The aim of this paper will be to construct a general equilibrium model of an open economy with trade restrictions—in particular, tariffs on imports—and to apply the model empirically to Argentina. This application is especially timely since Argentina, a country with a long history of restrictive trade practices, is currently embarking on an ambitious program of trade liberalization. The aim of this program is essentially twofold: first, to increase the efficiency of domestic industry by forcing it to compete with foreign imports; and second, to alleviate inflation, which until recently was at an annual rate of over 150 per cent, through the fall in import prices that is brought about by a general reduction of tariffs.

Abstract

The aim of this paper will be to construct a general equilibrium model of an open economy with trade restrictions—in particular, tariffs on imports—and to apply the model empirically to Argentina. This application is especially timely since Argentina, a country with a long history of restrictive trade practices, is currently embarking on an ambitious program of trade liberalization. The aim of this program is essentially twofold: first, to increase the efficiency of domestic industry by forcing it to compete with foreign imports; and second, to alleviate inflation, which until recently was at an annual rate of over 150 per cent, through the fall in import prices that is brought about by a general reduction of tariffs.

The aim of this paper will be to construct a general equilibrium model of an open economy with trade restrictions—in particular, tariffs on imports—and to apply the model empirically to Argentina. This application is especially timely since Argentina, a country with a long history of restrictive trade practices, is currently embarking on an ambitious program of trade liberalization. The aim of this program is essentially twofold: first, to increase the efficiency of domestic industry by forcing it to compete with foreign imports; and second, to alleviate inflation, which until recently was at an annual rate of over 150 per cent, through the fall in import prices that is brought about by a general reduction of tariffs.

A liberalization of the trade regime entails, however, certain problems, one of which is a deterioration in the balance of payments brought about by increased imports and because the decline in the profitability of domestic industry induced by the relaxation of tariffs makes investment abroad relatively more attractive. The primary aim of this study is to estimate the deterioration in the balance of payments caused by a selective trade liberalization and to determine what changes in government policies might be undertaken to counter these effects. The government may also be concerned with the impact of the new trade regime on certain domestic industries, which may be forced to cut their production significantly in response to increased import competition. An additional concern is that this decline in domestic production may lead to an erosion in the domestic tax base that more than compensates for any gains in tariff revenue brought about by the increase in imports.

The theoretical model used for this analysis is related to an earlier paper, Feltenstein (1979 a), in which a general equilibrium model was constructed of an open economy with import tariffs and quotas. The previous paper, however, considered a Walrasian relative price system in which there was no domestic money, making it impossible to include a price level or government monetary policies in the model. In the present study, money is introduced but is given only limited roles as a store of value and as a necessary element in foreign transactions; transactions costs are not included in the model. 1 Using money as a store of value allows us to consider the wealth effects brought about by changes in the balance of payments, and hence in the money supply. In addition, the presence of money allows us to introduce two important government policy parameters that were absent from the previous paper. The first of these is the nominal exchange rate, which may now be included as an exogenous variable in the model, so that experiments can be carried out in which the government changes the nominal exchange rate in response to other changes in the economy. 2 Outcomes similar to those produced by changing the nominal rate may also be produced by changes in the domestic price level, which may occur when the second government policy parameter, domestic credit expansion, is introduced in the model. 3 This study will assume that this domestically created money is exogenous to the model and exists at the beginning of the period in question, so that alternative domestic monetary policies, which affect the price level and hence the balance of payments, can be represented by changing the initial monetary stock.

The addition of domestic money to the model’s structure has the further advantage of permitting the existence of a capital account and, hence, of an overall balance of payments. Also, since domestic money is included, there will be a well-defined domestic price level, which will change as the parameters of the model change, leading to an endogenously determined rate of inflation. If the demand for money is determined by an adaptive expectations scheme, then the successive rates of inflation will allow us to determine new parameters in a demand-for-money function for each period, thus endogenizing the demand for money when the model is placed in a multiperiod setting.

Although the structure of the model is essentially that of the Arrow-Debreu model, 4 its workings with respect to the external sector are designed in accordance with the monetary approach to the balance of payments. At the beginning of the period, consumers have certain holdings of capital, labor, and foreign exchange, along with holdings of the domestically created stock of money, while the rest of the world holds a certain stock of money (domestic currency) and foreign exchange. There are also a number of intermediate and final goods, which may be produced domestically, imported, or exported. At an arbitrary set of prices, there will not be equality between the supply of, and the demand for, these goods and financial assets. We will, however, derive a set of prices at which all markets clear. If, at this equilibrium, consumers desire to hold less domestic money than they did at the beginning of the period, a negative balance of payments will allow this change to take place; if, on the other hand, the rest of the world wishes to change part of its foreign exchange holdings into domestic currency, perhaps to take advantage of relative interest rates, then the change will be realized through inflows in the capital account.

In the next section, we will give a brief description of the mechanical structure of the model, while the second section will describe the introduction of money into the model. The third section will discuss possible government policy parameters, while the final section will be concerned with empirical results and simulations for Argentina.

I. Mechanics of the Model

Computational general equilibrium models are becoming increasingly popular among economists wishing to analyze the impact of discrete changes in parameters. 5 The major advantage of such models, compared with those of traditional macroeconomic analysis, is that they allow the consideration of disaggregated commodities and, hence, a precise examination of relative price effects. When, however, the user is confronted by the necessity of using empirical data from real economies in order to construct models that are intended to have some policy relevance, he is immediately confronted by a major technical problem. The time needed by a computer to solve a general equilibrium model via the algorithms that are currently in existence tends to rise geometrically with the number of commodities involved; and by the time there are approximately 25 goods, the time needed may be more than a half hour. Usually, the most precise disaggregated commodity information that is available for a country’s technology in a form suitable for use in a general equilibrium model is an input-output matrix. The sectors that are included in these matrices are quite standardized and usually number 23 in the most aggregated form. If one treats these sectors as commodities and adds on various scarce resources, such as capital and labor, one obtains a model the size of which makes computation time almost prohibitively long. 6 One could, of course, try to aggregate various sectors of the input-output matrix to decrease the number of goods, or, as some authors have done, one might estimate simple smooth production functions with the same goal in mind. 7 Doing so, however, would cause the loss of a large measure of the accuracy contained in the empirical observation of the input-output matrix. We would therefore like to construct an augmented version of the input-output matrix in which the original structure is preserved and to which various import and export activities are added. We will take advantage of certain properties of input-output systems to reduce considerably the size of the space of commodities on which the general equilibrium algorithm operates, thereby saving a great deal of computer time. 8 The special structure that we create for the model requires, however, what may seem to be rather artificial definitions of certain goods, in particular of financial assets. Rather than immediately turning to a description of the general characteristics of the model, we shall informally explain the problems presented by the computational method used in the hope that doing so will help to justify the structural definitions in the next section. 9

The computational algorithms currently in use for solving general equilibrium models consider a system of prices with one price, or dimension, for every commodity in the model in question. The algorithms then search for a set of prices at which supply and demand are equal. 10 If the technology of the model is given by an input-output matrix, then the task of deriving market-clearing prices is much simpler: by inverting the Leontief matrix, the equilibrium prices corresponding to a particular set of value added are determined, while the level of production is determined purely by the demand side of the economy. 11 Although the model contains an input-output substructure, it has certain key differences from an input-output model which do not allow the simple matrix inversion technique to be applied. In particular, an input-output model has only a single scarce resource, usually taken to be labor, and consumers have no initial holdings of any produced good. We wish to have both capital and labor as scarce resources in the model, which would be incompatible with an input-output structure.

When we attempt to include money, both domestic and foreign, we encounter yet another problem. Domestic money is a good that, within the context of the model we shall develop, can be produced either by exporting domestically produced goods or by exchanging foreign exchange for domestic money, while foreign exchange may be produced by exchanging domestic money for it. 12 A realistic modeling of money in the economy necessitates the assumption that positive initial stocks of domestic money and foreign exchange exist. Such an assumption, however, is incompatible with the structure of an input-output model.

The method of dealing with the computational problems created by the model constructed here is to treat the model as a general equilibrium system with an input-output subsection. This methodology then allows the algorithm to operate only on the space of scarce resources, of which there will be 5, rather than on the entire commodity space, which contains 45 goods. Intuitively, the workings of the model are as follows. Suppose that p = (pi); i = 1, …, 5 is an arbitrary vector of prices for each of the scarce resources, 13 and assume that there are ad valorem taxes levied on the uses of scarce resources. 14 Let t(i, j) denote that tax rate that the jth sector, or production activity, must pay on its inputs of the ith scarce resource, and let VA(i, j) represent the physical inputs of scarce resources to the jth sector when it is operated at unit level. 15 The total cost to the jth sector of the inputs required for production at a unit level is then given by vaj(p) where

υaj(p)=Σi=15p1VA(i,j)(1t(i,j))(1)

Given these total costs, or value added, it is possible, via the Koopmans-Samuelson nonsubstitution theorem, to find a set of activities that are optimal in the sense that they minimize cost for any output mix. 16 We may then form a (40 × 40) matrix A(p) that consists of these optimal activities (there are 40 produced goods) and derive Leontief, or zero-profit, prices pL(p) as

pL(p)=va(p)(IA(p))1(2)

where va(p) is the vector of value added corresponding to A(p). Combining the assumed prices p of the scarce resources with the consumers’ holdings of these commodities (the only goods that they initially possess) gives us each consumer’s wealth. 17 We also have from equation (2) a complete set of prices for all goods; and, given the demand parameters for each consumer, we may compute the individual, and hence the aggregate, demand for each commodity. 18

Let xL(p) = {xL1(p)}; i = 1, …, 40 represent aggregate consumer demand for each of the produced goods. Production levels yL(p) =- {yLi(p)}; i = 1, …, 40 that will satisfy this level of demand are then given by

yL(p)=(IA(p))1xL(p)(3)

The vector of total inputs of scarce resources to production, z(p), is then given by

z(p)=VA(p)yL(p)19(4)

Thus, if r = {r}i; i = 1, …, 5 is the vector of aggregate initial holdings of scarce resources, then the total market supply, y(p), of scarce resources is

y(p)=z(p)+r(5)

and excess market demand, u(p), is

u(p)=x(p)y(p)(6)

where x(p) represents the vector of aggregate consumer demands for the scarce commodities. At an arbitrary set of prices, there will be some scarce goods for which the excess demand in equation (6) is positive, indicating that markets do not clear. By equation (3), however, supply and demand for produced goods are always identically equal, and the computational algorithm will search on the price space of scarce resources for a vector of prices for which supply and demand are equal. By employing the shortcut of utilizing the input-output component of the model, one achieves a dramatic reduction in the size of the computational problem to be solved.

In the next section, we will give a detailed description of the definitions of financial assets that shall be employed, but let us briefly note here the general structural problem brought about by the computational technique. In order for the underlying input-output matrix to be employed, no good that is produced may be part of the initial holdings of consumers, so that financial assets, which are part of the initial wealth of the public, may not be produced by any commercial activity. Formally, however, we wish to be able to treat financial assets as commodities that may be received as the outcome of certain commercial activities, such as exporting. We must therefore introduce, for each financial asset that is held as an initial stock, a corresponding commodity that may be interpreted as a flow of that asset and may thus be treated as a produced good.

II. Structure of the Model

We have briefly described the formal structure of the model and have explained the difficulties involved in implementing the computational techniques, but have not yet spoken about the specific real and financial assets to be used. This section will be divided into three parts, in the first of which production is described, while the second and third describe the tax system and consumption, respectively.

production

Scarce resources in the model will consist of five commodities: (1) capital, (2) labor, (3) initial holdings of foreign exchange, (4) initial holdings of M1, which denotes currency and demand deposits, and (5) initial holdings of D, which denotes savings and time deposits. Capital and labor are the two non- produced physical inputs to production. Scarce resource 3 is defined as the total private holdings, both in Argentina and the rest of the world, of the major currencies valued in terms of dollars. There are obvious empirical difficulties in estimating such a sum, so we have used the simplifying approach of taking it to be equal to the dollar value of the sum of savings and time deposits in each of the following countries: the Federal Republic of Germany, the United States, the United Kingdom, Japan, France, and Italy, plus the dollar holdings of the Argentine private sector in banks outside Argentina. 20 The holdings must, of course, all be measured at the same time period, and for this exercise we have chosen to use the fourth quarter of 1978 as the base period for the entire model. Scarce resource 4, currency and demand deposits, is assumed to be held only by Argentine citizens and is taken to be the officially given value of M1, while scarce resource 5, savings and time deposits, is divided among Argentines and private citizens from the rest of the world.

If D represents total initial holdings of domestic savings and time deposits, and DF represents short-term foreign liabilities of the banking system, 21 then DA, initial holdings by Argentines of savings and time deposits, is given by

DADDF(7)

Initial holdings of M1 and D are stocks that are given by the corresponding stock of the previous period plus the domestically created component of that particular monetary asset in the current period. Thus, if DC denotes domestic credit and k denotes the money multiplier, then by definition

kΔDCΔMid+ΔDd(8)

where ΔM1d represents the domestically created change in M1 and ΔDd represents the domestically created change in D. The initial stock of M1 (i.e., M1tI) is then given by

M1tIM1(t1)+M1d(t)(9)

where M1(t-1) represents the total stock of M1 in period t - 1 and M1d(t) represents the domestically created component of M1 in period t. The initial stock of D (i.e., DtI) is similarly given by

D1tID(t1)+Dd(t1)(10)

where D(t-1) represents the total stock of D in period t - 1 and Dd(t - 1) represents the domestically created component of D in period t, so that

M2(t)IM1(t)I+DtI=M1(t1)+D1(t1)+M1d(t)+Dd(t)=M2(t1)+kΔDC(t)(11)

Thus, the current initial stock of money is equal to last period’s total stock plus the change in domestic credit in the current period. It is quite possible for the money market to be in disequilibrium if, at the set of prices prevailing at the beginning of the current period, the stock of money is not equal to the equilibrium stock. 22

Intermediate and final goods consist of the 23 commodities in the Argentine input-output table 23 plus good 24, foreign exchange created “at the end of the period”; good 25, M1 created “at the end of the period”; and good 26, D created “at the end of the period.” 24 Each of the commodities (goods 24-26) may be interpreted as representing a change, or flow, in the corresponding asset. Thus, if the only source of monetary expansion endogenous to the model were export receipts, then M1 at the end of the period would be that amount of domestic money created by exports. This paper will assume that the only sources of creation of good 26, D at the end of the period, are capital inflows, although it is possible to have a spot market in which consumers may exchange initial allocations of M, and D without increasing the total supply of either one. Foreign exchange held either by Argentines or by the rest of the world may be created by changing domestic currency held at the beginning of the period into dollars. 25

Domestic production consists of the 23 activities given in the Argentine input-output matrix, and each domestic activity requires inputs of capital and labor, but not of the three other scarce resources. We have used the 1970 input-output matrix as a base, so that all physical commodities are measured in terms of constant 1970 Argentine pesos. 26 Labor inputs are measured by the wage bills for the respective industries, while capital inputs are given by the profits of the respective industries. Our justification for this derivation of capital is that in an economy characterized by constant returns to scale and efficient markets, profits are given by the returns to capital. 27

The import sector contains 14 activities, which are those importing the foreign counterparts of goods 3-16 of the Argentine input-output matrix—namely, the commodities produced by the industrial sector of the economy. As in Armington (1969), we have chosen to distinguish imported goods both by their physical characteristics as well as by their place of origin, so that we consider imports of goods 3-16 to be different from the corresponding domestic commodities; hence, these imported goods are designated goods 27-40 in the model. In addition, there will be no use of imports as inputs to domestic production, since at the degree of aggregation that we are working, import categories are not really comparable with the corresponding domestic categories. 28 In order to import one unit of one of these goods, inputs are required of capital, labor, and scarce resource 4 (M1 at the beginning of the period), so that, in particular, it is not necessary to export in order to import if there is a sufficient initial stock of M1. Denoting the dollar price per ton of the good in question at time t as Pt, the exchange rate at time t by et (in terms of units of domestic currency per dollar), and the 1970 values of the corresponding variables as P0 and e0, 29 it is easy to show that the cost aM1 in terms of domestic currency (at the official exchange rate), of importing the amount of that good costing 1 peso in 1970 is given by

(et/e0)(pt/p0)=aM1(12)

so that an import activity has the form

(ak,aL,aM1,00,1,0,,0)30(13)

Here aK, aL, and aM1, represent the amounts of the respective commodities needed to import one unit of the good in question; and as world prices and the official exchange rate change, am1 will also change. Also, since imports require domestic money as an input, they are directly reflected in decreases in the domestic money supply.

There are 17 export activities, the first 14 of which correspond to (the industrial) goods 3-16. 31 These require only the scarce resources of labor and capital as inputs and produce M1 at the end of the period, so that an export activity has the form

(ak,aL,0,,1,0,,aM1,0,,0)(14)

In summary, M1 created at the end of the period is produced by export activities, and D created at the end of the period is produced by capital inflows. Initial stocks of M1 are used to import, while initial stocks of D may be exchanged for dollars, 32 so that if M2I is the initial stock of money, as in equation (11), then the balance of payments is given by the change in M2I plus goods 25 and 26 (M1 and D, respectively, created at the end of the period). Thus

B=ΔM2I+M25+D26=ΔM1I+ΔDI+M25+D26(15)

where M25 and D26 represent goods 25 and 26 created at the end of the period. Here ΔM1IandΔDI will be nonpositive, representing domestic money from the beginning of the period that has been used to pay for imports or that has been exchanged for dollars, while M25 and D26 will be nonnegative, representing export receipts and capital inflows, valued in terms of domestic currency. Thus, equation (15) simply represents changes in reserves, valued in terms of domestic currency.

taxes and tariffs 33

In the real Argentine economy, there are innumerable taxes and tariffs; but in order to make the model manageable, we will include only the two most important types. There is a value-added tax of 16 per cent that, for domestic production, is levied upon the value of the inputs of capital and labor for each activity; for imports, it is levied upon the value of the sales price of the imported good. There are also tariffs that vary according to commodity and are charged on the basis of the sales price of the imported good, so that by compounding the value-added tax with the commodity tariff rate, we obtain a single tax rate levied upon a particular imported good. 34 This tax will then be applied to scarce resource 4 (M1 at the beginning of the period), in which the official domestic price of the good is denominated. The study assumes that there are no export taxes, since this is currently the case in Argentina. If t(i, j) is the tax rate on inputs of the ith scarce resource levied upon the jth sector, then revenue collected from the jth sector, Rj(p), is given by

Rj(p)=Σi=15pit(i,j)y(i,j)(16)

where y(i,j) denotes the level of use of the ith scarce resource by the jth sector. 35 Let us recall that these taxes are collected by the government and are distributed among consumers, who will be the next object of our attention. 36

consumers

The empirical model has two consumers (who are, of course, really categories of consumers), the first being the Argentine private sector and the second being the private sector of the rest of the world. Our first task is to derive demand functions to represent each of these consumers. In order to properly estimate such functions it is necessary to have time series of personal consumption of each of the commodities that enter the model. Such a series is not, however, available, and personal consumption of the commodities of the model was estimated most recently in 1963, corresponding to the input-output matrix of that year. We have therefore chosen to assume that Argentina has a Cobb-Douglas demand function for goods and services, thus spending a constant fraction of national income on each commodity, and we have taken these fractions to be equal to the expenditure pattern of 1963. The expenditure weights thus derived are for the first 23 intermediate and final goods produced by the input-output matrix. Since we are treating imports as different commodities from their domestically produced counterparts, we must also derive expenditure weights for goods 27-40, imports. We have done so by simply taking the fractions of national income that imports of each commodity represented during the fourth quarter of 1978, and scaling the consumption of domestic goods accordingly. 37 Since consumption of imports is strongly affected by government policies that are not included in the model, it is quite difficult to estimate a time series of expenditure patterns for import consumption.

We need also to derive expenditure weights for the scarce resources, including financial assets at the beginning of the period, and for goods 24-26, financial assets at the end of the period. We derive these weights for the Argentine consumer in the following way. We have assumed that there is no demand for leisure, so that at any set of prices there will be full employment. Although there is some unemployment in Argentina, it is, at least officially, very slight and has been in existence for some time, therefore justifying our claim of a full employment. Since estimation of an investment function representing capital formation is beyond the scope of this paper, demand for capital is assumed to be zero; were one to estimate such an investment function, it would be partially determined by the demand for capital. In order to derive demand for financial assets as a fraction of national income, one must estimate demand-for-money functions, first for M1 and then for D.

The functional form used here is chosen to facilitate the prediction of money demand in future periods. 38 The equation, which is derived from an adaptive expectations formulation for the anticipated rate of inflation, is estimated as

(1(1B)L)logmt=λa0+(1(1B)L)a1logytBa2LΔlogPt+(1(1B)L)(1λ)logmt1(17)

where m denotes nominal money deflated by the wholesale price level; y denotes real gross domestic product (GDP) at market prices; Pt denotes the wholesale price level; and B denotes the adaptive expectations adjustment parameter. 39 Demands for both M1/p and D/p are estimated in this way, and the results are reported in Section IV.

In order to derive the expenditure weights for financial assets, let us note that national income is given by the value of expenditure on both real and financial assets, so if I represents real national income, m1 represents demand for real M1, d represents demand for real D, and fa represents demand for foreign assets in real terms, then

I=y+m1+d+fa(18)

We may thus derive the fractions of national income that are given by m1 and d, which have, in turn, been estimated from equation (17). Thus, d4 and d5, the demand parameters for m1 and d, respectively, are defined as

d4=m1I;d5=dI(19)

We assume that d3, the demand for foreign exchange as a fraction of income, is determined exogenously by, for example, real relative interest rates in Argentina and the rest of the world. If demand for foreign exchange, fa, is given by

fa=d3I(20)

then equation (19) becomes

(1d3)I=y+m1+d(21)

from which I is derived.

The income of the rest of the world, for purposes of this study, is given by its holdings of foreign exchange plus its holdings of savings and time deposits in Argentina, and we have taken the demand parameters for these two assets to be the fractions of world income that they represented in the fourth quarter of 1978. Ideally, demand-for-money functions for the rest of the world would also be estimated, but such an exercise is beyond the scope of this study.

The workings of consumer demand for financial assets are designed in accordance with the monetary approach to the balance of payments. 40 Suppose that at a given set of prices, the value of a consumer’s initial holdings of, say, M1 is greater, as a fraction of his income, than he desires to hold—that is,

p4r4I>d4(22)

where r4 denotes his initial holdings of M1, and p4 denotes the price of M1. In this case, he will rid himself of the excess holdings of M1 by exchanging them for imported goods. A similar result occurs if he holds excess balances of D, as he will exchange them for foreign currency. If, on the other hand,

p4r4I<d4(23)

then the consumer will add to his initial holdings of M1 by demanding good 25 (M1 created at the end of the period) in an amount sufficient to make up the difference. Thus,

p25x25=(d4p4r4I)I(24)

in this case. Here x25 denotes the amount that the consumer demands of M1 created at the end of the period, and p25 denotes its price. The consumer’s demand for savings and time deposits and for foreign exchange behave similarly, with excess demands for these assets being made up by their corresponding assets created at the end of the period—produced goods 26 and 24, respectively.

Let us recall that this demand for increased holdings of M1 can be satisfied only by a positive current account, since exports produce M1 and imports use it as an input. A demand for increases in holdings of D can, in turn, be satisfied only by a positive net capital account, which would be produced by consumers exchanging part of their holdings of foreign exchange for D.

In summary, an excess supply, measured as a fraction of income, of a particular financial asset will lead to the consumer’s divesting himself of that asset and exchanging it for either another financial asset or for imported goods. If the consumer demands more of the asset than he has as an initial allocation, then his demand will be satisfied by a positive flow of the corresponding asset at the end of the period. 41

III. Price and Exchange Rate Stabilization

We have constructed a general equilibrium model of an economy with trade restrictions—namely, tariffs—and would now like to be able to answer a number of questions concerning the Argentine economy by using the model. In particular, we wish to be able to estimate what the effects will be if certain tariffs or taxes are relaxed; and, if the changes have adverse effects, what policies need to be undertaken to counteract these effects. A general consequence of a tariff relaxation will be an initial deterioration in the balance of payments, as imports become cheaper and thus are demanded in greater quantities than before. 42 The government would presumably wish to be able to stabilize the balance of payments while it carries out its relaxation of restrictions, but how should it do so? One possibility would be to devalue the exchange rate, thereby making imported goods more expensive and compensating for the tariff reduction. Although such an approach has obvious intuitive appeal and is clearly a step in the right direction, it is not immediately apparent by how much the exchange rate should be devalued. In the first place, even if one can predict the immediate price impact on the commodity affected by the tariff reform, one cannot necessarily predict the overall price effects. Since a devaluation affects the prices of all traded goods directly, and since there is substitution between domestic and imported goods, it may also affect the overall price level. One thus cannot hope to estimate accurately the effects of either the liberalization or of the various stabilization programs without using a general equilibrium framework.

The basic aim will be to develop a method for neutralizing changes in the balance of payments caused by shifts in the tariff regime. Because it is not possible to solve analytically the general equilibrium model, it is not possible to derive explicitly an exchange rate that will exactly compensate for tariff changes; but by looking at various price indices, one may derive certain conclusions about possible stabilization targets. Suppose that at the initial set of tariffs, the equilibrium prices for all intermediate and final goods are given by the vector pL*. Expenditure weights, wdi*,wdLi* for all domestic goods are defined by 43

wdi*=pi*yi*GNP;i=1,,5wdLi*=pLi*yLi*GNP;i=1,,26(25)

where gross national product (GNP), the total value of supply (including financial assets), is given by

GNP=Σi=15pi*yi*+Σi=140pLi*yLi*

Here yi* represents the equilibrium total supply of the ith primary commodity, and yLi* the corresponding equilibrium output of the ith intermediate and final good, so that GNP is the value of supply at equilibrium. Thus, wdi* is the proportion of total income spent on the ith primary good, while wdLi* is the proportion of income spent on the ith domestically produced intermediate or final good. Prices may be normalized by letting scarce resource 4 (initial stocks of M1) be the numeraire, so that GNP reflects the overall price level. A price index of domestic goods may then be formed, in terms of domestic currency, as

Pd=Σi=15wdi*pi*+Σi=126wdLi*pLi*(26)

One may similarly form a price index of imported goods by

Pm=Σi=2740wmLi*pLi*(27)

where

wmLi*=pLi*yLi*GNP;i=27,,4044

The relative price indices of imported and domestic goods are given by

R=PmPd(28)

One might wish to think of this relative price index as representing a measure of the exchange rate, in terms of units of domestic currency per dollar. 45

Removing or relaxing a tariff will, in general, lower Pm and hence also cause R to decline, as could also have been brought about by a revaluation of the exchange rate. Since this tariff relaxation will normally lead to a deterioration in the balance of payments (in the short run), one way the government might try to counteract this deterioration would be to bring R back to its original level, thus maintaining the price ratio of imported to domestic goods. There are two policy instruments for doing so within the context of the model. The first is to devalue the nominal exchange rate by the inverse of the change in R. Thus, if R is the relative price index before the tariff change and R’ is the index after the change, then the degree of devaluation required by this method would be given by R/R’ - 1. It cannot be guaranteed that this devaluation will exactly correct the imbalance in the balance of payments, since any change in a parameter of the model has repercussions throughout the entire model, but it should, theoretically, be a step in the right direction.

One interpretation of the connection between the relative price index and the balance of payments is that the latter implicitly reflects the relative costs of domestic and imported goods. Thus, a naive explanation of the deterioration in the balance of payments is that the relaxation of tariffs has caused imported goods to become cheaper relative to domestic goods than they were before. One way to compensate for this effect is to increase the price of imported goods by the nominal exchange rate devaluation mentioned earlier, but another way would be to decrease the price of domestic goods by an amount equal to the decrease in the price of imported goods. Let us recall that the supply of money at the beginning of the period, the total of scarce resources 4 and 5, consists of the money supply from the previous period plus whatever money it is assumed will be created in the current period by domestic credit expansion. Since a major portion of domestic credit expansion is determined by variables controlled by the government, such as the size of the fiscal deficit, it would be possible to start the period with different money supplies, given different sizes of these exogenously determined government control parameters. Suppose now that one wishes to lower the price of domestic goods. From the demand-for-money equations, we have estimated the real quantity of money that will be demanded by the Argentine public in this period. Combining this amount with the estimated demand for domestic currency by the rest of the world, we arrive at the total demand for domestic money in this period. If, then, the Government decreases the rate of increase in the domestically created supply of money by, for example, decreasing the size of the deficit, then the domestic price level must also increase less rapidly than would otherwise be the case if the real stock of money is to equal the desired stock. 46 The parameters of the demand-for-money function will then allow us to compute the amount that the supply of money must fall from its originally projected level if domestic goods are to keep their same price level relative to imported goods.

It may be claimed that in taking this monetary approach, it is incorrect to look at only the relative price index of traded goods, since this will directly affect only the current account but will not affect the capital account, which may tend to compensate for current account shifts while also being affected by the relative prices of traded goods. Rather, in order to take into account the overall balance of payments, one should look at the aggregate price index. If, for example, the relaxation in tariffs causes a decline in the overall price level given by

P=Pd+Pm(29)

as it normally would, then this decline will lead to an increase in the real quantity of money. In the short run, however, people will not wish to increase their real cash balances and will therefore change their excess holdings of domestic money into other assets. If all of these holdings are spent on domestic commodities, then the impact of this divestiture of money will be simply to increase the domestic price level. Usually, at least part of the excess cash balances will be spent on imported goods or will be changed into foreign currency. In these cases, there will be net outflows of domestic currency, indicating deterioration in the balance of payments. Thus, by stabilizing the overall domestic price level, and hence the real quantity of money, changes in the balance of payments may also be at least partially neutralized. If one takes the point of view that it is the overall price level that should be controlled, it will imply a different initial stock of money than if the target is the relative price index of traded goods. One might also try to stabilize the domestic price index by devaluing the exchange rate in much the same fashion as described earlier. The target for the devaluation would be different, however, from the case where the government desires to stabilize the relative price indices of imported and domestic goods, since a tariff relaxation has price effects upon domestically produced goods, as well as imported ones, as a result of consumers shifting the relative structure of their demands as prices change.

One cannot say in advance which of these targets will be more effective in controlling the balance of payments; and, indeed, in different circumstances, one may work better than another. It may also be claimed that the goal of insulating the domestic price level against shocks caused by the external sector should, indeed, be the primary target of economic policy within the context of the model. In the next section, by means of several experiments with the model using Argentine data, the question of whether it is effective to use a combination of different targets and instruments in response to a tariff relaxation will be examined. It will also be possible to draw a number of conclusions concerning the effect of these targets on other variables that are of interest to the government, such as tax revenues and consumer welfare.

IV. Simulation Analysis

In this section, we shall analyze the results obtained from solving a model of the structure described incorporating Argentine data. We have undertaken four simulations, the first of which looks at a particular historical situation and attempts to judge how closely the model approximates the true Argentine situation. The second simulation predicts the impact of a relaxation in tariff rates, while the third and fourth simulations consider the effects of monetary policy and a combination of monetary and exchange rate policy, respectively, as policy instruments used by the Government in attempting to counteract the deterioration in the balance of payments caused by the tariff liberalization.

historical simulation: fourth quarter, 1978

In order to use this model for policy simulations, it is necessary to first have at least some indication that it offers a reasonably accurate representation of the current Argentine economy. We have, accordingly, constructed a set of data representing the exogenous parameters in the model, which are derived from data for the fourth quarter of 1978. 47 We have then solved the model to derive the endogenous variables—output, consumption, balance of payments, etc.—and have compared the results with those that actually occurred in the fourth quarter of 1978. 48

The estimated parameters for the money demand equation (17) are

logmt=3.338(1.141)+0.730(1.208)logyt0.462(2.074)ΔlogPt1+0.595(3.843)logmt1(30)B=0.5,R2=0.734,DW=1.624
logdt=3.138(0.787)+0.671(0.827)logyt0.841Δlogpt1(3.960)+0.631(6.301)logdt1(31)B=0.5,R2=0.776,DW=1.694

where the numbers in parentheses are t-statistics. Here B = 0.5 was the value that maximized the log-likelihood function for both equations. 49 If one interprets yt as the level of real income that consumers anticipate for period t, it may then be assumed that one is in period t - 1 and may simulate the desired real levels of M1 and D for period t, thereby deriving the expenditure weights for these two assets. For the purposes of this simulation, we have taken the necessary parameters from the third quarter of 1978 and have assumed anticipated real income for the fourth quarter of 1978 to be equal to actual income for that period.

We shall now describe the results of the model when it is solved to predict the outcome of the fourth quarter of 1978. In Table 1, we give the predicted gross outputs (calculated as real value added) of each sector and compare them with the actual values for that period. Since the model uses constant 1970 prices, we have deflated the predicted sectoral outputs by the corresponding national account deflators in order to obtain units based on 1960 prices that correspond to the national accounts.

Table 1.

Argentina: Gross Domestic Production, Fourth Quarter, 1978

(In millions of 1960 pesos)

article image

The national accounts at this level are slightly more aggregated than the input-output matrix, and these categories are subsumed in other categories. In addition, there is not a precise correspondence between certain other sectors.

As we notice in Table 1, the predicted and actual values of total GDP are remarkably close to each other. It is difficult to make direct comparisons between some individual sectors because of differences in aggregation; but in those sectors where comparisons are possible, there is a reasonable similarity between predicted and actual values. In Table 2, we make a similar comparison between the actual and predicted balance of payments. Since we measure the balance of payments as the change in the money supply, we show the initial and final stocks of money.

Table 2.

Argentina: Balance of Payments, Fourth Quarter, 1978

article image

Good 25 is produced by export receipts.

Good 26 is produced by net capital flows. A capital outflow will be shown as a decrease in holdings of M2.

We are using the fourth quarter of 1978’s exchange rate of 1,003.5 pesos per dollar.

We are excluding official capital movements.

We notice that there is a fairly close correspondence between the predicted and the actual balance of payments. In particular, we have been able both to predict the directions of change and to approximate the magnitudes of the current and capital accounts.

There are a number of other predicted results of the model that may be useful to mention at this point. In Section III (equation (29)) we described how an overall domestic price index could be formed. The price index P corresponding to this example, which is derived in the same way as equation (29), is 50

P = 61.09

giving a real quantity of money, M2P, at equilibrium of

M2P=311.77million1960pesos

Total government revenues T were 51

T = 6,713 billion current pesos

while the value of nonmonetary expenditure C (i.e., the value of consumer expenditure on nonfinancial assets) was

C = 58,409 billion current pesos

so that, in particular, taxes and tariff revenue are 11.49 per cent of total expenditure. 52

Since the Government may be interested in the welfare implications of its tax and tariff policy, it may be useful to note the utility levels of consumers in this simulation and the following ones. 53 These levels have, of course, no absolute significance, but ordinal changes in them may be of interest.

Utility of Argentina = 4.20

Utility of the rest of the world = 17.85

The relative prices, consumer demands, and excess demands for the five scarce resources are given in Table 3. Thus, excess demands for all scarce resources are approaching 0, with the exception of capital, which is in excess supply. There is zero demand for both capital and labor; labor, which is a binding constraint on production, has a positive price, and capital, which is in excess supply, has a price that is approaching 0. It is also interesting to note that the relative price of good 24 (dollars created during the period) is 598.91 with respect to scarce resource 4 (initial holdings of M1). The official exchange rate for the period was 1,003.5, however, which explains why consumers would wish to increase their holdings of M1 (since the exchange rate was being held artificially high).

Table 3.

Argentina: Equilibrium Relative Prices and Demands, Fourth Quarter, 1978

article image

M1 is the numeraire.

tariff reduction

Suppose now that the Government decides to carry out a trade liberalization of the type that is currently being implemented in Argentina. We have thus allowed individual tariff rates to be reduced by a uniform 50 per cent and have resolved the model. As Table 4 indicates, there is a considerable deterioration in the balance of payments, as compared with the original simulation.

Table 4.

Argentina: Balance of Payments Subsequent to 50 Per Cent Tariff Reduction

article image

We thus note that the balance of payments deteriorates by a further US$421 million because of the reduction in tariffs. In addition, the overall price index P declines by 12.98 per cent to

P = 53.17

The real quantity of money demanded at equilibrium, M2P, has risen to

M2P=350.25million1960pesos

because the tariff reform has raised real income. In addition, government revenues have fallen to 5,911 billion pesos, or 10.73 per cent of the total expenditure of 55,091 billion pesos.

tariff reduction combined with reduction in initial money stock

Suppose now that the Government decides to stabilize the overall domestic price index, so that there will not be an increase in the real quantity of the initial money stock, leading to an outflow of reserves. Accordingly, operating on the assumption that for the quarter in question, the desired real stock of money is fixed as a fraction of income, the Government reduces the initial nominal stock of money by 12.98 per cent, the amount of the decline in the price level caused by the tariff relaxation. 54 The results obtained are given in Table 5.

Table 5.

Argentina: Balance of Payments with 50 Per Cent Tariff Reduction Combinedwith 12.98 Per Cent Decrease in Initial Stock of Money

article image

Thus, the reduction in the initial money stock has caused an improvement of US$100 million in the balance of payments, compared with the previous example. The price level has declined further to

P = 52.31

so that the removal of tariffs, combined with a tightened money policy, has a deflationary impact of 14.37 per cent. Finally, the real stock of money at equilibrium is

M2P=309.75million1960pesos

or approximately what it was in the original simulation.

We note that decreasing the money supply by the same amount as the decrease in the price level does not fully neutralize the effects on the balance of payments caused by the tariff liberalization. The intuitive reason for this is that the wealth effect caused by the reduction in the money supply tends to decrease the deficit in the balance of payments, while the relative price effect raises the relative price of money, hence lowering the domestic price index. This lower domestic price index tends to counteract the initial impact of the wealth effect, hence partially negating the stabilizing impact on the balance of payments.

It is, perhaps, unrealistic to hope that a change as drastic as a 50 per cent tariff relaxation could be completely neutralized in one quarter. Suppose that the Government, after having tightened the money supply during the period in question, decides to allow this experiment to continue for a second quarter. In the second period, there will be no expansion of domestic credit, so that the initial allocations of financial assets in this second period are equal to the equilibrium holdings of them at the end of the first period. We also assume that there is no change in the initial allocations of capital and labor. This experiment is of particular interest because it allows us to get some idea of how quickly the model stabilizes after having been shocked.

As Table 6 shows, in the second period, after the initial shock, there is an improvement of more than US$400 million in the balance of payments, compared with the previous period when the shock took place. There is, indeed, an improvement of about US$175 million over simulation 1, which was reported in Table 2, that represents the situation before the shock. The model is hence rapidly stabilizing the balance of payments and after one or two more periods of restricted credit expansion should be virtually in equilibrium. If it is felt that such rapid stabilization is not necessary, then a controlled expansion of domestic credit may be allowed, rather than the somewhat unrealistic assumption of zero expansion used in this example.

Table 6.

Argentina: Balance of Payments in Second Period After Initial Tariff and Money Supply Reductions

article image

use of devaluation as stabilizing instrument

In the previous simulations, the only government policy instrument that has been considered has been the control of domestic credit expansion and, hence, of the domestic component of the money supply. Normally, the exchange rate would also be used to compensate for shocks to the external sector, so the final example will examine the effects of combining a devaluation with restrictive monetary policy. As was mentioned in Section III, the relative price index of imported to domestic goods, as defined in equation (28), may be thought of as a measure of the real exchange rate. When the tariff liberalization was carried out in simulation 2, this index declined by approximately 15 per cent. Suppose, then, that the Government carries out the same restrictive monetary policy as in the first part of simulation 3, and at the same time devalues the nominal exchange rate by 15 per cent, attempting to stabilize the shift in the relative price indices of equation (28). The balance of payments now shows the results that are given in Table 7.

Table 7.

Argentina: Balance of Payments Combining Tariff Liberalization, Money Supply Reduction, and 15 Per Cent Devaluation

article image

Recall that we are using an exchange rate that has been devalued by 15 per cent, so that there are 1,154 pesos per U.S. dollar.

Adding the devaluation of 15 per cent onto the already restrictive monetary policy has thus led to a further improvement of more than US$80 million in the balance of payments, and an improvement of more than US$180 million compared with simulation 2, where the tariff relaxation was first considered. Clearly, further tightening of monetary expansion and more rapid devaluations would lead to faster stabilization of the balance of payments. In this example, the overall price level P rose to

P = 59.27

which is approaching the level reported in simulation 1, before the tariff relaxation took place. In addition, the real stock of money, M2P, is now

M2P=273.24million1960pesos

This decline in the real stock of money has taken place primarily because the combined effect of tight money and devaluation has been to decrease real income, thus lowering the absolute level of the demand for money.

In the initial case of simulation 1, total government revenues T were 6,713 billion pesos, or 11.49 per cent of expenditure, while now they are

T = 5,145 billion current pesos

which represents 10.73 per cent of total expenditure C, which is

C = 47,936 billion current pesos

The utility levels of Argentina and the rest of the world are

Utility of Argentina = 3.41

Utility of the rest of the world = 17.83

Hence, Argentina has suffered a moderate welfare loss, compared with the original simulation, where its utility level was 4.20; this indicates that, from the point of view of overall consumption, the negative wealth effects of the stabilization program have dominated the favorable relative price effects of the tariff liberalization.

Finally, it is interesting to compare relative prices in this simulation and those in the initial simulation, which were reported in Table 3. In particular, we notice in Table 8 that the relative price of labor has declined significantly from the original price of 10.2161. If each of these prices is deflated by the corresponding overall price index, real wages in the two cases are

Real wage (simulation 1) = 0.1672

Real wage (simulation 4) = 0.1415

so that the real wage declines by 15.41 per cent when the stabilization program of simulation 4 is combined with the tariff relaxation.

Table 8.

Argentina: Equilibrium Relative Prices

article image

We thus have quite specific measures of the costs, both to the Government in terms of lost revenue and to the public in terms of decline in welfare levels, of the stabilization program. The question of whether or not these costs are acceptable is, of course, purely subjective and up to the discretion of the Government, which may, indeed, decide to accelerate the stabilization process at the cost of greater losses in consumer welfare and real wages.

V. Summary and Conclusion

We have constructed a theoretical general equilibrium model of a small, open economy that has been disaggregated and that has ad valorem taxes on domestic production and tariffs on imports. In addition to physical commodities, the model also contains financial assets, in the form of domestic currency and foreign exchange, and permits the existence of current and capital accounts, a nominal exchange rate, and a price level. We have demonstrated the existence of an equilibrium within the model and have simultaneously constructed a computational method for solving the system.

The model has been implemented with Argentine data and has been found to approximate reasonably closely the actual Argentine outcomes of the fourth quarter of 1978. We have resolved the model after assuming a 50 per cent tariff reduction, and have estimated the effectiveness of monetary and exchange rate policies in stabilizing the balance of payments against the shock caused by the tariff relaxation. It was found that the changes in both domestic credit expansion and the nominal exchange rate that are required if the balance of payments is to be completely neutralized were greater than might have been expected. In addition, we have estimated the total government revenues that result from the various policies that have been considered and, via the calculation of utility indices, have examined the welfare implications of these policies.

The model is sufficiently general to be used for a number of other purposes. If data on expenditure patterns of disaggregated consumer groups were to become available, it would be interesting to examine the differential welfare effects of government tax, tariff, and monetary policies. We have also considered only the objectives of neutralizing the balance of payments against external shocks, although many other policy targets might be of interest to the Government. It might, for example, wish to stabilize total tax revenues. Yet another possible use of the model would be to examine various programs for income redistribution while simultaneously trying to prevent these programs from having too severe an impact on the price level. Finally, a general modification of the model that would be of considerable interest would be made by constructing a dynamic framework with endogenous savings and investment behavior, so that the long-term implications of government policies could be better analyzed.

REFERENCES

  • Argentina, Ministry of Economy, Instituto Nacional de Estadistica y Censos, Intercambio Comercial, various issues.

  • Armington, Paul S., “A Theory of Demand for Products Distinguished by Place of Production,” Staff Papers, Vol. 16 (March 1969), pp. 159-76.

  • Chu, Ke-Young, and Andrew Feltenstein, “Relative Price Distortions and Inflation: The Case of Argentina, 1963-76,” Staff Papers, Vol. 25 (September 1978), pp. 452-93.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Debreu, Gerard, Theory of Value: An Axiomatic Analysis of Economic Equilibrium (New York, 1959).

  • Dervis, Kemal, and Sherman Robinson, The Foreign Exchange Gap, Growth and Industrial Strategy in Turkey: 1973-1983 (unpublished, World Bank, 1978).

    • Search Google Scholar
    • Export Citation
  • Feltenstein, Andrew (1979a), “Trade Restrictions and the Exchange Rate: A Computational General Equilibrium Approach” (unpublished, International Monetary Fund, July 10, 1979).

    • Search Google Scholar
    • Export Citation
  • Feltenstein, Andrew (1979b), “Market Equilibrium in a Model of a Planned Economy of the Soviet Type: A Proof of Existence and Results of Numerical Simulations,” Review of Economic Studies, Vol. 46 (October 1979), pp. 631-52.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fonseca, Marcos G., “The General Equilibrium Effects of International Trade Policies” (unpublished doctoral dissertation, Yale University, 1978).

    • Search Google Scholar
    • Export Citation
  • Gale, David, The Theory of Linear Economic Models (New York, 1960).

  • Grandmont, J. M., “Temporary General Equilibrium Theory,” Econometrica, Vol. 45 (April 1977), pp. 535-72.

  • Hahn, Frank H. (1971), “Equilibrium with Transaction Costs,” Econometrica, Vol. 39 (May 1971), pp. 417-39.

  • Hahn, Frank H. (1973), “On Transaction Costs, Inessential Sequence Economies and Money,” Review of Economic Studies, Vol. 40 (October 1973), pp. 449-61.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • International Monetary Fund, The Monetary Approach to the Balance of Payments:. A Collection of Research Papers by Members of the Staff of the International Monetary Fund (Washington, 1977).

    • Search Google Scholar
    • Export Citation
  • Khan, Mohsin S., “Monetary Shocks and the Dynamics of Inflation,” Staff Papers, Vol. 27 (June 1980), pp. 250-84.

  • Kurz, Mordecai, “Equilibrium in a Finite Sequence of Markets with Transaction Cost,” Econometrica, Vol. 42 (January 1974), pp. 1-20.

  • Merrill, O. H., “Applications and Extensions of an Algorithm that Computes Fixed Points of Certain Non-Empty Convex Upper Semi-Continuous Point to Set Mappings” (unpublished doctoral dissertation, University of Michigan, 1971).

    • Search Google Scholar
    • Export Citation
  • Miller, Marcus H., and John E. Spencer, “The Static Economic Effects of the U. K. Joining the EEC: A General Equilibrium Approach,” Review of Economic Studies, Vol. 44 (February 1977), pp. 71-93.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Nogues, Julio J., Caracteristícas Factoriales Asociadas a las Exportaciones Manufactureras: Análisis del Caso Argentine, Series de Estudios Técnicos, No. 39 (Banco Central de la República Argentina, October 1979).

    • Search Google Scholar
    • Export Citation
  • Polak, J. J., “Monetary Analysis of Income Formation and Payments Problems,” Staff Papers, Vol. 6 (November 1957), pp. 1-50.

  • Richter, Donald K., “Existence and Computation of a Tiebout General Equilibrium,” Econometrica, Vol. 46 (July 1978), pp. 779-805.

  • Scarf, Herbert E., “The Approximation of Fixed Points of a Continuous Mapping,” SIAM Journal of Applied Mathematics, Vol. 15 (September 1967), pp. 1328-43.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Scarf, Herbert E., and Terje Hansen, The Computation of Economic Equilibria (Yale University Press, 1973).

  • Shoven, John B., “The Incidence and Efficiency Effects of Taxes on Income from Capital,” Journal of Political Economy, Vol. 84 (December 1976), pp. 1261-83.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Shoven, John B., and John Whalley, “General Equilibrium with Taxes: A Computational Procedure and an Existence Proof,” Review of Economic Studies, Vol. 40 (October 1973), pp. 475-89.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Whalley, John, “The United Kingdom Tax System 1968-1970: Some Fixed Point Indications of its Economic Impact,” Econometrica, Vol. 45 (November 1977), pp. 1837-58.

    • Crossref
    • Search Google Scholar
    • Export Citation
*

Mr. Feltenstein, economist in the Special Studies Division of the Research Department, received degrees from Harvard and Yale Universities. Before joining the Fund, he taught at the University of Massachusetts at Amherst.

The author would like to thank Tomás Baliño of the Central Bank of Argentina, who was responsible for many of the ideas that allow the introduction of money into this model. In addition, Beatrice Tchinossian, Julio Nogues, and Agustín Uriarte of the Bank, and Mario Teigeiro of the Ministry of Finance, have given him information that was crucial to the empirical implementation of the model. The author would also like to thank Roque Fernandez and Carlos Rodriguez of the Centro de Estudios MacroeconOmicos (Argentina), and Anna Martirena-Mantel and Rolf Mantel of the Instituto Torcuato di Tella (Argentina), for useful criticisms given during seminars at their respective institutions.

1

Transactions costs, the requirement that transactions use some inputs of money, were first dealt with in the context of an exchange economy in Hahn (1971 and 1973) and Kurz (1974). There has been no satisfactory model of an economy that has dealt with transactions costs in production (seeGrandmont (1977)).

2

The earlier paper constructed an exchange rate in terms of relative price indices. Since there was no domestic money there could be no nominal exchange rate.

3

We are thus implicitly assuming that the real quantity of money that consumers wish to hold is given at the beginning of the period, and that changes in the money supply go partly into changes in the domestic price level and partly into changes in the balance of payments.

6

By adding several scarce resources, we lose the input-output structure that would allow a solution by simply inverting the Leontief matrix. The problem must therefore be solved by the type of algorithm described in Scarf (1967 and 1973). We use a modified version of this type of algorithm that was developed by Merrill (1971).

7

This is the approach of Shoven (1976) in analyzing the impact of capital taxation on the U. S. economy.

8

In the model, the use of this reduced commodity space allows a cut in computation time of about 95 per cent.

9

A formal description of the computation technique, plus a proof of the existence of equilibrium in the model, is available on request from the author, whose address is the Research Department, International Monetary Fund, Washington, D. C. 20431.

10

Actually, since an approximating technique is used, only a “best” approximation to an equilibrium is achieved.

11

These results are analyzed in Gale (1960).

12

These processes, which are formally described in the next section, correspond to a current account inflow, a capital inflow, and a capital outflow, respectively.

13

These scarce resources are (1) capital, (2) labor, (3) initial holdings of foreign exchange, (4) initial holdings of M1, and (5) initial holdings of D, which denotes savings and time deposits.

14

In the next section, we will show how the use of taxes on scarce resources allows us to incorporate both ad valorem taxes and import tariffs in the model. One could also include sales taxes levied on intermediate and final goods, but, since actual rates for these are difficult to determine, we have chosen not to do so.

15

As in Shoven and Whalley (1973), we adopt the convention that t(i j) has the same sign as the corresponding input.

16

See Gale (1960) for a discussion and proof of this theorem.

17

When we speak of an individual consumer, we are, of course, referring to groups of consumers with similar demand functions. It should also be noted that initial holdings of financial assets are included in the consumer’s wealth.

18

For the purpose of this paper, it is assumed that consumers have Cobb- Douglas demand functions, although other functional forms could, of course, be used.

19

Here VA(p) represents the columns of scarce resources corresponding to A(p).

20

It should be noted that this is not the same as foreign reserves, which are held by the central banks. These countries have been chosen because they are Argentina’s main trading partners and because they have no significant restrictions on capital movements.

21

The amount of short-term foreign liabilities of the banking system would depend on the choices made by foreigners as to what proportion of their financial assets they wished to hold as Argentine currency. The choice would presumably depend on, among other things, relative interest rates in Argentina and the rest of the world.

22

In this case the excess stock of money will, as we shall see in the next section, be released by outflows into the balance of payments. It is clearly not strictly correct to speak of the domestic creation of money as taking place separately from the external sector; but because of the static nature of the model, we are compelled to do so.

23

These goods are listed in Table 1.

24

This use of what, admittedly, is unfamiliar terminology is motivated by the activity analysis characteristics of the model. Production takes place “during the period” while output emerges at the end of the period.”

25

This study thus does not consider, for example, the possibility of receiving domestic currency, say from export receipts, and then immediately exchanging it for foreign exchange. Such a transaction would be viewed as taking place in the next period.

26

A more detailed description of how these units are defined is given in Chu and Feltenstein (1978).

27

We have chosen to retain a linear structure for inputs of capital and labor, keeping the relative proportions of 1970, although it would be possible to use smooth production functions to produce value added. It is not possible at this point, however, to estimate the parameters of such functions, since annual data on profits are not available on an industry-by-industry basis.

28

The price indices of corresponding categories of domestic and imported goods are, in general, very different, as would not be the case if they were directly comparable. It should, however, be noted that a large proportion of Argentine imports are actually intermediate goods.

29

We have derived these unit import prices from various issues of Intercambio Comercial, which is published by Argentina’s Ministry of Economy, for use in the empirical work in the final section.

30

The symbol (’) denotes the transpose of a vector.

31

Clearly, there are imports and exports of goods other than those of the industrial sector, but we have chosen to use only these, since they are the ones for which unit price information is readily available. The export commodities are viewed as being the same as the corresponding domestic goods.

32

This distinction between M1 and D is, admittedly, artificial and is made for a technical reason. In equilibrium, because of the structure of the model, each commodity will be produced by only a single activity. Thus, if there were only a single type of money, which could be produced either by exporting or by capital inflows, then exporting and capital flows could not take place at the same time. Clearly, this would not be a realistic outcome, since the two do occur simultaneously. Therefore, the model has two monetary assets, one produced by exports and one by capital flows at equilibrium.

33

The author is grateful to Mario Teigeiro of Argentina’s Ministry of Finance for help in clarifying the tax structure of the Argentine economy.

34

We have taken the estimates of Nogues (1979) of average effective tariff rates for each sector, in order to give values to the tariffs in the model. These rates are often quite different from the official tariffs because of, among other things, exemptions and difficulties in collection.

35

As in Shoven and Whalley (1973), we adopt the convention that taxes on inputs have a negative sign, so that RJ(p) is positive. Total taxes collected are then given by R(p)=ΣjRj(p).

36

In practice, this distribution would take place via government expenditures on various social benefits and public goods.

37

This scaling is necessary because the 1963 figures merge consumption of domestic goods and imports.

38

This form is adopted from Khan (1980).

39

See Khan (1980) for a derivation and discussion of this formulation. A search procedure is used to find a value for B that maximizes the log-likelihood function.

41

This is a simple version of the Polak model (see Polak (1957)), although in that model the distinctions between M, and D, and between initial stocks and flows of assets, are not made.

42

This need not always be the case, and there should, in the long run, be no price effects caused by such a relaxation.

43

Here primary, intermediate, and final goods are included.

44

Thus, wmL*i is the expenditure weight of the ith imported good. It should be noted that many of the weights will be equal to the Argentine demand parameters, since there is only one Argentine consumer group. If we had disaggregated consumer categories, however, this would not be the case.

45

For a further discussion of the construction of this type of exchange rate, see Feltenstein (1979 a).

46

Hence there would be a decline in the rate of inflation.

47

The entire data set is available on request from the author, whose address is given in footnote 9.

48

This use of a “benchmark” year is quite common in empirical general equilibrium studies and is described in greater detail in Whalley (1977).

49

The equations were estimated on a quarterly basis for the period extending from the first quarter of 1971 to the fourth quarter of 1978 using ordinary least squares, while an autocorrelation correction was used in the second equation.

50

We have used scarce resource 4 (M1 at the beginning of the period) as numeraire.

51

These revenues are the sum of collections from the value-added tax and the tariff on imports.

52

Actually, national (i.e., nonprovincial) taxes were 12.25 per cent of GDP in 1978.

53

The utility levels are derived from the utility functions that underlie the demand functions being used.

54

Recall that the initial stock of money is formed by combining the previous period’s stock with the domestically created component of the money supply from the current period. Since the domestic component of the money supply is determined by domestic credit expansion, which is at least partly controlled by the Government, the desired decrease in the initial stock of money may be achieved by cutting domestic credit.