Yu~,oslsvEccnomic Performance i n t h e 1940s: A l t e r n a t i v e Scenarios Sherman Robinson University of California, Berkeley Laura D. Tyson Ijniversity of Califc=:iia, Berkeley NathLas 3 2 ~ a t r i p o n t Harvard Lnivers i t y January 1984 Develop~entResearch Department -. economics ar.d Research Staff World Bank The virr;s presented here a r e t h c o e of tir. a u t i ~ o r ,azd tkey silouid no; 5 e t r r 03 r a i l e c t i n ~those of t h e Kf:>rid Sank Yugoslav Economic Performance in the 19808: Alternative Scenario8 Page No. Introduction ...................................................... 1 ..................................................... The CGE Model 3 ..................................... Production and Employment 4 ................................................. Foreign Trade 5 Demand, Prices and Macro Closure .............................. 7 ........... External Conditions and Economic Performance, 1981-1984 8 ....... Alternative Policy Regimes and Macro Projections, 1983-1990 18 ........................................................ Conclusion 29 ............................................. References 31 Abstract This paper utilizes a multisector, computable general equilibrium (CGE) model to analyze Yugoslav economic performance during the 1981-1990 period. . The model is used to generate both counterfactual historical simulations for the 1981-1984 period and alternative forward-run scenarios for the 1984-1990 . period. The counterfactual simulations assess the relative contribution of domestic policy errors and adverse changes in external trade and credit market conditions on deteriorating economic performance between 1981 and 1984. These simulations illustrate the impact of the growing foreign exchange crisis on domestic growth and productivity performance. The forward-run scenarios examine growth and debt repayment prospects dur- ing the 1984-1990 period on the assumption that Yugoslavia will successfully implement a change in development strategy t~warda more open economy with increased exports. The new development strategy is assumed to be realized by an appropriate exchange rate policy that eliminates the excess demand for for- eign exchange and the complicated rationing schemes it has engendered. As a result, the bias in incentives against exports is removed, and both export and productivity performance improve. The forward runs also assume that Yugoslavia will maintain austerity measures to control domestic absorption, especially agqfegate investment, throughout the 1984-1990peri~d.~Alternative simula- tions analyze the implications of differences in the severity and composition 2 -- offs between the growth of domestic investment and real wages on the one hand # and the balance of payments and debt rqayment on the other. - YUGOSLAV ECONOMIC PERFORMANCE IN THE 1980s: ALTERNATIVE SCENARIOS* Sherman Robinson University of California, Berkeley Laura D.-Tyson University of California, Berkeley Mathias Dewatripont krvard University Introduction In the last decade, the Yugoslav economy has undergone a number of shocks. 'Ihere have been major changes in both domestic policy, including major institutional refons, and shifts in the world economy, including oil price increases and worldwide recession. Since about 1979, problems with foreign trade have dominated policy discussions as economic performance has become steadily more constrained by shortages of foreign exchange. Until quite recently, the Yugoslav response to foreign exchange shortages was to impose increasingly severe rationing of imports. Now, however, they are considering ways to increase exports and to implement what is essentially a shift in development strategy. Policy debate is currently intense and centers & on issues of moving toward a more open economy, with more reliance on market incentives, and removing the severe bias in incentives against exports that has characterized the system over the past decade. - - * w * *The work descrcbed in this paper was supported by the World Bank. fie views and interpretations in this document, however, are those of the authors and should not be attributed to the World Bank, to its affiliated organiza- tions, or to any individual ucting on their behalf. The policy debate i n YugosZavia has p a r a l l e l s i n many other countries. To what extent are t h e i r problems due t o external conditions beyond t h e i r con- t r o l , and t o what extent a r e they due t o f a i l u r e s i n domestic policy? Given the variety and severity of the shocks the economy has faced, what are the best policies t o pursue i n t h e 1980s? In t h i s paper we seek t o analyze these questions within the framework of a multisector, computable general equili- brium (CGE) model of the Yugoslav economy. Such models have been used t o analyze issues of structural adjustment i n other countries. l The frame- work is useful because it permits analysis with an empirical model that ex- p l i c i t l y incorporates structural relationships among many sectors and market interactions between producers and demanders. In a separate paper we have used the model t o analyze Yugoslav performance i n the 1976-1980 period. In that paper we concluded that, while external shocks were important, the policy response within Yugoslavia was inadequate f o r dealing with the problems which emerged during t h i s period. Resorting t o quantitative restrictions on imports led t o serious distortions i n the domes- tic market, especially large biases in incentives against exporters, under- u t i l i z a t i o n of capacity, and losses in kfficiency. Between 1978 and 1981, the second o i l crisis h i t , international interest r a t e s rose, and the world recession hurt export markets. The Yugoslav response t o these events was, 6 again, inadequate and created a complicated and confused incentive structure b e e , for example, Dervis and Robinson [I9821 and Dixon et a l . [1981]. Dervis, de Melo, and Robinson [I9821 survey the use of CGE models in develop- ing countries. %ee Robinson and Tyson [1983]. In this paper we take up the story in 1981. We first analyze the course of the economy in the 1981-i"94ileriod and then consider different medium-term scenarios for the 1984-1990period. In the next section, we describe the CGE model which we used for the analysis. The discussion is very brief, focusing only on the essential features of the model. More detailed descriptions, including specification of equations, are available elsewhere.' In the succeeding sections, we use the model to generate various counterfactual his- torical simulations and alternative forward simulations which explore the eco- nomic trade-offs that have faced, and will face, Yugoslav policymakers. The CGE Model The CGE model for Yugoslavia, like CGE models implemented for other developing countries, operates by simulating the operation of markets for factors, products, and foreign exchange. The model is highly nonlinear and involves the specification of conditions of supply and demand for all the markets. A solution for a given year generates market-clearingprices and quantities, including all the elements comprising the circular flow in the economy. The Yugoslavia model has 18 production sectors; 4 labor categories; 2 household types (rural and urban); and enterprises differe~tiatedby sector, . L government, a "rest of the worldu institution, and an aggregate capital ac- count which serves the role of tho banking system in gathering savings and allocaring lnvestmene runds to sectors. Depending on how one counts, the - model has around 1,Om to 1,500 equations which are solved for each period in * a dynamic simulation.' 'see World Bank [1983], Methodological Annex, pp. 301-334. The emphasis on markets and market-clearingmechaliisms in the CGE model reflects the view that the independent behavior of decentralized producers and consumers plays a significant role in influencing economic performance. How- ever, this view of the economy does not imply that markets are perfect or that the decision making of producers and consumers is necessarily guided by neo- classical rules of profit maximization or utility maximization. Instead, the Yugoslavia model explicitly recognizes the existence of rigidities and imper- fections in the economy and of special behavioral features of self-managed enterprises. Production and Employment In most CGE models, it is assumed that producers maximize profits given a neoclassical production technology in an environment of perfectly competitive markets. In the Yugoslavia CC;E ~odel,the assumption of profit maximization is replaced by a set of rules that attempts to capture the behavior of self- managed firms. The specification incorporates two basic features of such behavior: (1) rigidities in employment levels that limit the supply respon- siveness of the firm to changes in product market conditions and ( 2 ) payment rules that produce a divergence between the marginal value product of labor and the personal income or wage that labor receives. , We assume that output decisions of self-managed firms are guided by a form . . - ..... ".- .:.--I-. .-...- ..'!,----....-... n c - - L :-.- ?: "CC..,."-A +-,f-,,,,+ 5nr+;n., ^."A ? -their - labor force as fixed in the short term, analogdus to sectoral capital .. which is also assumed to be immobile in the short ru& Decisions on the amount of variable labor to be used in a given period are guided by a "plan- ning" or "accounting" wage. Distinguishing between variable and fixed labor has the effect in the model of reducing the employment variability and sec- toral supply responsiveness relative to what they would be in a world of perfect factor mobility and unconstrained profit maximization. The model of f i n behavior distinguishes between the planning wage that a f i n uses in deciding the level of labor utilization and the actual personal . income paid to members of the firm. Income payments can be seen as consisting of two parts: a planning wage component and a component that reflects the worker's share in the fin's net income. The size of the second component depends on both the legal and contractual obligations affecting the distri- bution of enterprise income and on the firm's decision as to what part of its disposable net income should be distributed to workers. In the model, these shares are set by various ccefficients which can either be fixed or determined endogenously in response to policy choices. The variation of these distri- bution shares in response to government policy is an important feature of the Yugoslav institutional system and of the forward simulations with the model (discussed further below). Foreign Trade Exports are treated in two different ways in the version of the model used here. For the counterfactual experiments, exports are exogenous and are var- 4 ied parametrically to explore the impact of different scenarios. In the for- * ward simulations, exports are dctenined endogenously. For each sector. t h ~ 5 - model incltldes separate world demand functions and domestic export supply - * functions (the latter specifi& the enterprise's decision to sell on the world * 8 - market rather than on the domc'stic market). On the import side, the model assumes that domestically produced goods and imports are imperfect substitutes. Elasticities of substitution vary across sectors, with the lowest e l a s t i c i t i e s i n raw material and capital goods sec- tors. Given t h i s specification, the demand for imports depends on the rela- t i v e price in the domestic market of domestically produced and imported goods. The world price of imports is assumed fixed (the "small country" assumption); but their price i n the domestic market depends on the exchange rate, tariffs, and premia, i f any. Thus, trade policy w i l l have an important effect on import demand. The relative importance of these factors in deter- mining domestic prices and demand depends on the relative shares of imports and exports in total domestic supply a s well a s on the trade substitution elasticity. The total demand for foreign exchange i n the model is determined by sum- ming desired imports across a l l sectors. The total is compared with the supply of foreign exchange arising from exports and net foreign capital in- flows (including remittances and reserve decumulation) . An adjustment mechan- ism must be specified to equate the supply and demand for foreign exchange. There are two versions used for the simulations. In the counterfactual ex- periments for the 1981-1984 period, the model uses an endogenous mechanism of import rationing which is intended t o capture in a stylized manner the elabo- r a t e foreign exchange ra~ioningsystem that the Yugoslavs have built up over the past few years. The mechanism' and its implications for economic perform- ~r ance are discussed further below. In the forward simulations, we assume that * .'._., ru.,,t.s--r.c,\ii u,.\-r ..ld-"l IL..h . 2 d C L . 8 . U - A ~ . - V . . L l l b -I - A I I . - . I Y " C . - . A . . _ ) C I I I - L D - i - the model assumes that Yugoslavia w i l l pursue an exchange rate p d i c y which is designed t o keep the real exchange rate roughly constant (for e x a p l e , a "crawling peg" policy), and then capital inflows w i l l adjust t o clear the foreign exchange market. kmand, Prices, and hlacro Closure The demand side of tne model works by tracing through the incomes gener- ated in the productive sectors of the economy and the various demands they induce. There are three categories of income recipients whose behavior is modeled : productive enterprises, households, and the government (or non- productive sector). The model contains an elaborate set of accounting and * behavioral rules to determine how value added (or fzctor income) is dis- tributed among income recipients.' To complete the circular flow, the saving and expenditure behavior of each ir!come recipient is specified, leading to demands for sectoral output for consumption and investment purposes. The problem of achieving macroeconomic savings-investment balance is a separate issue involving what has come to be called the macro "closure" of the model. There are a variety of ways discussed in the literature for achieving savings-investment equilibrium in CGE models. Different approaches are based on different theoretical views of how the macro system works: neoclassical, Keynesian, monetarist, Kaldorian, "structuralist," etc., with many variations and combinations. In the Yugoslavia model, we use different specifications for the counterfactual simulations and the forward simulations. In the counterfactual simulations, the model is run with neoclassical closure. In- vestment is determined by the sum of savings from a l l sources: enterprises, government, households, and foreign or rest of the world. The institutional - lThese rules are dkcribed in detail in World Bank [1983]. - 2 ~ o further discussion of issues of macro closure, see Taylor [1983] and r Robinson and Tyson [1984]. rate, are all exogenous. Some of these parameters are varied in the counter- factual simulations as part of the alternative scmarios. In the forward simulations, we specify a different macro story which is designed to reflect the particular institutional framework in Yugoslavia. In this jpecification, inflation is endogenous and nominal personal incomes (or wages) are treated as exogenous. The details are discussed below when we present the forward simulations. External Conditions and Economic Performance, 1981-1984 As a result of outstanding debt repayment requirements, coupled with poor export performance, Yugoslavia was forced to strengthen its domestic austerity program in 1982-83 in order to cut imports. In 1982 and 1983, had Yugoslavia been able to export more, or to borrow more (or, equivalently, to pay 5ack less), a smaller dose of austerity would have been possible and the perform- ance of the economy would have been better. To a large degree, domestic policy errors were responsible for Yugoslavia's difficulties. In particular, / the persistence of an overvalued dinar exchange rate and import rationing schemes produced a bias in incentives against exports that has played a major role in Yugoslavia's poor export growth since the mid-1970s.l Nonetheless, Yugoslavia's 1982-836conomic difficulties were not all of its own making'. Like other net-debtor,newly industrializing countries, Yugoslavia was ad- - r r r -.-A1 U L - C L L C d U I C,IC/LIIU.l~bJA11i C 3 i r \ C b L I I u L c L V I I ~ I I I A L~ I I d L L U I I I I I b I L C . \ A ja sharp slowdown in the growth of its export markets and (2.) a tightening of - extkrnal credit market conditions. l~ora discussion of the extent of dinar overvaluation and the bias against exports, see Robinson and Tyson [1983]. In this section, we discuss the results of simu1at;ons of the CGE inodel designed t o assess the impact of these changes on the Yugoslav economy. All the simulations begin in 1981, the base period for the current version of the CGE model. 'Ihe f i r s t simulation (the base run) reproduces the actual course of the Yugoslav economy in 1982 and 1983--consistent with information avail- able a t the time the simulation was completed (August, 1983)--and produces a projection for 1984. 'Ihe 1984 projection, in turn, depends on a number of endogenous variables whose values are determined by 1982-83 history and on a number of exogenous variables whose values are based on various Yugoslav and World Bank estimates. The base run scenario depends on a number of behavioral assumptions about the functioning oE the Yugoslav economy and also on assumptions about external conditions confronting it. For the purposes of this paper, the most important behavioral assumptions regarding the domestic economy concern the effects of a foreign exchange shortage. The base run simulation takes the supply of for- eign exchange and the exchange rate a s given in each year. The 1982-83 values for the s ~ p p l yof foreign exchange are based on actual historical values for export earnings, net remittances, net foreign capital flows, and reserve ac- cumulation (decumulation). 'Ihe 1984 values for these exogenous variables are based on projections as of summer 1983. The 1982 di~ar-dollarexchange rate is set a t its actual period average value. 'Ihe 1383 period average exchange , . .- - L devaluaton realized during the f i r s t half of 1983 by a combination of internal - stabilization measures and further exchange rate adjustinentc ~hroughthe end . of the year. For 1984, it is assumed that the Yugoslavs w i l l continue t o maintain a constant price level deflated exchange rate. Therefore, the period :.erage dinar-dollar exchange rate in nominal t e n s is projected t o depreciate just enough t o offset the projected differential between Yugoslav and world inflation rates. Given the exchange rate, the demand for foreign exchange is detevlned endogenously i n the model for each year by summing desired imports across a l l sectors. In 1982 and 1983, the model results indicate continued substantial excess demand for foreign e.tchange, given the actual exchange rate policy i n those two years. The projection for 1984 is for more of the same, although the actual extent of excess demand is expected t o declinz, primarily a s a result of improved export earnings. In the presence oE excess demand for foreign exchange, the Yugoslavs have relied on a complicated s e t of rationing rules t o allocate the available sup- ply among competing users. The CGE model contains an elaboratz specification of how these rules operate and captures their effects on real economic per- formance in several ways.' First, would-be importers who are unable t o import a l l that they desire arc forced t o substitute domestic goods. Since most imports are intermediate inputs into the production process and since domestic goods are d i f f i c u l t t o substitute for imported goods, a shortage of foreign exchange has a direct real cffect on domestic output and growth. . , Second, the model reflects the fact that import rationing schemes cause uncertainties and interruptions in the flow of crucial imports t o domestic i users wlth consequent dlsruptiorls i n domes~icproduct~onand growth. 'Ihe i z model captures these second effects in its assumptions about the exogenous s - I rate of growth of total factor productivity. For 1982 and 1983, the rates of t o t a l factor productivity growth are actually negative (-4.0 and -2.0 percent l ~ o more discussion of the workings of the rationing rules in the CGE r model, see Robinson and Tyson [I9831 and World k n k [1983]. in manufact1 1r jr,;l,, respectively) reflecting the sharp slowdown in domestic output that occ~rred,despite continued growth in both labor and capital inputs. The rate of total factor productivity growth is expected to remain negative through 1984 (at -2.0percent a year). Finally, the model assumes that there are real output costs associated with the rent-seeking activities engendered by Yugoslavia's import rationing mechanisms. The existence of substantial unsatisfied demand for foreign ex- change implies the existence of substantial scarcity or rental income, which provides powerful incentives for "rent seeking" behavior as would-be importers seek access, through nonmarket means, to foreign exchange at the overvalued official exchange rate. Such rent-seeking behavior generates real costs as resources that might otherwise be used in production are diverted to rent- seeking activities.' In the Yugoslav institutional setting, these activi- ties include such things as enterprise lobbying for import allocations in communities of interest, negotiating complex hidden arrangements for the sale of foreign exchange at a premium rate among enterprises, and political lobby- ing to restrict the flow of foreign exchange across regional boundaries. 'Ihe CGE model assumes that these activities waste domestic resources and reduce output. . U I The effects of the actual 1982-83 foreign exchange shortages and the pro- jected 1984 shortage are apparent in the 1981-1984growth rates generated in the base run simulation, whish are given in Table 1. The base run results show an average annual growtfi rate of 1.4 percent for gross domestic product * (GDP), 1.6 percent for personal consumption, and -2.3 percent for gross fixed investment. These growth rates are dramatically lower than the average annual 1See Krueger [I9741 for a discussion of the theory of rent seeking. 7% 7 growth rates of 5.9 percent for GDP, 7.2 percent for private consumption, and 6.1 percent for gross fixed investment realized in the 1976-1980period; and they are lower still than comparable rates for the 1971-1975 period.1 To assess the impact of adverse changes in external trade and capital mar- ket conditions on Yugoslav economic performance during the 1981-1984period, the results of the base run simulation can be compared with the results of appropriately specified counterfactual simulations. In the first counter - factual simulation, called C-1, we assume a relaxation of capital market constraints that allows Yugoslavia to increase its net foreign borrowing to reduce sharply import rationing in 1982 and 1983 and to eliminate rationing altogether by 1984. In addition to increasing borrowing, simulation C-1 assumes that the relaxation of the foreign exchange constraint and the concomitant increase in crucial imports would lead to an increase in total factor productivity growth of one percentage point a year in the nonagricul- tural sectors. All other assumptions are the same as in the base run. Table 2 presents cumulative trade flows and balance of trade figures for the simulations, and Table 3 presents various macro variables for 1984. To achieve the objective of eliminating 'importrationing by 1984, the model results indicate the need for additional net cumulative borrowing of $5.1 . I billion between 1982 and 1984. The substantial increase in borrowing required to eliminate the excess demand for foreign exchange by 1984, even with a major real devaluation in 1983 and a projected improvement in exportrgrowth in 1984, attests tro the severity of the foreign exchange crisis during Xhe 1982-83 - period . - * 'see World Bank [1983], p. 362. TABLE 1 Counterfactual Simulations: Growth Rates, 1981-1984 Simulation variablea Base run C-1 C-2 er,-e*t . . . . . . , . . . . 0 ,,, r, Gross domestic product 1.4 3.1 3.6 Private consumption 1.6 3.4 3.4 Gross fixed investment -2.3 2.2 1.3 Exports 2.0 2.0 6.4 Irnpor ts -1.3 2.3 2.6 Real wage, manufacturing -1.1 0.2 3.7 aExports and imports include services. All variables are in real terms (1981 prices). TABLE 2 Counterfactual Simulations: h u l a t i v e Trade Flows, 1982-1984 - Simulation Variable Base run C-1, C-2 billions of current dollars Exports 47.89 47.89 53.24 Balance of trade - 2.19 - 2.90 0.74 Reflecting the model's treatment of the direct and indirect costs of a foreign exchange shortage and the rationing mechanisms it sets in motion, the C-1 simulation indicates the effects of the additional borrowing on Yugoslav economic perfonance. The simulation results show 1981-1984 growth rates of 3.1 percent for GDP, 3.4 percent for private consumption, and 2.2 percent for gross fixed investment (Table 1). Even with substantial additional borrowing, aggregate growth rates remain much lower than those realized in the 1970s. This is, in part, the result of the projected slow growth in investment, albeit at a positive rate, in the C-1 simulation. Equally important, however, are the assunptions about total factor productivity which is projected to con- tinue to decline during the period, although at slower rates than in the base run. While an improvement over the base run, these assumptions are, nonethe- less, very conservative in that total factor productivity is not ass~nnedto increase in the short run at rates realized in the 1970s. The fundamental underlying assumption is that past disruptions and continued nonmarket ad- ministrative interventions in the economy would prevent realization of im- provements in efficiency, even if the shortage of intermediate imports had been relaxed.1 Although the C-1 simulation indicates that Yugoslavia would have been able . , 6 to grow more rapidly in the 1981-1984period if it had been able to borrow more, it says nothing about the wisdom of-sucha borrowing stratew. Addi- ticnal external loans during this period iould have added to Yugoslavia's debt-servicingproblems in the future, an; a prudent borrowing strategy would b l ~ e eNishimizu and Page [1982], who have measured total factor produc- tivity growth by sectors and regions in Yugoslavia, and discuss reasons why productivity growth has been so slow there; see also World Bank [19833. have required a comparison of the real cost oE external funds with the ex- pected real benefits from their use. The C-1 simulation is instructive, none- theless, because it gives an idea of the growth-borrowing trade-off that confronted Yugoslav policymakers during the 1981-1984 period. A comparison of the additional borrowing in the C-1 simulation with an estimate of Yugoslavia's cumulative interest payments in 1982 and 1983 also suggests the effects on the Yugoslav economy of the shortening of maturities and the climb in interest rates in international capital markets. Yugoslavia paid an estimated $2.1 billion in interest in 1982 and is projected t o pay about the same amount in 1983. Relative t o the outstanding stock of debt, these figures convert to an effective interest burden of about 12 percent i n both years. Similar figures for the 1979-80 period indicate that interest payments were only 6.5 percent of the outstanding stock of debt in those years. If, in the absence of worsening credit market conditions, Yugoslavia had been able to maintain the same relationship between interest payments and debt i n the 1982-83 period, its interest payments would have amounted t o about $2.3 billion. Seen from t h i s vantage point, about half of the additional funds that would have been required for Yugoslavia t o eliminate its foreign exchange shortage by 1984 can be attributed to the effects of deteriorating 6 conditions in external credit markets. In the second counterfactual simulation, called C-2, we assume that Yugo- b l d v ~ A ~ U T L S grew at an average annual rate of 6.4 percent in 1982 and 19th. Such perfonance implies a reasonable (nonrecessionary) growth in world t&de * E - and assumes that Yugoslavia would have been able t o maintain its share in world markets. This perfonance is in sharp contrast t o the base run i n which the world recession leads to no projected growth in world trade in 1982 and 1983 and only very slow growth of Yugoslav exports (2 percent a year). We also assume, as i n C-1, that the Yugoslavs would have been able t o borrow any additional funds required t o eliminate their foreign exchange shortage by 1984 under these more optimistic export earning assumptions. In addition, in simulation C-2, we assume a further increase i n total fac- tor productivity growth of one percentage point a year in the nonagricultural sectors (yielding rates of -2.0, 0.0, and 0.0 percent for 1982, 1983, and 1984, respectively). This l a t t e r assumption reflects the view that increased exports would have improved productivity through a combination of better re- source allocation and, implicitly ,a lessening of administrat ive interference i n market incentives. Thus, the scenario does not involve just an improvement of world market conditions for Yugoslav exports but also assumes a Yugoslav response t o the improved export opportunities. The results presented in Table 2 indicate that the more rapid export growth i n simulation C-2 yields additional cumulative export earnings of $5.35 billion between 1982 and 1984 and that additional cumulative foreign borrowing of $1.45 billion would have been needed in the same period--all compared t o the base run. The GDP growth is better in simulation C-2, but consmption growth is the same a s i n C-1 and investment growth is lower able 1). The reason for t h i s mixed aggregate performance is that the increased export earn- ings are used partly t o offset the increased borrowing generated in simulation compared t o C-1, which represents about 6 percent of GDP in 1984. Once again, the results a t t e s t t o the severity o q the foreign exchange constraints facing Yugoslavia in the 1981-1984 period. In the C-2 scenario, Yugoslavia is able t o reduce the constraint through both additional export earnings and additional borrowing. This is a preferable solution compared t o the pure borrowing scenario of the C-1 simulation bec.luse it implies a smaller debt-servicing burden i n the future. Table 3 provides some macro indicators for 1984 f o r the base run and the two counterfactual simulations. A s one iiould expect, the higher borrowing simulation (C-1) increases t o t a l investment and the share of foreign savings in t o t a l savings. The higher export simulation (c-2) leads t o roughly the same investment share i n GDP as i n the base run and a similz? structure of savings. Manufacturing real wages a r e also highest i n t h i s simulation, a s would be expected with a significant increase in manufactured exports. Finally, it is important t o note that a comparison of the C-2 results with the base run simulation provides only a rough estimate of the cumulative effects of adverse changes i n external credit and trade conditions on the Yugoslav economy between 1981 and 1984. The comparison undoubtedly overstates these external effects because the C-1 simulation implicitly assumes that Yugoslavia's actual export performance in 1982-83 was largely the result of adverse external conditions. In fact, an overvalued dinar and other domestic policies that produced a bias in incentives against exports played a large role. The counterfactual simulations a r e optimistic i n that they assume that , I Yugoslav exporters would have been able to respond t o the improvement in export opportunities. r Alterna'tive Policy Regimes and Macro Projections, 1983-1990 - * * In making future projections of Yugoslav perfonance, it is necessary t o project the policy choices that w i l l be implemented. In t h i s section we explore the implications of a variety of scenarios, focusing on the trade-offs among different constraints and macroeconomic policies. In all these experi- ments, we start from the assumption that Yugoslavia will implement success- fully a change in development strategy toward a more open economy with increased exports. Also, we assme gradually improving world conditions, especially with regard to Yugoslav export markets. Tables 4 to 6 give the results. In the first experiment (F-11, the change in development strategy is modeled by assuming that policymakers quickly remove the historical bias in incentives against exports. After the large real devaluation in 1983, it is assumed that Yugoslavia will maintain a constant price level deflated exchange rate throughout the projection period (to 1990). The macro assumption is that policymakers do not try to use the exchange rate to counteract inflation but use nominal devaluations to correct for any differential between Yugoslav in- flation and inflation in her trading partners. It is also assumed that import rationing is quickly eliminated, although the levels of official tariffs are assumed to remain unchanged. As discussed earlier, in the forward-runningversion of the model, exports are detenined endogenously with explicit foreign demand curves for Yugoslav goods. In simulation F-1, the result is that exports grow at an average annual rat8 of 6.5 percent for the 1983-1990period (Table 41.' Compared to the experience of other semi-industrial countries, this export growth rate is ( ~ U LG L IIIUULAd LC. vci, I U W C & A VGII pd5 L ~ u & O > l d \ l)IeLLulll~dr1ct:d~lu~nt:COIUUS~Q . i - incentive structure, it is probably best to err on the conservative side in I * . - forward simulations. Experiment F-1 assumes that there is a very slight fall in the share of investment in GUP over the 1983-1990 period and that the share of foreign TABLE 4 Foward Simulations: Growth Rates, 1983-84 and 1984-1990 - Simulation variable" p-1 F-2 F-3 F-4 percent Growth rates (1983-84) Gross domestic product 2.9 2.3 2.0 2.9 Private consumption 2.7 4.3 3.0 3.5 Gross fixed investment 0.4 -10.6 -11.1 -10.0 Exports 8.0 9.2 10.1 11.2 Imports 6.3 2.3 0.5 1.0 Nominal wage, manufacturing 26.2 25.5 20.8 20.9 Real wage, manufacturing 0.6 3.0 1.8 2.3 Gross domestic product deflator 25.6 20.4 17.7 17.1 Growth Rates (1984-1990) Gross domestic product 4.1 3.8 3.8 4.4 Private consumption 3.5 3.3 2.8 3.2 Gross Fixed investment . 3.6 3.0 3.0 3.5 Exports 6.3 6.3 6.7 7.6 Imports 3.9 * 3.9 3.7 4.1 Nominal wage, manufacturing 17.5 17.5 16.7 16.9 deal wage, lrrarluracruring 1.5 1.1 0.6 1.0 1 - Gross domestic product deflator 15.3 15.5 15.4 15.0 -- ** R - QNational accounts variable; are in real tens (1981 prices). TABLE 6 Forward Simulations: Selected Indicators, 1990 Simulation Variablea F-1 F-2 F-3 F-4 Indices (1983 100) Cross domestic product 130.9 128.3 127.6 133.6 Private consunption Capital stock Real wage, manufacturing Gross domestic product deflator Ratios (percentlb Private consumption/gross domestic product 53.1 54.9 52.7 51.8 Fixed invcstment/gross domestic product 22.9 19.2 19.3 19.2 Export s/gross domestic product 34.0 36.1 38.5 39.8 Imports/gross domestic product 28.7 28.8 29.1 29.3 Foreign srrvlngs/total savings - 19.3 - 30.4 - 39.14 - 43.4 Ijltcrprise savings/total savingsC 48.3 46.5 56.1 61.4 Government savings/total savingsC * Ijltcrprise savings/value addedC - h t i o n a l accounts variables are ln current domestic prices. - 8 Je - eEconmywide, including agriculture and services. savings in t o t a l savings f a l l s significantly a s Yugoslavia repays its foreign debt (Table 6). These projections reflect stated Yugoslav policy objectives. It is assumed that total factor productivity growth w i l l recover by 1985, gradually increasing t o one percentage point a year by 1987 and remaining a t t h a t rate. This projected increase is assumed t o f o l l o r ~from the s h i f t in development strategy, a s has occurred in other countries pursuing such p o l i - cies.' The assumed increase i n t b t a l factor productivity growth, i n f a c t , 2 is quite small when compared t o that in other semi-industrid countries. The result is t h a t the GDP growth r a t e is 3.9 percent a year for th2 1983-1990 period with some acceleration i n the l a t t e r p a r t of the period. Simulation F-1 represents an optimistic scenario but is conservative in many of its behavioral. assumptions. The fundamental optimistic assumption is t h a t the Yugoslavs are able t o implement a s h i f t in incentives t o achieve a more open development strategy. Given t h i s s h i f t i n policy, the concomitant assumpt ions about world condit ions and the responsiveness of the economy a r e conservative, especially when compared t o the experience of other semi- industrial countr'es that have faced similar d i f f i c u l t i e s , for example, l'brkey. Simulation F-1 is a l s o based on f a i r l y conservative assumptions about the a b i l i t y of Yugoslav policymakers t o maj.ntain macro balance. The fixed invest- Ir ment rate, which the Yugoslavs a r e seeking t o lower, is assumed t o decline slightly--to 22.9 percent of GDP in 1990 (down from aboat 26 percent i n the . I ~ U -I,U.J A LUUI I I I C ~ ~ ' U J ~ C LllulaCloIl ~ C I raLe IS assumed LO be 16. L percent - w *- bee Nishimizu and Robinson [1983] for evidence of this relationship. Balassa [I9821 a l s o discusses why one would expect links between an open development strategy and f a s t e r productivity growth. 2 ~ e eGenery [I9841 for a survey of evidence from a number of semi- industrial countries. a year on average over the 1983-1990period with significant deceleration over time (Table 4). The assumed maintenarce of a roughly constant price level deflated exchange rate leads to balance of payments and debt-repayment projectior.5that are consistent with projections by Yugoslav authorities and interrlational agencies. Given the projections of real investment and the balance of payments, domestic savings adjusts to achieve macro balance. In simulation F-1, the major adjustment is through changes in government savings which are projected to increase as a share of total savings. In this simula- ti~nthe government does not implement a strong incomes policy (i.e., it is unwilling to restrict enterprise payments to workers) and, hence, penits a small, gradual increase in real personal income of about 1.1 percent a year. Thus, these macro projections reflect a continued policy effort to control government expenditure (collective consumption and nonproductive investment) and a relaxation of administrative controls on personal incomes. We have also done a number of simulations with the CGE model to explore the implications for future Yugoslav perfonance of different assumptions about the mcroeconomic policy options available to policymakers. In these experiments we have specified a different macro-equilibrating mechanism (or macro "closure") in the model than that used in the historical simu1atio"n. In these experiments, the aggregate investment rate out of GDP, the exchange rate, and rhe level ot personal Irlcui~rt:oi4ulKels ~ L G L,ubLi.uuaA,. aggregate price level adjusts endogenously to achieve a savings-investment balance. The implicit assumption is that the government, by controlling the level of nominal personal income, is able to control the distribution or sav- ings decisions of enterprises. The price level adjusts so that, through inflation, real incomes are made consistent with macroeconomic equilibrium. The inflationary mechanism implicit in this specification is a kind of "cost push," but inflation is also sensitive to the level of investment demand. For a given level of investment, an increase in personal income will lead to higher prices. muever, for a given level of nominal income, increased in- vestment will also lead to higher prices, lower real income, and, hence, higher enterprise saving. This view of the inflationary mechanism in Yugoslavia is rather special in that it reflects particular Yugoslav institutional features. For example, the way personal incomes are set by enterprises is assumed to be independent of decisions about employment, so that inflation has no feedback on aggregate employment. mere is evidence that this macro specification is appropriate for Yugoslavia, but one would not want to apply it to other countries indiscriminately.1 In addition to simulation F-1, which might be considered as a forward base run, we have done three other simulations which explore the trade-offs among macro forces and policy choices. All these experiments start from F-1 and are thus also conditional on its major policy assumptions: (1) a similar exchange rate policy, (2) an export incentive system that allows exporters to respond to change in prices, and (3) ipprovements in world conditions after 1983. The simulations all start from the same 1983 solution and include the same behav- 9 and inflation.- - - Simulatioe-2 assumes a lower investment rate out of GDP than in F-1, - starting in 1984 and continuing throughout the period. The fixed investment rate falls to 19.2 percent (in 1990) compared to 22.9 percent in F-1. 'see Tyson [1977]; also, see Horvct [19711 and Tyson [19801. Simulation F-3 adds t o F-2 a policy regime limiting nominal wage in- creases. As a consequence of wage restraint, the rate of growth of nominal wages is 4.7 percentage points lower in 1984 and about 1 percentage point lower thereafter. Simulation F-4 adds t o F-3 higher total factor productivity growth assumed t o accompany the s h i f t i n development strategy. The t o t a l factor productivity growth rates are -1.0, 1.5, 1.5, and 1.5 percent a year for the nonagricul- tural sectors i n 1384, 1985-86, 1987-88, and 1989-90. The comparable rates for the other simulations are -2.0, 0.5, 1.0, and 1.0 percent a year. The impact effect of these experiments h i t s i n 1984, and then the economy moves along a different path for the r e s t of the period. The results are given i n Tables 4 t o 6. Table 4 indicates clearly the impact of the different scenarios i n 1984. Compared t o F-1, the f a l l in investment i n the other ex- periments is dramatic--a minus 10-11 percent growth rate in 1984 compared t o a 0.4 percent rise i n F-1. The impact of the wage restraint in simulations F-3 and F-4 is also evident. In simulation F-2, the decline i n investment is significant. The invest- ment share in GDP is about 4 percentage points lower throughout the period, and the capital stock i n 1990 is about 5 percent lower than that i n , t h e F-1 simulation. The policy has the desired effect on the balance of trade. Lower i - - rate, leads t o a real devaluation in 1984. Exports Increase and, due largely * - t .. o lower growth, import demand decreases. The net m s u l t is that cumulative net foreign capital inflows are $15.5 billion less than in simulation F-1 (Table 5). m e slower inflation actually leads t o higher real wages in 1984, since the ~ominalwage growth is roughly the same a s i n F-1. However, the slower GDP growth over the period leads t o slower real wage growth a s well, finally eroding most of the i n i t i a l gain in real wages (see Table 6). In simulation F-3, the assulned wage restraint policy leads t o a lower in- flation rate i n 1984 but lower growth of real wages compared t o F-2. The lower inflation leads t o an even higher real devaluation i n 1984 and a higher export growth rate. The net effect is a much greater change in cumulative foreign capital inflows ($23.2 billion less than i n F-1 and $7.7 billion less than i n F-2). The wage restraint policy is thus successful i n that, coupled with the decrease i n investment, it leads t o even better balance-of-payments performance. GDP growth is lower in 1984, although the difference is slight by the end of the period. Real wages, however, never recover and remain sig- nificantly lower a t the end of the period. ?he final simulation, F-4, assumes that the s h i f t i n development strategy t o increased exports and reliance on market mechanisms leads t o higher t o t a l factor productivity growth. In t h i s more optimistic scenario, exports in- crease even more a s t o t a l output is higher. GDP growth is higher than i n F-1 and much higher than i n F-2 and F-3. Higher exports lead t o improved balance- of -trade performance, with cumulative net foreign capital inflows of $3.2 billion lower than F-3 (and $26.6 billion lower than F-1). Real wages, of ..-.....-.. ,.7 ..._ ..,... -.- , P-.&-. . , . . 4 . 1..... t . P . . I 7 . . - n n n level. - Table 6 indicates that the four scenarios have quite different implica- t tions for the composition of savings. They a l l s h ~ wa dramatic increase i n the share of government savings over time and a dramatic decline i n the share of foreign savings.. The better the trade performance, the more debt repayment and the higher the negative share of foreign savings in total savings. In simulations F-1 and F-2, the share of enterprise savings falls over time. In the two wage restraint scenarios (F-3 and F-41, enterprise shares of total I savings increase slightly over time, although enterprise savings as a share of total value added ends up in 1990 slightly loher than in 1984. I 'Ihe results from the various forward scenarios indicate the importance of links between a change in incentive policies and supporting macro policies in determining the success of a new development strategy. A real devaluation is I needed early in the period in order to remove the bias in incentives against exporting which has been very strong in the Yugoslav system. The real de- valuation will work only if enterprises can respond to the shift in relative prices and reallocate resources to increase the production of tradables--both exports and import substitutes. A real devaluation, however, cannot be achieved without first achieving control over macro balances. The reduction of aggregate investment is a major part of the macro package. A nominal in- comes policy is also useful because it lowers inflation and so supports a real devaluation early in the period that will not be eroded quickly over time. Note that, in contrast to an IMF policy package, the goal of an incomes policy is'no'tto reduce real wages (and's0improve "competitiveness") but to help control inflation and support a real devaluation. The evolution of real -.*--.*.-- --r"."a "." L.,' -'." .>.A d ' L L - ' L.. IL.b&",aII"l.C =? i - strategy. In particular, it depends on the speed of recovery in total-factor ) productivity growth and capacity utilization and on policies with regaFd to Conclusion The forward-run experiments are cautiously optimistic in their assumptions about Yugoslav policy choices through the end of the 1990s. All of the ex- periments are optimistic about Yugoslavia's ability to pursue an appropriate exchange rate policy, thereby eliminating the excess demand for foreign ex- :' change and the complicated rationing schemes that have been used in the past and are in place currently. As a result, the bias in incentives against exports is removed, and the attendant productivity and rent-seeking costs of these schemes disappear. All of the forward runs are cautious in their assessment of the likely consequences of these developmellts for both export and productivity performance. If the recent experience of other semi- industrial countries is any guide, the actual consequences should be even more favorable than those assumed in the simulations. As far as macro policy is concerned, all of the forward simulations re- flect Yugoslavia's stated policy objective to reduce its external indebtedness through the end of the 1990s. Consistent with its experience in the 1981-1983 period, and in earlier stop phases of its recurrent stop-gocycles, Yugoslavia is likely to continue to pursue this objective through a combination of aus- terity measures to cut domestic investment'and/or to increase domestic sav- ing. Austerity measures are likely to continue to include administrative . ..---*.-..--..-O f U S A I u &LAIC.~ U l t 6 1 L U A I I lUL I I~ULULLUII i of enterprise decisions withjegard to personal incomes and savings. The - forward simulations analyze 3 e implications of differences in the severity and composition of such austerity measures for economic performance. The results illustrate the trade-offsbetween the growth of domestic investment and real wages--indicatorsof domestic absorption--onthe one hand and the balance of payments and debt repayment on the other. None of the macro policy assumptions are unrealistically severe--atleast in the light of actual 1981-1983 performance. Despite the assumption of continued domestic austerity measures, all of the forward runs show improvements in overall indicators of ecor~omicgrowth and absorption relative to the 1981-1983period. The recovery implied in the results critically depends on the assumption that Yugoslavia successfully in- troduces a more open, export-orienteddevelopment strategy. ' As the his- torical scenario of the 1981-1984 period suggests, austerity in the absence of such a strategy is likely to produce continued economic stagnation. lAs of this writing (December, 19831, there are some worrisome signs ~hnt continued policy errors in the system of foreign exchange allocation will undermine the shift in incentives necessary to support the new strategy. References Balassa, B., and associates [1982]. Development Strategies in Semi-Industrial Economies. Baltimore: #JohnsHopkins University Press. Chenery, H. B. [1984]. "The Semi-Indust-rialC~untries,"Development Research Department, The World Bank, mimeo. Dervis, K., and S. Robinson [1982]. "A General Ejquilibrium Analysis of the Causes of a Foreign Exchange Crisis: The Case of Turkey," Weltwirtschaftliches Archiv, Band 118, Heft 2. Dervis, K., J. de Melo, and S. Robinson [1982]. General Equilibrium Models for Developnent Policy. Cambridge: Cambridge University Press. Dixon, P. B., et al. [1982]. Orani: A Multisectoral Model of the Australian Economy. Amsterdam: North-Holland Publishing Company. Horvat, B. [1971]. Business Cycles in Yugoslavia. White Plains: Interna- tional Arts and Sciences Press. Krueger, A. [1974]. "The Political Economy of the Rent-Seeking Society," American Economic Review, Vol. 64, No. 3, pp. 293-303. Nishimizu, M., and J. Pagc [1982]. "Total Factor Productivity Growth, Technological Progress and Technical Efficiency Change: Dimensions of C Productivity Change in Yugoslavia, 1965-1978," Economic Journal, Vol. 92, No. 368, pp. 920-936. in Semi-Industrialized Countries," Discussion-Paper No. 52, Development s Research Department, The World Bank. Forthcoming, Journal of Development Economics. Robinson, S., and L. D. Tyson [1983]. "Foreign Trade, Resource Allocation and Structural Adjustment in Yugoslavia: 1976-80," Discussion Paper No. 57, Development Research Department, The World Bank. Robinson, S., and L. D. Tyson [1984]. Wodelling Structural Adjustment : Micro and Macro Elements in a General Equilibrium Framework," in H. Scarf and J. Shoven, eds. Applied General Equilibrium Analysis. Cambridge: Cambridge University Press. Taylor, L. [1983]. Structuralist Macroeconomics: Applicable Models for the Third World. New York: Basic Books. Tyson, L. D. [1980]. The Yugoslav Economic System and I t s Performance in the - 1970s. Berkeley: Institute of International Studies, Research Series No. 44. Tyson, L. D. [1977]. "The Yugoslav Inflation: Some Competing Hypotheses," Journal of Comparative Economics, Vol. I, No. 3, pp. 113-146. -> World Bank 119831. Yugoslavia: Adjustment Policies and Development Perspec- - tives. Washington: World Bank.