstitute.
Dombusch, Rudiger. 1974. "Tariffsand Nontraded Goods." Journal of International
Economics4: 177-8S.
. 1987. "ExchangeRate Economics."EconomicJournal97: 1-18.
Edwards, Sebastian. 1988. "Real Monetary Determinantsof Real ExchangeRate Be-
havior: Theory and Evidencefrom DevelopingCountries." Journal of Economic
Development29: 311-42.
Mundlak, Yair, Domingo Cavallo, and Roberto Domenech. 1987. "Agricultureand
Growth: The Experienceof Argentina,1913-84." Paper presented at International
Food PolicyResearchInstituteSeminaron Trade and MacroeconomicPolicies'Impact
on Agriculture,Annapolis,Md.
. 1989a. Agricultureand EconomicGrowth in Argentina,1913-1984. Research
Report 76. Washington,D.C.: InternationalFood PolicyResearchInstitute.
. 1989b. "Data Supplement."In Agricultureand Economic Growth in Argen-
tina, 1913-1984. ResearchReport 76. Washington,D.C.: InternationalFood Policy
ResearchInstitute.
Rodriguez,Carlos, and Larry Sjaastad. 1979. El Atraso Cambiarioen Argentina:Mito
o Realidad?Centro de Estudios Macroeconomicosde ArgentinaWorkingPaper 2.
BuenosAires.
Sjaastad, Larry A., and KennethW. Clements. 1981. "The Incidenceof Protection:
Theory and Measurement."Departmentof Economics,Universityof Chicago. Chi-
cago. Processed.
Snape, RichardH. 1989. "Real ExchangeRates, Real InterestRates, and Agriculture."
In Allen Maunder and Alberto Valdes, eds., Agricultureand Governments in an
InterdependentWorld. Aldershot,England: GowerPublishing.