An economic model for Argentina

This paper is prepared for staff use and is not for quotation. The views are those of the author and not necessarily those of the Bank. INTERNATIONAL BANK FOR RECONSTRUCTION AND DEVELOPMENT Bank Staff Working Paper No. 177 April, 1974 An Economic Model for Argentina This model was formulated during the 1973 Economic Mission to Argentina. It highlights the following points: a. Role of relative and absolute prices (domestic and foreign) in resources allocation, income distribution and growth. b. The interdependence between money, wages, prices, and real variables. ce The cost and demand elements of inflation. Methodologically, the model tries to improve in the treatment nf the following neglected areas. a. The ex post equalization between the savings and trade gaps by changes in-prices and resource allocation. b. The importance of testing the predictability of a model by historical simulation. c. The need for a feasibility test as against a pure consistency test in a model. The model has been used for (i) prediction, (ii) sensitivity to policy changes, and (iii) testing feasibility of the plan, with alternat'.Te specifications. Prepared by: Syamaprasad Gupta Assisted by: R.D. Bhakta Comparative Analysis and ProjectionEt Division Economic Analysis and Projections Department Development Policy Staff  AN ECONOMIC M)DEL FOR ARGENTINA TABLE OF CONTENTS Page A. Introduction 1 'B. Argentine Econorty and Purpose of this Model 4 0. Existing Other Models of the Argentine Economy 6 D. MethodoloV and Structure of the Model 7 E. Tests of the Model's Validity 25 F. Projections with Alternative Policies 28 APPENDIX I: Table s APPENDIX II: Description of the Equation System Mathematical Convergence APPENDIX III: Graphs APPENDIX IV: Bibliography  A. Introduction 1. The core purpose of model building is to get an insight into the nature of the relevant set of forces shaping the economic reality of a country. This reality of a country is determined by four set of forces: (a) behavior of individuals and institutions; (b) technological know-how; (c) policy plans of the Government; and (d) exogenous forces outside the country's control. The third set impinges on the first two, often with a conscious design to alter them, in the light of a set of explicit or implicit objectives. 2. Barring the random element in these sets of forces, it is possible to express most of them in quantitative terms, setting out explicit relation- ships between variables, and express them in the form of equiations or identi- ties or bounded inequalities. These relationships can be static or dynamic. Their specifications and method of estimations would depend largely on the specific nature of the problem and may not admit generalization. As for the random elements, standard econometric methods can be used. 3. The derived interest from this knowledge lies in the attempt to explore the future of the economy. These exercises are known as "Prediction or Policy Models.t" The tests of such models are their predictability (i.e., closeness to reality or actuals). As reality represents a consistent and feasible set, all such models must satisfy a consistency and a feasibility criteria. 4. In an attempt to classify economic models, all "Prediction Models" should be regarded as a special qase of "Policy Models," In "Prediction Models" the policy changes are assumed to take place on the basis of likeli- hood or probability rather' than on the basis of any desired (normative)  -2- prescription. To be useful, they should satisfy both feasibility and consistency conditions. However, most of them satisfy ohly the consistency conditions. The majority of the models, built in academic institutions, by planning bodies and even the TWorld Bank, suffer from this deficiency. To be more specific, some of these models could satisfy consistencies only in a strict mathematical sense that is when calculated values (endogenous variables) are made consistent by being derived from solving a system of equations. But the consistencies between the assumptions arnd parameters are generally ignored. This could happen when the structured coefficients of the model is not estimated from real life values (or simulated against real happenings). For example, it is fashionable to use ICOR's to estimate the investment requirements for a given growth of GDP and at the same time to assume alternative labor coefficients (in production) rMd import coefficients independently. Their inconsistencies do not come out on the surface as they are not estimated from real life data. I have seen models where ICOR's, import and labor coefficients are changed ad hoc and independently, over the projected period. The similar objections could be raised if input/output value ratios are isolated from the relative prices in the economy and relative prices are changed consciously in the model. Indian planning models can be given as a specific examplei! All these models used a set of ad hoc para- meters to solve a system of equations. But the consistencies between the parameters and their feasibilities have rarely been queationed. In this connection an optimum disaggregation of any model becomes very crucial. Many macro models could give a pseudo sense of consistency and fall apart 1/ Gupta (1971), Tendulkar (1974), Roy (1972).  -3- when they are properly disaggregated. To give an example, an overall tax rate of 6% of income for a country may seem to be within an easy feasibility range but it might Diean an infeasible tax rate of 98% for the higher income group, when the existing income distribution is taken into consideration. Indeed a suitable guide for disaggregation of a model is the exact speci.;"ica- tion of its policy tools. If, or example, the Government wants to influence the investment activities of a society, having a large free private economy, it is essential to treat the banking sector and government's fiscal and monetary sectors separately, rather than the easy way of treating investment or quantity of money as aggregate policy variables. Further, setting "para- meter values" in the context of real-life "benchmarks"t with appropriate disaggregation w7ould take the model closer to a feasibility test. But before feasibility conditions can be satisfied, more rigorous tests will have to be applied extending to a series of relations, including the behavior and institutional structure of the economy. 5. Finally, all models must fulfill the convergence test. This is a mathematical property without which a model becomes indeterminate. But inconsistent models can also converge. Therefore, "convergence" is a necessary but not a sufficient condition. A model can have the property of non-convergence (i.e., explosiveness) in the very long run, but it may converge over the predicted period. This is true for many linear models. But as the real test of a good model is its predictability over any relevant range, there are occasions when the so-called explosive models could qualify for predictive purposes. On the other hand, by the same line of logic, many "non-explosive" models may turn to be very poor for predictive purposes.  B. Argentine Economy and Purpose of this Model 6. The most recent economic paper on Argentina in the IBRD with periods of stagnation, and development has been unbalanced. Argentina's crucial economic problem is to diagnose correctly the reasons for this state of affairs and to develop corre*tive policies and programs. '" And somewhat further in the text: "... Argentina seems to be living through an era of precarious political balance ..., and economic policy makers enter and leave the stage without time to carry through consistent economic policies ..."1 The results of the work presented here are an attempt to address the diagnos- tic problem cited above in the form of an econometric model of the economy of Argentina. It is specifically designed for assisting the economic mission's work on the preparation of the FY73 economic report on the country both in terms of economic analysis and of medium-tenr projections. 7. Historical trends and developments in the Argentine economy have been characterized by generaUly low and uneven growth rates; an almost chronic balance-of-payments deficit, aggravated during years of relatively rapid growth and accompanied by substantial speculative capital movements; rapid rates of inflation which in turn required frequent exchange rate adjustments but sometimes considered inadequate in size and too late in time; strong pressures to increase real wages in urban and industrial sectors leading to a chronic deterioration in the terms of trade for agriculture, rising unemployment and serious difficulties in diversifying exports. The model presented here tries to explain some of those features by locating their causes, to the extent that those are registered by economic variables.  8. This attempt to bring quattitative causality to bear on the Argentine econorty is not the first one; the present exercise starts in fact from the testing of several alternative hypotheses put forward in current literature. In general, it is asserted in these other studies that the primary problem is the foreign exchange constraint with import requirements found to be rather elastic whereas the exports are constrained by slow growth of output in the export-producing sector and substantial growth of domestic demand for exportables. This situation is linked by some to the income redistribution policies of successive Argentine Governments. Alternatively, others suggest that a shortage of labor has caused the slow growth of the rural sector of the country. A third opinion emphasizes slow technological progress, lack of entrepreneurs and unfavorable fiscal treatment of the rural sector of the economy. Similarly, there are differing views regarding the cause of inflation in Argentina. There is a strong monetarist school but others consider structural problems as the main cause of inflation. Finally, there are those who ascribe Argentina's growth problem largely to distorted relative prices which in turn are the result of fiscal policies or their absence. 9. The character of the economy and the identification of major policy issues and alternatives has been decisive in the choice of the sectoral di.saggregation of the model and of some key variables. Paucity of data, although a handicap in some respects, has not significantly hampered the selection of 11 sectors and/or subsectors of the economy, which provide adequate detail to study production and consumption decisions and the effects of fiscal and monetary policies. This also allowed the introduction of such variables as prices of beef, export taxes, credit conditions, changes in  -6- import duties, world prices of major exports and imports, technological progress and changes in relative wages between the urban and rural sectors. The linkage at';empted in the pres;ent model between the real and monetary sectors, introducing prices as a variable in a growth model, is a novelty usually omitted in earlier studies of Argentina. C. Existing Other Models of the ine Econom 10. Models available for Argentina can be divided roughly into three broad groups: (1) economy-wide models, (2) models for specific sectors and (3) models for testing specific economic hypotheses. Of the economy-wide models, the most well-known and recent are those prepared by Gomez (UNCTAD, in connection with Project LINK; 1972) and Nugent (1965). Nugent's model consists of 97 linear equations of which 45 are definitional. It covers the period 1947 to 1963, both for the purpose of estimation and for testing the predictive capability of the model. The model has mainly been used for policy simulations. It is not suitable for present purposes because of inadequate sector specification and problems of updating the statistical base which derives partly from unofficial sources. The Gomez model is equally unsuitable: because it seems to be mainly concerned with predictions and has insufficient specification to serve for explaining structural change in the economy. 11. As to the second category (the sector models), two important ones are: (1) by C.F. Diaz Alejandro on "the problems of balance of payments" and (2) by R.E. Yver on the investment behavior and the supply response of the cattle industry in Argentina. Both these studies are extremely useful, but partly depend on data drawn from the results of ai hoc inquiries which  -7- cannot be updated for current years, nor projected for the future. Besides, both these studies emphasize short-term issues at the expense of long-term growth problems. Other models are by Conesa (IBRD) on import suLibstitution, and by Elson (IMF) on exchange rates and export development. Although these models are useful in pointing to lines of enquiry which could be incorporated in the present exercise, their limited focus is a handicap. In general, these models suffer from inadequate [inkage between the real and monetary variables in the economy. D. Methodology and Structure of the Model 12. The present model does not start with a stipulated national income growth rate to estimate the foreign capital requirement, nor with a given trade gap and capital inflow to calculate the consequential growth rate. In fact, it derives an equilibrium growth rate endogenous by subject to a set of constraints in terms of policy, behavior, natural endowment and/or inter- national forces. In this sense, it is rather close to the "skill constraint" phase of the familiar gap approach, although being more liberal in identifying the constraining sectors of the growth process, by exploring the constraints in a more disaggregated form. Foreign capital inflows have been treated both endogenously and exogenously, in alternative versions of the model. 13. The model defines the national income identity and thus equality between the savings and trade gaps at current prices. This differs from the more comnnon "two-gap" presentation, which gives essentially a disequili- brium picture by defining equality in constant prices. The model does, of course, allow presentation of the two gaps at constant base period price, with GDP deflator as residual. As has been mentioned, the ex ante "Saving Gapt'  -8- and "Trade Gap" have been equalized by reallocating resources via price changes. But care has been taken not to violate the original behavior relations in this process. For example, from an initial position: Ypy . CPc + IPi + EPe + (-kDm) + Si where Y, C, Is E, and M are respective quantities and py, Pc, Pi, Pei pm are the corresponding prices and Si the slack variable showing the extent of supply demand imbalance, we reached a new set of prices and a new set of variables when S1 fell to zero.2/ The condition of convergence have been explained algebraically with the help of a simplified case in Appendix II. 1h. Econometric relations have been computed on the basis of past time series data for investment, consumption and international trade. This procedure has the disadvantage that parameters thus estimated are influenced by past policies and conditions which may not be relevant or of the smune importance for the projection period. Even explicitly incorporating appropriate variables in the equation system may not always provide an acceptable solution, as the introduction of values of policy variables outside the range of past observations may reduce the reliability of the projections. For example, when the consumption propensity is estimated without due regard to the fiscal restrictions which applied to consumption in the past, it will understate the consumption behavior or overstate the savings propensity. Similarly, import propensities may be influenced by existing import restrictions. The model contains the maximum number of safeguards that could be incorporated, 1/ In this model, price changes depend either on the changes in the wage rates or exchange rates on quantity of money or government's borrowing or expectations for future price changes or a combination of them. Besides, it is also made to be affected by supply-demand imbalance, as is shown above. As the price adjustment mechanism has been used only in the aggregate supply-demand balance, aggregative "price-quantity response equations'" are formulated in the National Income account block (No. IX).  -9- partly through disaggregation and for another part through inclusion of appropriate variables or their dummy forms. Disaggregation has therefore not been restricted to a representation of producing sectors only, but extended to savings, consumption and international trade. 15. The model-frame consists of 11 submodels, thus forming a simul- taneous equation system with a large block-recessive structure. The fol- lowing blocks are distinguished: a. Agriculture: divided into wheat, corn, sorghum, other agriculture, cattle and other animal products; b. Non-agriculture: consisting of capital goods industry,, non-capital goods industry especially based on agrri- culture, services; c. Consumption: beef, wheat, fruit, manufactured products, private consumption, agricultural consumption and the rest; d. Price Formation: beef, wheat, corn and sorghum, industrial goods, capital goods and consumption goods prices; e. Exports: wheat, sorghum, corn, beef, other animal products, manufactured products; f. pnport: capital goods, consumption goods, intermediate goods; g. Wage Determination: rural wages, urban wages, cost of living index; h. Treatment of Exogenous Variables; i. Monetary Sector; j. National Income Accounting.Balance.  - 10 - 16. The model contains three, sometimes mutually overlapping, alter- native sets of equations;!/ the first group explaining the structural rela- tions between different economic variables in terms of single equations and "Igeneral equilibrium relations;" the second group, dealing with historical simulation for the purpose of testing the predictive capability of the model; the third one being used for the purpose of projections. The difference between the first two is the number of endogenous variables and speciiications. For example, in the structural investigation the relative wage rates have been used as explanatory variables for the labor allocation between sectors of the economy, whereas in the historical simulation, labor allocation has been made exogenous to the system. Il this way, the comparatively weaker relations are removed in the latter case to improve the predictive capability of the model over the sample period in the major basic blocks. In the pro- jection exercise, the number of exogenous variables is further reduced by building new equations and making some of the exogenous variables endogenous to the system. These additional equations form block nine of the complete equations system as used for projections purposes. Block 10 refers to the Monetary Sector which has been used in the last simulation to explore the implication of the Prv. sent Goverrnment policy. 17. The data base of the estimated equations covers a period of 1955-1969 (in few cases a shorter period of 1960-1969)j and is mainly confined to official sources (see bibliography). Alternative sources have also been used to check the reliability. 1/ See for notations and equations, Appendix II.  18. The structure of the Argentine economy as revealed from the model: the agricultural sector has been divided broadly into two groups - (1) the livestock and (2) the crop sector. Livestock sector has been further sub- divided into three groups: (1) the cattle herd, (2) the cattle slaughter and (3) the residual. The crop sector, on the other hand, has been sub- divided into wheat, corn, sorghum and other agriculture. Live stock 19. The cattle slaughter decisions in Argentina has been largely guided by speculative price expectations. For example, whenever the cattle price has moved upwards, the slaughter of cattles has always declined. In order to explain this apparently paradoxical supply response, two alternative hypotheses have been tested. 'Wen the cattle price goes up in the slaughter market, cattle owners try to increase the stock of their capital (i.e., the cattle herds) in anticipation of a higher profit. This they achieve.by reducing the slaughter of young cows or cows of child bearing age. But this accentuates a further rise in the price of beef in the market. Eventually a point comes, as the cattle population of the country reaches a certain age-structure, when it becomes imperative that a given section of this popu- lation must be slaughtered, for attending an optimuI slaughtering age. Such a point is reached normally three to four years after the calves are born._/ But as soon as the number of slaughters increases, the price of beef starts falling, when the reverse phase of the cattle cycle begins. 1/ CATS = 12.21 + 2.796 RPB (-3) - 5.718 RPB (-1) + .184 t (cattle slaughter) (Real price of (Real price of (Tech. progress) beef 3 years beef one year ago) ago)  - 12 - 20. The second hypothesis is that when the prite of beef rises in the market, the profitability per catrle mainly meassured by the difference between the selling price and the feeding cost per cattle, increases and hence the owners try to increase the number of cattle by reducing their slaughterAi/ But in this process, a part of the cultivated land in the economy gets transferred from corn and fodder production to cattle breeding. This leads to a rise in the price of corn and fodder. As a result the 1tholdingtt price of cattle herd becomes less profitable, valued at the ruling market price and feed price when the slaughtering of cattles, starts increasing. 21. The changes in the cattle stock population has been explained in terms of the existing cattle population, the current rate of slaughtering and a time trend representing the improvement in cattle breeding process. In explaining cattle slaughter, hypothesis (1) has been found to be more suitable for the projection purposes. This is because, to work on the second hypothesis, one needs the estimate of the relative prices of beef and corn, which are extremely volatile, and are very difficult to predict. On the other hand, working on the first hypothesis is seen to be comparatively easy, especially for reproducing the cattle cycles over the projected period. However, over the historical period, on the basis of actual prices of beef and corn, the second hypothesis gave a better result, in predicting the actual cycle s. 1/ CATS - .465 - 6.685 RPB + 5.234 IPOO/IP (cattle '(real price (real price slaughter) of beef) of corn)  - 13 - Agriculture Crop Sector 22. The outputs of wheat, corn and sorghum have been explained in a t"three tier"t block structure. The decision as to how much land is to be brought under cultivation for a given crop, has seen to be a function of its profitability vis-a-vis other crops (measured in terms of their relative prices), the total availability of land and the adjustment time needed for switching over from one crop to the other. In this context, it has been seen that mainly the wheat, beef and sorghum are competitive between them-. selves, in the use of land vis-a-vis their relative price and cost structure. 23. The area to be harvested from a given total crop area (that is total sown area) has been made dependent, in most cases, on the rainfall in the relevant regions.- Finally, output has been made largely dependent on the area harvested. An attempt is made also to explain the output of the agriculture crop sector and of the livestock sector separately, by a simplified Cobb-Douglus procluction function.- The main explanatory variables, chosen for the agriculture crop sector, are (i) total capital stock and (ii) avail- ability of fertilizer and rainfall. The regression coefficients for this production function show the presence of a diminishing returns to scale in this sector. Moreover, in an alternative formulation it is found that any incremi,ents in labor input has practically no impact on the level of agri- cultural outputs in Argentina. The result is very interesting since 1/ e.g., LAWH.H - -2506 + .9875 LAWH + 15.6 RIKW4 (land wheat harvested) (land wheat (rainfall) planted) 2/ Log OAGC = 4.29 + .28 Log Kagst (-1) + .h7 Log Fert + .06 Dum. (Agricrop) (capital stock) (fertilizer) (rainfall)  -14- according to an alternative hypothesis, the low output of agriculture in Argentina over this period has been ascribed to labor migration away from agriculture. In our view, the "diminishing returns" in agriculture in Argentina could be due partly to a lack of technological improvements and partly a migration of skilled labor into the urban area (which has affected the quality of rural labor force). The total availability of agriculture labor force has been made a joint function of' total available labor force, (working population) and the relative wage rates between the agriculture and the non-agriculture sector.- It seems from our present analysis, the labor movements between the two sectors were largely responsive to the wage changes over the sample period. Non-Agricultural Block 24. This sector has been sub-divided into six broad categories: (1) the output of the capital goods industries (mainly metal based), (2) the output of consumption goods industries (mainly agricultural based), (3) the output of the construction industry, (4) the output of services, (5) the output of power and fuel and (6) the output of mining and extractive industry. Of the six ou4tputs sectors, construction has been treated residually although in an alternative formulation, it is seen to be related significantly with the capital formation of the country (with one year time lag). 25. The growth of output of the capital good industries has been largely demand-oriented; this demand has been, mainly generated by the level of activity in the capital formation sector of the economy. On the other 1/ LABAG - -33.7 + .23 FXT - 42.7 ERT/ERAGR - 16.1 T (labor in (employment) (wage rate/ (time agriculture) wage rate in trend) agriculture)  hand, the growth of the consumption good industries has been largely due to a very heavy import substitution in the economy. The output of power sector has been largely in response to the growing demand of the final consumers and of the intermediate sector. 26. We have formulated an aggregate production function for the non- agriculture sector '-r Argentina in terms of its total capital stocks.V available and labor force employed in the non-agriculture sector. It is observed that the output of the non-agriculture sector has largely been dependent on the last years capital stock (year ending) and the labor force of the present year. Marginal capital output ratio, in the non-agriculture sector (as partial derivative), is seen to be as high as 4.09 arnd is very close to the conventional measurement of capital/output ratio in this sector. This suggests a heavy capital intensity in the non-agricultural production in Argentina. 27. Investment in the non-agriculture sector has been treated mainly, by acceleration princiqles2/ although in the policy version of the model we treated it exogenously in order to trace the implications of alternative investment decisions. Labor supply for the non-agriculture sector has been made dependent on the relative sectoral wage rates and no attempt has boen made to deal with the idle capacities in the labor or capital stock. This has been necessitated because of the lack of suitable time series information on capacity utilization in Argentina. 1/ The capital stock vector has been computed by "perpetual inventory" method from a 50-year time series information on fixed investment in Argentina by assuming 30 years life time fQr a capital asset. 2/ KNAGR = -219 + .789 KNAGR (-1) + .30 ONAGR .23 ONAGR (-1) + 324 DUM2 (inv. in (output (output non- non-agri) non-agri) agri last year)  - 16- Consumption Block 28. This block in the model has been sub-divided into three broad sub- groups: (1) the total consumption and consumption of the private sector; (2) the consumption of specific items like beef, wheat and food, (3) the consumption of broad groupings like agriculture, industry, services and the residuals. 29. From equations number 1 and 2 of this block, it is clear that both beef and wheat have a high income elasticity of demand (beef having higher than that of wheat) on the other hand beef has a very low price elasticity. For wheat, there is no significant relationship between consumption and prices. This is so because the total expenses on wheat is so low in the aggregate expenditure basket of the household (wheat being one of the bare necessities of life) that the absolute level of price has rarely been taken into consideration by individuals in deciding their level of consumption. From equation 4,1 it is further noticed that over and above the impact of total income anid tax deductions, private consumption has been largely affected by the relative wage and non-wage components of the national economy. This can be easily explained when the wage earners are likely to have higher propensity to consume than the non-wage earners. From equations 6 and 7, it is clear that, in general, the price eleasticity (calculated at the mean) of food is lower and for industrial goods, higher, in Argentina. Price Formation Block and Monetary Block 30. In the price block, an attempt is made to explain the price changes in the aggregate an-d for eight different sectors. Some of them are in absolute 1/ PRICONS = 2oh + .77 DI + 8.2 W/NW (private (disposable (wage/non-wage)' consumption) income)  - 17 - terms and others are in index form. The general price level is explained in equation (1) mainly in terms of last years quantity of money (defined in a broader sense including currency and deposits) and changes in the nominal exchange rates. In an alternative more detailed specificationr2/ (block X), price changes have been explained in terms of Public Sector's borrowing from the Central Bank, Banking Sector Reserve Ratios, foreign assets and expecta- tions for future inflation. This is basical-ly a monetarist approach to the problem of inflation. But in micro study, we notice that in some sectors prices have been influenced by the movements of the world prices and by the changes in the domestic wage rates with different time lags. Theoretically, the ideal approach would be to compute a general price level calculated as a weighted average of all the sectoral prices. But in this model as we have not been able to cover all the sectors, we had to formulate the general price level, as distinct, from the sum of the sectoral prices. Furthermore, an inter-temporal sub-division of our sample period shows that different factors are predominant in contributing to inflation in different periods of Argentine history. To use a sweeping generalization, it seems that the price changes in the agricultural sector in Argentina have been more geared to cost changes (i.e., wage rate changes) whereas in the non-agriculture sector, prices have largely been affected by credit expansion by the Public Sector and exchange rate devaluations. Export Block 31. The export sect,or has been divided into seven subsector, six of them have their origin from agriculture and the last one from industry. 1/ Log IP - 18.2 + 1.13 Log PUBCB + .08 Log FORASST - .31 Log RR - Log PP'- 2.4 LogYPC (price index) (government borrowing) (foreign (reserve ratio) (popu- (GDP per assets) lation) capita) + 1.2 S C (Block X 1 and 2; changes in price capitation)  - 18 - This last sector has a contribution of nearly 20% to the total value of exports. 32. Exports of beef and wheat have been treated as residuals in this model. They have been expressed as a difference between the domestic production and domestic consumption. Ini an alternative version wheat has been made a function of domestic output, which is a slightly different specification, suggesting a supply constraint, in the exports. In this context, domestic production of beef has been assumed to be equal to slaughter of cows in the same year. This means that the possibility of any change in the holding of stocks has been ignored. Furtherg cattle slaughters, in this model, has been recordod as number of "heads." On the other hand, cattle consumed is recorded in physical units, in terms of metric tons. Hence, a functional relation between the number of cattle slaughtered and the average weight has been formulated in the equation ni 1.!/ Tis relationship is of the inverse form, as the heavier cattle are slaughtered mostly at the begin- ning of the upswing. This would mean that meat production cycles would be less intense than cattle slaughter cycles in the economy. 33. The estimates for exports of wheat have been calculated from domestic production after domestic consumpti6n has been satisfied. In such an approach, one sometimes wonders what the relevant role of world prices vis-a-vis domestic prices in deciding the volume of exports. In fact, the qxport duty of the Government (in the form of imposing an effective exchange rate) has been so operated as to make the domestic price almost equal to the 1/ NB - 240.4; - 2.9 CATS /rlock V, 9) (weight (no. of in lbs per cattles cattle) slaughter)  - 19 - world prices including duties, with a minimum of time lag. Therefore, the dealers are always kept on an indifference line between the domestic and foreign market. This has presumably been deliberately practiced in Argentina, partly to keep domestic prices downl/ and thereby people happy and partly for fiscal purposes. It has indeed been demonstrated convincingly in Argentina, that any change in the price of beef and wheat have an immediate impact on the general price level and the wage rates of the laborers. Therefore, in an attempt to control prices, this country has always tried to control the price of beef. But at the same time, it has overlooked that any lowering of the price of beef in an artificial manner would result in a shortfall in the export receipts and a balance of payment crisis. Further, its attempts to tackle this crisis by tariff raised in the price of its imported materials continuously and introduced an added inflationary element into the economy. So far as the export of corn2 is concerned, it seems that its expansion has been arrived at exclusively by opening a new market and reducing the cost of freight charges for transshipment abroad. For industrial exports, it seems that exports in general have been dependent on the volume of domestic produc- tion and the price competitiveness of domestic goods in export markets. 34. The level of imports of capital goods- in, Argentina has depended directly on the expansion of domestic demand (maitLly for domestic capital formation and inversely on the growth of the "import substituting capital 1/ The world dollar price of beef was rising faster than the domestic one. 2/ EXCO - -1526 + .38 EXITA - 2245 FCO (export of (new (freight corn) markets) charges) 3/ MCAP - 611 + .39 K(T) + 113 DtTM - 318 DP (capital (domestic (restrictive (import price/ goods invest.) measures) domestic price) imports)  - 20 - goods producing" sector of the economy). Thei average import component of capital. formation fell during the sample period from 20% to 10%. This heavy substitution seems to have been made possible partly because of the differ- ential price movements between the domestic and the foreign market, resulting from protective tariffs and exchange rate depreciation. Besides, there have been many other quantitative types of import restrictions especially in the years 1964, 1965, and 1966 which contributed to the heavy import substitution in these three years. But such a heavy import substitution in the capital goods seems to have affected the quality of capital over this period and can be regarded as one of the explanation for the very low capital productivity over the same period 1950 to 1969. 35 In the intermediate goods sector,- similarly a large import substitution has been arrived by introducing a differential price movement between comparable domestic and foreign'goods. But in spite of this, the incremental intermediate import component in Argentina is very high. Wage Block 36. The wage rates, both in the agricultural and non- agricultural sector, seem to have been influenced by the changes in the general price level (with a given lag) and by the price of beef with an almost instantaneous effect. An optimum relation between a change in the price level and wage rate has been worked out in Block X of the 'model. Public Finance 37. As to the public income_, it seems that in Argentina direct taxes at constant price are inversely related to the rate of growth of inflation 1/ MINT - -843 + .220 NAGR - 1.367 PIM (inter (output non- (world price adi. for duty) imports) agriculture) ( domestic price )  - 21 - and directly related to gross domestic product at constant price9l/ This means that with the control of inflation, direct tax receipts are likely to go down in Argentina. As to the export and import taxes, it seems that the import tax receipts are more elastic than export tax receipts in this country. As to current expenditure, it is largely dependent on the wage bill in the public sector and therefore wage rates are very important in control- ling both government deficits and inflation. 38. Coming to Block IX, which deals with certain economic rolations which have been used in the projection exercise, the most noteworthy are the relations explaining credit creation and the changes in the nominal exchange rates vis-a-vis general price level. It seems that the exchange rates changes in Argentina have been largely dependent on changes in the general price level of the country. This means that the Government of Argentina has adjusted the exchange rates very promptly with every change in the domestic price level. Summary 39. The Argentine economy seems to exhibit a dualistic structure, comprising of a rural agriculture sector and an urban industrial sector. Dualism has a very important economic bearing in the development process of this country. The rural sector in Argentina mainly contributes to the domestic consumption and exports, whereas the urban sector mainly contributes again to domestic consumption and investments and is the main recipient of imports. Hence, any imbalance in the growth between the two sectors immediately affect the balance of trade position of the country. 1/ DT = 534 + .416 OT - 616 j iP/IP (direct tax) (GDP) (rate of change of price)  - 22 - 40. As to the agricultural sector, it is suffering from diminishing return to scale. Marginal capital coefficients seems to be very high both in agriculture and industry (nearly 4 to 1). But as one looks at the aggre- gate ratio between total capital stock and labor employed in agriculture, it does not seem to be high at all. All the statistics point to the fact that agriculture has a very inefficient production contour because of lack of technological progress over the past few decades in Argentina. This has presumably happened because of the uncertainties of market conditions arising partly from the uncertainties of the international situation, regarding the wide price fluctuation of primary goods and partly due to uncertain govern- ment policies (with frequent revisions) regarding the exchange rates and lastly because of uncertainties of rainfall. This view has been further supported when we noticed that only that part of agriculture which is non- traditional and which supplies mostly to the stable protected domestic market, has employed better technology in the form of using fertilizer and better production techniques. Coming to the question of extensive cultiva- tion it has been noticed that the total land area although has been increased over the sample period, most of it has gone to,non-traditional, non-exporv,able sector of agriculture. Another feature of the Argentine agriculture which emerges from this model is its very uneven and cyclical nature. These fluctuations are partly to natural causes, partly to speculative behavior of the economy, partly to the nature of production and largely to bad policy measures;l/ the second, third and fourth contributing causes can be largely c%ontrolled by prudent policy measures and changes in the technological 1/ See equations' 1, 2, 10, 11, 13, 17, 22, etc.  - 23 conditions. For example, a proper control of the corn/beef prices and a credit policy in agriculture and in livestock mnight reduce the fluctuatiolis to a large extent. The removal of these fluctuations would also reduce the uncertainties of the market and thereby might create congenial conditions for bringing greater technological progress. As to the traditional crop sector, io is the relative price movements which can be regarded as the most likely important instrument variables in controlling their fluctuations. The non-agriculture sector of the Argentine economy shows a high return for labor. But its growth seems to have been constrained by the availability of imported inputs because of the foreign exchange constraint of the country. The domestic output in this sector has a very large import component and a high price elasticity for imports. This has led to a "policy of import substitution'? by protection for the growth of this sector. But the limits to this policy have been set by the pace of the domestic inflation when the gap between the domestic and foreign price is continuously eaten up by ci creeping inflation within the economy. Furthermore, this process of con- tinuous import substitution has affected the quality of the new capital goods and thereby the efficiency of the capital stock. Domestically produced capital goods in many fields have not become as productive as the imported ones. Evidently, in such a situation, one wonders whether any improvement in the growth of the economy is at all possible without a thorough change in her economic structure.I' Indeed, unless the economy moves to the higher production transformation by technical progress and by heavy technological 1/ With the recent changes in the terms of trade and political atmosphere of the country, the statement is not valid. Our third alternative simu- lation of the model presents the up-to-date picture of the economy.  - 24 - import substitutions and adopts behavioral disciplines in consumption and wage demands, the higher rate of growth will always bring with it a creeping inflation) a balance of payment problem and a typical stop go cycle (which in fact the economy is experiencing now). Consumption in the aggregate in Argentina is not high at all. This means investable'surplus available in the economy is not very low. However., because of the high replacement component of investment, a further addition to the investable surplus is desirable. But what is more important in Argentina is not so much the surplus, but the correct composition of the surplus. Attempt should be made to generate the surplus in those sectors which contribute mainly to the capital formation and to the supply of the material inputs for production. Bi., in the absence of sufficient and efficient capital goods producing sector within the country, it can only be achieved by exporting agriculture goods and importing corresponding.capital goods and intermediate goods from abroad. But unfortunately, the scope of curtailing consumption in the relevant sectors with exportable commodit-ies is not very high in Argentina using tradition policy instruments. For example, in the case of beef the' price elasticity is very low. Therefore, an indirect tax policy is not likely to succeed in curtailing consumption. On the other hand, it will add to inflation. Although the consumption of most of these commodities can be reduced significantly (as they have a high income elasticity) by direct taxes, the tax administering machinery in Argentina unfortunately has not proven to be particularly efficient in collecting direct taxes, as the tax evation is very common and tax recording is very poor. Therefore it seems the only solution is direct controls or rationing.  411. We would like to close the discussion with one or two words regarding the problerm of price formation in Argentina. It seems that the general price level of the country has largely been influenced by credit expansion (via the Public Sector), massive price speculation and the periodic exchange rate devaluations, although it is very difficult to conclude which of them starts the process. However, they all point to the need for controlling the infla- tion by monetary and fiscal discipline. But again, when we study the price of an individual sector it becomes clear that world price movements and changes in the wage rates within the economy also have a large part to play in generating inflation in the country. This is especially true for the agricultural sector. Also changes in wage rates seem to be significantly influenced by inflationary pressures in the economy, and especially by pr.ice changes in the beef market. 42. Hence, any attempt to control inflation should be combined both from the monetary side and from the structural side. This means (i) con- trolling wages and government expenditure, (ii) curtailing consumption by rationing, (iii) insulating the influence of the changes in the world price levels, (iv) improving productivity and following a well regulated monetary policy. 13. Tests of the Model's Validity 43. The statistic of individual equations give the predictability of partial relations: for testing the predictability of the overall relation- ships (in a general equilibrium sense), the model has been simulated over a historical period and its solution values have been tested against actuals.  - 26 - 44. In this connection the following variables have been treated as exogenous elements: a. population; b. rainfall; c. land under food grains; d. quantity of money in circulation; e. nominal exchange rates; f. world prices for beef, wheat, corn and sorghum; g. wage bill in the public sector; 45. Altogether;the model contains 117 equations and 117 urnmowns; the solution is obtained by iterative methods, solving period by period. In order to reach a convergence (i.e., the solution), we had to rearrange the equations in almost recursive blocks. The solution values are given in details in Table I. 46. In this exercise, the national income identity is satisfied at current prices by treating the "stock" variable as a residual. Thus, the stock variable represents the "error in prediction ,",in the model. In the "projection exrercise,It however, the unplanned stock estimates have been regarded to influence the price structure in the economy, and new solutions are reached by further price and resources reallocation, by an iterative process, until the unplanned stock vector had vanished. 47. The predictability of the model in this sense, has been seen to be quite satisfactory. The mean error!/ for GDP is seen to be as low as 2% over the whole sample period: 1960-69. But the errors in the prediction 1/ See graphs.  - 27 for "investments" are found to be high, althc,'.gh admitting the volatile nature of investment activities in Argentina, the results are not very disappointing. They are never more than 3.0% in the sample period. 48. In the cattle slaughter activity sector, the mean error is as high as 10.8%, but the directional predictions are correct in 9 of the 10 cases. However, the estimated cattle slaughter cycle is more damped than the actual. As for the other sectors, the estimates of imports for inter- mediate goods give the worst prediction with an average error of 20% over the sample period, although again directional predictions are always correct. The movement of macro estimates vis-a-vis their actuals are shown for major variables in the graphical annex. 49. In terms of average growth, the eatimates for gross domestic product at factor cost, the fixed investment, the exports (f.o.b.) and imports (c.ief.) compare with their actuals, as follows: Fixed Price GDP Investments Exports i