Output, Devaluation and the Real Exchange Rate in Developing Countries By
Pierre-Richard Ag~nor C o n t e n t s : I. Introduction. - II. Contractionary Devaluation in LDCs: the Evidence. - III. A Model of Output Determination with Intermediate Imports. IV. Estimation Results. - V. Summary, Conclusions and Extensions. - Data Appendix.
I. Introduction he use of devaluation as a stabilization tool in developing countries has increased in the past few years, as shown by the growing use of active exchange-rate policies in adjustment programs supported by the International Monetary Fund.l At the same time, however, the role of exchange-rate policy in economic adjustment has become the subject of considerable debate. The "orthodox" view [see, e.g., Guitifin, 1976, and Dornbusch, 1988] has maintained that devaluation plays a positive and important role in balance-ofpayments stabilization (through its expenditure-switching effects and increased production of tradables), while the New Structuralist school has stressed the contractionary and otherwise perverse effects of exchange-rate adjustment. Several theoretical arguments have been advanced by New Structuralist economists to explain why, contrary to the traditional view, a
T
Remark." I would like to thank, without implicating him, Peter Cornelius for helpful comments on an earlier draft. The views expressed in this paper are my own and should not be attributed to the International Monetary Fund. According to data presented by Johnson et al. [1985], of 85 upper credit tranche stand-by arrangements approved from 1963 to 1972 by the I M F (a period during which emphasis on maintaining par values prevailed), 26 (31 per cent) involved action on the exchange rate or the exchange system. O f the programs that did not include such action, about half were designed to correct over-expansionary demand management policies, and the remainder were intended to facilitate external debt rescheduling or to deal with problems (such as the impact of a recession and temporary shortfalls in export receipts) for which exchange-rate action was not considered a necessary solution. F r o m 1973 to 1980, exchange-rate action figured in 53 per cent of programs supported by stand-by arrangements and extended arrangements supported by the I M F - a considerably higher proportion than in the previous decade. More than half the stand-by arrangements involving exchange-rate changes included an initial devaluation. During 1981-83, the incidence was 64 per cent.
Ag6nor: Devaluation and the Real Exchange Rate
19
devaluation can have a negative effect on output. The emphasis was first put on the demand side, most notably by Diaz-Alejandro [1963]. 2 Using a three-goods model of exportables, importables and home goods with relatively price-inelastic exports and imports and consumption functions based on higher saving propensities for non-wage earners, he shows that a devaluation may lead to a reduction in real income. The mechanism producing this result lies in the transfer of real income from workers, who receive a fixed nominal wage, to capitalists with higher propensities to save out of factor income. Krugman and Taylor [1978] have formalized and extended this view, and various others have followed [e.g., Barbone and Rivera-Batiz, 1987]. In addition to these demand-related effects, the New Structuralists have recently discussed a number of supply side channels through which devaluations can be contractionary. Bruno [1979], for example, rationalizes the negative impact of devaluation by postulating that in a typical semi-industrialized country where inputs for manufacturing are largely imported and cannot easily be substituted for by domestic production and where the working capital that firms depend on from banks is subject to rationing, a sudden devaluation will sharply increase firms' input costs and need for working capital. But with credit rationed, these funds can only be obtained in the informal loan market where the sharp rise in demand may drive interest rates to such prohibitive levels that firms may choose instead to reduce production. The negative impact from the higher cost of imported inputs will weigh against the production stimulus from higher relative prices for domestically traded goods. If, as seems likely, there is some delay in realizing the positive effects of the devaluation, the negative impact of higher cost imports on aggregate supply may dominate in the short run. More recently, van Wijnbergen [1986] has also provided several mechanisms, focused on the supply side of the economy, through which a devaluation may cause an adverse effect on output (besides being inflationary). In particular he stresses, like Bruno, the impact of a devaluation on the local currency costs of intermediate inputs, as well as its negative effect on the volume of real credit to firms needing funds to finance working capital. He also discusses the effect of parity 2 One of the first analyses of the impact of exchange-rate devaluation on output and employment is to be found in the work of Hirschman [1949], who showed that a devaluation, starting from an initial trade deficit, may lead to a fall in real income as long as the increase in spending on importables exceeds export receipts.
20
Weltwirtschaftliches Archiv
changes o n external debt p a y m e n t s 3 and on n o m i n a l wages in the presence o f wage indexing and f o o d imports. T h e i m p o r t a n t result f r o m the new a p p r o a c h is that once supply-side effects are t a k e n into account, it is possible for a d e v a l u a t i o n to be c o n t r a c t i o n a r y even if the net effect on aggregate d e m a n d were e x p a n s i o n a r y (i.e., the expenditure-switching effect d o m i n a t e d the e x p e n d i t u r e - r e d u c i n g effect), as long as aggregate d e m a n d shifts by less t h a n aggregate supply. Empirical evidence so far o n the c o n t r a c t i o n a r y effect o f exchange-rate changes on e c o n o m i c activity has been largely inconclusive. T h e aim o f this p a p e r is to p r o v i d e some f u r t h e r evidence o n the N e w Structuralist view, by using a f o r m a l m o d e l o f o u t p u t d e t e r m i n a tion and estimating the m o d e l o n a cross-section d a t a set for 23 developing countries. T h e rest o f the p a p e r is o r g a n i z e d as follows. Section II provides a brief overview o f empirical evidence available so far o n the relationship between d e v a l u a t i o n and o u t p u t b e h a v i o u r in developing c o u n tries. Section III presents a simple m o d e l o f aggregate o u t p u t , based on a competitive m a c r o - e q u i l i b r i u m f r a m e w o r k with rational expectations. Section IV d i s c u s s ~ the e c o n o m e t r i c m e t h o d o l o g y , a n d presents the empirical estimates o f the model. Section V p r o v i d e s a summ a r y o f the results, and discusses perspectives for f u r t h e r research.
II. Contractionary Devaluation in LDCs: the Evidence 4 F o u r alternative empirical a p p r o a c h e s have been followed in the literature to analyze the effects o f a d e v a l u a t i o n o n o u t p u t (and, m o r e generally, on m a c r o e c o n o m i c p e r f o r m a n c e ) . T h e first, s o m e w h a t inf o r m a l or factual a p p r o a c h , examines changes in c o u n t r y perform a n c e at the time o f d e v a l u a t i o n a n d is usually referred to as the 3 The deflationary pressure induced by a devaluation deriving from the presence of a large private external debt denominated in foreign currency was first stressed by Cooper [1971, p. 492]. His argument was that a devaluation increases both the outstanding debt and the debt-servicing burden in terms of domestic currency. The former development may throw some firms and individuals into technical bankruptcy, and the latter will reduce their net earnings. Gylfason and Risager [1984] have attempted to estimate empirically the importance of the contractionary effects of debt servicing. They conclude - unsurprisingly - that it can have a significant output-reducing impact in heavily-indebted countries. 4 Lizondo and Montiel [1989] have recently provided a critical survey of the literature on contractionary devaluation in developing countries. They, however, concentrate on analytical issues and do not discuss available evidence. The literature survey by Hamilton [1988] provides only a sketchy view of empirical results.
Ag6nor: Devaluationand the Real ExchangeRate
21
"before-after" methodology. The second procedure, known as the "comparison" or "control" group approach, consists in comparing performance in devaluing countries with performance in a group of non-devaluing countries. The third strategy applies econometric methods to time series in order to determine the impact of exchangerate changes on real output. A fourth, somewhat less direct, approach to the problem uses simulation models or reduced-form equations to analyze the effects of exchange-rate variables on economic activity. 1. B e f o r e - A f t e r
Approach
One of the first studies using the before-after approach was DiazAlejandro's [1965] analysis of the experience of Argentina over the period 1955-61. He showed that the 1959 devaluation of the peso was contractionary, because it induced a shift in income distribution towards high-propensity savers, which, in turn, depressed consumption and real absorption. In a seminal paper, Cooper [1971] surveyed twenty-four devaluation episodes involving nineteen different developing countries that took place between 1959 and 1966 and assessed statistically the extent of the response of the balance of trade and payments, inflation, and the elements of aggregate demand. While overall his results showed that the trade balance (measured in foreign currency) and the balance of payments did improve in most cases, he also found evidence of contractionary tendencies following a devaluation. Cooper's study suffers, however, from two major weaknesses. First, he examined only short-run or impact effects of a devaluation (i.e., effects which take place within a year), while in most cases a devaluation can hardly be expected to have its principal - let alone its sole - effect in the following year. Second, the before-after approach he used is based on a strict ceteris paribus assumption and will not yield an estimate of the independent effect of a currency depreciation on output whenever other determinants (domestic as well as external factors) of the outcome are changing as between the pre-devaluation and the post-devaluation period. 2. C o n t r o l - G r o u p
Approach
The control-group methodology allows in principle to overcome the inability of the before-after approach to distinguish between the effect of devaluation per se and the effect of other factors on output. The basic assumption is that devaluing and non-devaluing countries
22
Weltwirtschaftliches
Archiv
are subject to the same external environment. Therefore, by comparing before-after changes in output growth in devaluing countries to those in the control group of non-devaluing countries, the effect of the external environment (or factors unrelated to devaluation per se) will cancel out - leaving the difference in group performance to reflect only the effect of exchange-rate changes. Recent studies applying the control-group approach to the specific question of devaluation in developing countries are those of Kamin [1988] and Edwards [1989 a, 1989 b]. Kamin's research exploits data for a set of 50 to 90 devaluations out of a sample of 107 that took place between 1953 and 1983. He uses a control-group methodology and performs formal statistical tests for significance of changes over time for each economic indicator he considers, for both the devaluing country's performance and the comparison's group performance, and for the difference between the two. Among many counter-intuitive results, s Kamin finds that a contraction in economic activity, in the sense of an actual decline in output levels, is not typical of most devaluations; the growth rate remains in general positive. Second, a sharp and significant decline in output growth is registered by the devaluing countries, but it occurs in the year preceding the devaluation. This lower rate appears to be maintained virtually unchanged through the year following the devaluation. Growth in subsequent years turns upward in absolute terms and improves relative to the comparison group. In sum, in Kamin's view, there is little evidence that devaluations are followed by significant contractions in output; recessions prior to devaluation appear to be more typical of the devaluation process. A somewhat different perspective is provided by Edwards [1989 a; 1989 b, pp. 320-324], who analyzes the evolution of a number of key variables during the three years preceding and the three years following 18 devaluation episodes that took place between 1962 and 1982 in Latin America. Like Kamin, he also constructs a "control group" consisting of 24 developing nations that maintained a fixed nominal exchange rate over the same period, and compares its behaviour with that of the devaluing countries. Non-parametric statistical tests (variants of the Mann-Whitney U-tests for differences in means) are used in these comparisons. His results suggest that the observed decline in 5 Kamin shows, for example, that the reduction in trade deficits found to follow devaluations is attributable to sharp increases in exports rather than decreases in imports, a result which seems to contradict the conventional "short-term elasticity pessimism" found in much of the literature.
Ag6nor: Devaluationand the Real Exchange Rate
23
output growth in periods surrounding devaluations may not in fact be a consequence of exchange-rate changes but may rather reflect the effects of a massive imposition of restrictions and other policies that have accompanied devaluations in Latin America. The point is indeed important; devaluations in developing countries are almost always one of many components of a stabilization package. It is thus difficult to separate the effect of the devaluation itself from that of the accompanying macroeconomic policies (and, in particular, from the trade liberalization reforms) often implemented alongside a nominal parity change. Several studies based on the control-group approach have also examined the impact of devaluations on output in the broader context of stabilization programs supported by the IMF. In an initial study, Donovan [1981] examined twelve IMF-supported devaluations between 1970 and 1976, comparing the performance of the devaluing economies with that of all non-oil exporting developing countries. Among other results, he found that meaningful reductions in G D P growth were registered only for those programs specifically aimed at import restraint. However, in a subsequent study [Donovan, 1982], where the sample of IMF-supported programs was extended to 78 (covering the period 1971-80), he found that the rate of economic growth fell by more than the average decline experienced by non-oil developing countries in the one-year comparisons, but by less in the three-year comparisons. Gylfason [1987] also used a comparison-group approach in his study of 32 IMF-supported programs implemented during 1977-79. He found that differences in output growth between countries with IMF programs and the reference group of non-program countries were not statistically significant. Finally, Khan [1990] has recently conducted a comprehensive evaluation of the effects of IMF-supported programs on the balance of payments, the current account, inflation and growth in a group of 69 developing countries covering the period 1973-88. Using regression analysis to isolate the impact of policy variables on macroeconomic aggregates, and performing twoyear comparisons (i.e., averages of target variables over periods t and t + 1), he finds that real exchange-rate changes have a negative effect on the rate of growth of output, but the coefficient is small and not highly significant in statistical terms.
24
Weltwirtschaftliches Archiv
3. E c o n o m e t r i c
Approach
Little econometric evidence on the relationship between devaluations and real economic activity in LDCs has been produced so far. A well-known paper is the study by Edwards [1986], who uses data on 12 developing countries for 1965-80, and estimates a model of real output behaviour. His results (which are discussed in greater detail below) indicate that real exchange-rate changes have a small contractionary effect in the short run. In the medium run, however, the perverse effect is completely reversed. In the long run, devaluations are neutral. Econometric evidence has also been reported by Sheehey [1986]. He employs a Lucas-type supply function, and uses cross-country data for 16 Latin American countries to estimate the impact on shortrun output growth of unanticipated inflation, changes in the relative cost of foreign exchange, and business-cycle fluctuations in the industrial countries. His results provide strong support for the contractionary devaluation hypothesis for Latin America, as well as suggesting a strong impact of external economic activity on real growth rates. In subsequent work, Edwards [1989b] has developed a multi-sector macroeconomic model of a dependent economy with imported intermediate goods, foreign debt and wage indexation to analyze the way in which devaluations affect aggregate output and employment. The model makes, however, no distinction between anticipated and unanticipated changes in the behaviour of key macroeconomic variables, a dichotomy which plays a crucial role in the model developed below. The regression results (based on a sample of 12 developing countries with annual data covering the period 1965-84) corroborate his earlier findings that devaluations have a short-run contractionary effect on real output. The long-run neutrality result, however, is found to hold only in two cases. 4. M a c r o - S i m u l a t i o n
Approach
Evidence from macro models on the impact of devaluation on real income is provided by Gylfason and Schmid [1983], Gylfason and Risager [1984], Branson [1986], and Solimano [1986]. 6 Gylfason and Schmid, for example, construct a log-linear macro model of an open 6 In addition, Khan and Knight [1985] have examined the simulation properties of a small dynamic econometric model estimated on a pooled cross-section time-series sample of 29 developing countries. They, however, discuss the impact of alternative policy packages rather than devaluations p e r se.
Ag6nor: Devaluationand the Real Exchange Rate
25
economy with intermediate goods. Theoretically, they show that a devaluation has conflicting effects: on the one hand, it generates an expansionary effect through aggregate demand; on the other, a devaluation has, through its effect on the cost of imported intermediate inputs, a negative effect on aggregate supply. The empirical relevance of the model is evaluated for a sample of ten countries (both industrialised and semi-industrialised) using a "calibration" procedure to estimate the key parameters of the model for each country. Their results show that a devaluation is expansionary (it raises real income) in eight out of ten countries which form the sample. The exception, among developing countries, is Brazil where the Marshall-Lerner condition is not satisfied. The expansionary impact, through expenditureswitching effects, of a devaluation seems therefore in general to dominate contractionary effects. In contrast, Gylfason and Risager [1984] develop a model which stresses the effects of exchange-rate changes on interest payments on external debt. Using also a calibration procedure, they show that while a devaluation is likely to be expansionary in developed economies, it is in general contractionary in developing countries. A similar result is obtained by Gylfason and Radetzki [1985] for a larger group of developing countries, in a model emphasizing the role of wage indexing. In addition to these multi-country studies, several authors have constructed country-specific models to analyze the effectiveness of exchange-rate changes on output. Solimano [1986] for example, using a computable macro model for Chile, shows a currency devaluation to be contractionary in the short to medium run. His analysis focuses on three factors: the structure of the trade sector in terms of the response of trade flows to changes in relative prices; the relative intensity of domestic value-added with respect to imported inputs in production across export and import competing industries; and the degree of wage indexing. A stagflationary effect of devaluation is also obtained by Branson [1986] for Kenya, using a two-sector model with sticky prices, wage indexing, and imported intermediate goods. Finally, Roca and Priale [1987] have examined within the framework of a detailed macroeconomic model the experience of Peru during 1977-78 and 1980-82, a period during which the Peruvian authorities attempted to reduce the current account deficit by lowering the real exchange rate through large nominal devaluations. Because a large portion of loans to the business sector were contracted in US dollars, the successive devaluations not only increased the price of imported inputs, but resulted also (as discussed in the analytical papers by
26
Weltwirtschaftliches Archiv
Bruno [1979], and van Wijnbergen [1986]) in a rise in the real cost of credit, pushing up the cost of working capital for highly-indebted firms and generating a strong stagflationary effect. This brief survey of empirical work suggests that, so far, evidence concerning the contractionary effect of devaluation on real output is, at best, mixed. Part of the diversity of the results may be attributable to the diversity of the research methodologies employed by different authors. Each method is subject to a number of inherent limitations. As mentioned earlier, studies using the before-after approach do not take into account the behaviour of other variables like monetary and fiscal policies, external disturbances and structural changes. Another well-known drawback of this approach, moreover, is that focusing on "before" and "after" makes it harder to detect causality among variables. While the control-group approach copes with some of the problems of the before-after methodology, it also has limitations [see Goldstein and Montiel, 1986]. The crucial problem is that devaluing countries do differ systematically from non-devaluing countries prior to a devaluation episode, and this matters considerably when evaluating the impact of a currency change on output. In principle, this problem can be dealt with (see e.g., Gylfason [1987], and Khan [1990], in a different context), but considerable difficulties arise in practice. Studies using a macro-simulation approach have as a major advantage the fact that they usually provide considerable information on the transmission process of exchange-rate changes to output, contrary to factual approaches. However, most of the studies available so far use imputed parameter values; the reliability of results derived from a set of "guesstimates" and coefficients that are not consistently estimated in an integrated framework can rightly be questioned. Econometric studies available so far are also subject to a number of limitations. In two of the case studies discussed above [Edwards, 1986, and Sheehey, 1986], the specification of the output equation estimated is essentially ad hoc. This raises well-known simultaneity and specification bias problems. Moreover, more recent work by Edwards [1989 b] does not provide a rigorous basis for distinguishing between anticipated and unanticipated movements in variables. This distinction is, however, a critical one in a rational expectations framework. In what follows, further evidence on the contractionary devaluation issue is provided by using an econometric approach based on an explicit derivation of the aggregate output relationship under rational expectations.
Ag6nor: Devaluation and the Real Exchange Rate
27
IIl. A Model of Output Determination with Intermediate Imports7 Consider a small open economy with a fixed exchange rate, facing a perfectly elastic supply schedule for its imports. Two (composite) goods, which are imperfect substitutes, are traded in the economy. One good is produced domestically, and the other abroad. The foreign good is used both for domestic consumption and as an intermediate input in the production process of the domestic good; its price (in terms of foreign currency) is exogenously given to the small country. 1. A g g r e g a t e
Supply
Following, e.g., Marston and Turnovsky [1985] and Hardouvelis [1987], a two-level production process is assumed. Gross output of the domestic final good, Q = F [ N , q~(L, K)] is produced using imported goods (raw materials), N, and domestic value-added V = c ~ ( N , K), where L denotes labour and K the capital stock, assumed fixed in the short run. The production function F is assumed CES, while the value-added function ~b(.) is Cobb-Douglas in L and K. Thus: O = B [aN -Q + (1 - a)(Lx -v K v)-o] - 1/~ exp (g), 0
(1)
B denotes a multiplicative constant and es a white-noise productivity shock. Setting K = 1 and taking a Taylor series expansion around an initial equilibrium yields, as a first-order log-linear approximation to (1), q = c l n + c 2 ( 1 - v) l + e ~, (1') where c a =ct(Q/N) ~
0 < c 1< 1
c2=(1-oO(Q/P) ~ 0
.
Throughout the paper, lower-case letters denote the logarithm of the corresponding level variable, c 1 and c 2 are the initial shares of imported materials and domestic value-added (~7 and P') in domestic output. 7 The contractionary effect of a devaluation on aggregate supply via intermediate imports has been formalized not only by Bruno [1979], van Wijnbergen [1986], Gylfason and Schmid [1983], and Gylfason and Risager [1984], as mentioned above, but also by Buffie [1986], Hanson [1983], Schmid [1982], and Shea [1976].
Weltwirtschaftliches Archiv
28
Producers are assumed to be risk-neutral and thus choose the short-run inputs N and L to maximize expected profits. Let W denote the nominal wage rate and Pathe price of the domestic good. The price of imported raw materials, expressed in domestic currency is SP~, where S denotes the nominal exchange rate (the number of domestic currency units for one unit of foreign currency) and P~ the world price. From the first-order conditions, the derived demands for labour, Ia, and raw materials, na, can be expressed in log-linear form as la = -- (1/v) (w -- Pa)- - (C1/VC2)e + F.s/vc2 (2)
na= --[(1--v)/v](w--pd)--(1/c2) [tr +c, ~ - ] e +(1/e2)Ia+ lvV]es,
(3)
where e = (s + p,-Pa) denotes (the logarithm of) the real exchange rate, and tr = 1/(1 + Q)> 0 the elasticity of substitution between imported inputs and value-added. Equations (2) and (3) show that the higher the product wage or the relative price of raw materials, the lower the producer demand for labour and raw materials. Also, the higher the weight of raw materials in the production function, ct, the more sensitive the demand for labour and the demand for imported intermediate goods are to changes in the real exchange rate. Following the Friedman-Phelps hypothesis, labour supply (in loglinear form) is assumed to depend on the expected real wage
l~=fl(w-pa),
fl>0,
(4)
where pa denotes the logarithm of the price level expected to prevail in the current period, based on information available up to last period. For simplicity, and without loss of generality, the intercept term in (4) is set to zero. Equation (4) states that workers deflate the nominal wage by their expectations of the current aggregate price level. Equation (4) reflects therefore a behavioural asymmetry between firms and workers: while firms react to the price of their output (P~), consumption behaviour is based on the price of the basket of consumption goods. The overall price level is defined as a weighted geometric average of the price of domestic goods (Pd) and imported goods (P,), expressed in domestic currency:
e=~(se~) ~-~,
(5)
or, in log terms, p = ,~ p,, + (1 -,~) (s + p.).
(53
29
Ag6nor: Devaluation and the Real Exchange Rate
Labour market equilibrium is assumed to hold continuously, so that setting P = Id yields, from (3), (4) and (5') and since w-pal = ( w - p ) +(1--6)e, and w - p ~ w-p=
(6)
- - 2 f l v ( p - - p a ) - - 2 [ ( 1 --6)+ c2..]Cl]e+ c__s
where 2= 1/(1 +vfl)< 1. Substituting (6) in (2) and (3) yields in = 2 f l ( p _ p . ) _ v_~2 1 (1-6)+ ~
a e [ (1 - v ) ( / - 2)((1-6) + ~2) + b-~21
nd = 2 f l ( l - v ) ( p - p " ) -
+ F(1-v)(1-2) + z ] e . [_
(7)
e + vc2
VC2
(8)
C2J
In turn, substitution of (7) and (8) in (1') yields (9)
q = ~tl ( p -- pa) + o~2 e + ota es,
where a~ = 2/~(1 - v) > O,
<0,
_(I-v)(1-2) cl ~3 =+ -- ~>0. YC2
C2
Now, the measure of value-added defined in (1) is expressed in terms of physical inputs. In terms of final goods, we have Y = Q - (S P,/Pd) N ,
(10)
where Y measures the real income accruing to the domestic factors of production (capital and labour), in contrast to V which measures the real net output of these factors. Equation (10) indicates that real income differs from gross output by the amount of real factor imports. Taking a log-linear approximation to (10) yields y = (q/c2) -- (cl/c2) e -- (cl/c2)n.
(10')
Substituting (9) and (8) in Off) yields finally the aggregate supply equation for net output: f =~l ( p - p " ) + y2e + y3 es,
(11)
30
Weltwirtschaftliches Archiv
where 71 =0~1 > 0 ,
Y3-
(1 -v)(1 vc2
>0.
Equation (11) indicates, first of all, that real income is independent of the elasticity of substitution a, a and equals real (gross) output only when c 1 =0, i.e., when there are no imported raw materials. Second, it shows that a rise in the real exchange rate (a depreciation of the domestic currency in real terms), has an unambiguously negative impact on output. A real depreciation or, equivalently, an increase in the relative price of the imported input expressed in domestic currency, entails a fall in the real wage rate (equation 5). Domestic producers reduce the demand for both labour and imported inputs (equations 7 and 8), so that output falls. 2. A g g r e g a t e
Demand
We now specify the demand side of the model. The aggregate demand relationship is assumed to take the form
yd=bl(m-p)+ b2e + ba(g-p)+ b4ys +ea,
(12)
where all coefficients are positive and all variables are measured in logarithms except e d, which denotes a stochastic shock with zero mean. Equation (12) states that aggregate demand depends positively on real balances (m denoting the logarithm of the nominal stock of money), real government spending (g denoting the logarithm of nominal public expenditure), foreign real income, Yl, (which affects exports) and the real exchange rate. This last effect comes as a result of the assumption of imperfect substitutability between domestic and foreign g o o d s . 9 The positive value of b 2 reflects both the Laursen-Metzler effect on domestic absorption and the assumption that the Marshall-Lerner condition is satisfied.
8 The same result would hold in a disequilibrium Gray-Fischer framework where nominal wages are indexed on the price level and employment is demand-determined. See Hardouvelis [1987]. 9 A similar aggregate demand equation is used e.g. by Chopra and Montiel [1986].
Ag6nor: Devaluation and the Real Exchange Rate
3. T h e O u t p u t - R e a l
Exchange
Rate
31
Relationship
We now calculate the relationship between output and the real exchange rate, under the assumption of rational expectations. Imposing equilibrium in the commodity market (yS = ya) yields, from (11) and (12), p=f2-1[blm+ 71pa+(b2-72)e+b3o+b4 yf +e], (13) where O = bl + b3 + 71 and e = e a - Yae~. Taking expectations conditional on information available up to the previous period and solving for p~ yields pa = ( f 2 - 71)- 1 [bl ma + (b2 - 72) e~ + b3 ga + b4 y~].
(14)
Unanticipated movements in domestic prices are derived by subtracting (14) from (13), so that p - - p ~ = f 2 - x [bl ( m - m a ) + ( b t - ? 2 ) ( e - e ~ ) + b 3 ( g - g ~)
+ b 4 (yy - y~) + e].
(15)
Finally, substituting (15) in (11) yields the following equation for output: Y = zl (m - ma) + z 2 (e - e ~) + z3 (O - g") + z, (Yj" - Y~) + zse+r ,
(16)
where zx = 71 b x / O > O ,
Za =71 b4/t2 > O,
T2 ----71 (b2--72)/Q > O,
"c5----72< 0 ,
9 3 = 71 b3/~ > O,
~ = 71 ed + 73 [1 - (~,l/t~)] d.
Equation (16) indicates that real output depends on unanticipated movements in the real exchange rate, the money supply, government spending, foreign economic activity, as well as on actual changes in the real exchange rate. To derive the effect of an anticipated increase in e, set e = e ~ in (16) and differentiate. The result is d y / d e a = 7 5 < O . The effect of an unanticipated increase in e is d y / d e (i.e., e ~ is held constant), which, from (16), is z2 + zs. F r o m the definition of f2, it can readily be seen that this sum is positive, so that d y / d e > 0. Therefore, an anticipated depreciation of the real exchange rate has an unambiguously negative impact on real output, while an unexpected depreciation has a positive effect on output. The impact of a real exchange-rate devaluation on real output is, in this model, quite different from the results obtained in the literature.
32
Weltwirtschaftliches Archiv
In models where the distinction between anticipated/unanticipated movements plays no role (as in Hanson [1983], Schmid [1982], and Gylfason and Schmid [1983]), the impact of a devaluation is theoretically ambiguous, lo This result derives from two conflicting factors: on the one hand, a devaluation generates an expansionary effect through aggregate demand (via substitution effects between imported goods and domestic goods, as well as its effect on exports); on the other hand, it has, through its effect on the cost of imported intermediate inputs, a negative impact on aggregate supply. As a consequence, a devaluation can be contractionary even if the net effect on aggregate demand is expansionary. By contrast, in the rational expectations model developed here, the distinction between expected and unexpected changes in the real exchange rate plays a crucial role. F r o m (14), an anticipated depreciation of the real exchange rate - which can be seen as an adverse "supply shock" - translates into a rise in the expected price level. F r o m (4), workers will thus increase their nominal wage demands (by (12-71)-1 (b2-y2)de"). As a result, the demand for both labour and imported inputs falls and consequently, output also falls. By contrast, an unanticipated devaluation has no effect on prices and the real wage. it It gives rise, however, to an unanticipated increase in domestic demand as the relative price of domestic output (unexpectedly) falls. In turn, this implies an unanticipated increase in prices, which stimulates aggregate supply. The above results illustrate the importance of a proper derivation of the output-real exchange rate relationship under rational expectations. The arbitrary procedure of simply adding expected/unexpected components, whithout explicit formulation of the underlying supply and demand curves, may provide an incorrect formulation for econometric analysis.
lo It should be emphasized, however, that most authors have considered the case of a nominal devaluation rather than a real change in the exchange rate, as is done here. Although empirical evidence provided by Cooper [1971], Kamin [1988], Himarios [1989] and Edwards [1988, 1989a, 1989b] shows that large nominal devaluations have in general been associated with a significant (and, in some cases, lasting) real exchangerate depreciation, in a theoretical context the distinction is crucial. The nominal exchange rate can be treated as exogenous in a fixed-parity regime, but the real exchange rate is an endogenous variable in such a system. 1~ As a consequence, the negative output impact of an anticipated depreciation of the real exchange rate is larger, in absolute terms, than the expansionary effect of an unanticipated depreciation.
Ag6nor: Devaluation and the Real Exchange Rate
33
We now turn to econometric estimation of(16). As is clear from the above discussion, (16) is not a proper reduced form because the real exchange rate is endogenous and depends on the price of domestic goods. Accordingly, an instrumental variable estimation procedure must be used to ensure consistency. IV. Estimation Results
Estimation of a macro relation such as (16) raises several difficult statistical problems, mostly stemming from inadequacies in the data available in developing countries and their lack of comparability. Consistent series for long periods on aggregate data are available for only a few countries. Here, we follow the strategy adopted by many researchers in this area by considering a pooled sample of cross-section and time-series data for a number of developing countries. Specifically, a sample consisting of annual data covering the period 1978-87 for 23 developing countries is used (see the Appendix for more details), and a fixed-effect estimation procedure is applied to the pooled sampie. t2 Expectations are assumed "rational" or - more appropriately and perhaps less controversially - "consistent". To derive a proxy for the expected changes in the money supply, government spending, foreign output, and the real exchange rate, a two-step procedure is used, as suggested in the literature on policy ineffectiveness.13 In the first stage of the estimation procedure, a forecasting equation is generated for each variable. The information set for estimation consists of the lagged value of the variable itself, a time trend, and country-specific dummies. The fitted values from these prediction equations (which are not reported here to save space) are used as a proxy for the expected values. In the second stage, the predictions from the forecasting equations are used to estimate (16) by an appropriate method. As is well known, the resulting estimates will be consistent, although not fully efficient. Because of the endogeneity of the real exchange rate, the estimation of (16) is performed using an instrumental variable method on a fixed-effect procedure, with country-specific dummy variables included in the regression equation. Instruments used are the constant 12 For a detailed discussion of the fixed-procedure approach to estimation of pooled time series-cross sections, see Mundlak [1978]. t3 For a discussion of several two-step procedures used in the estimation of rational expectations models together with their limitations, see Pesaran [1987],
34
Weltwirtschaftliches Archiv
Table 1 - Real Output Equations in 23 Developing Countries: 1978-87 Equation
Constant Am - Am" A e - Ae" Ag - Ag"
Ay~--Ay"~ e
(1)
(2)
6.658 (3.631) 0.388 (1.610) 0.933 (2.719) 0.305 (1.443) 0.771 (1.709) --0.217 (--1.261)
3.728 (2.614) 0.176 (1.337) 0.965 (2.756) 0.201 (1.327) 0.628 (1.853) --0.008 (--1.297) -
e_ l
time
~2 ~r
-
-
0.813 0.074
0.679 (1.733) 0.815 0.071
(3)
(4)
8.530 (4.621) 0.344 (1.601) 0.516 (1.927) 0.292 (1.462) 0.819 (1.737) -
7.259 (3.113) 0.318 (1.548) 0.694 (1.915) 0.306 (1.461) 0.745 (1.802) -
--0.396 (--3.913) -
--0.367 (--3.965) 0.008 (0.336) 0.818 0.061
0.822 0.053
(5)
(6)
4.624 (3.283) 0.951 (1.259) 0.563 (1.987) 0.216 (1.413) 0.536 (1.927) --0.473 (--1.824) --0.185 (--2.330) -
9.103 (3.586) 0.954 (1.157) 0.589 (1.918) 0.165 (1.347) 0.657 (1.923) --0.427 (--2.346) --0.182 (--2.586) --0.051 (-- 1.485) 0.836 0.018
0.831 0.023
Note: Coefficients in brackets are t-ratios. ~z denotes the coefficient of determina-
tion adjusted for degrees of freedom, and ~r the estimated standard error of the regression. Country dummies are not reported.
term, country-specific time trends, an index of industrial output in industrialized countries, and the ratio of fiscal deficit to the monetary base, a measure of fiscal stance in each country. Table 1 contains a summary of the results. Consider first regressions (1) and (2), which differ by the inclusion of a time trend. Unanticipated movements in money supply and government spending have only a weak effect on output behaviour, while unexpected changes in world economic activity seems to exert a significant role - a result consistent with those reported for instance by Sheehey [1986]. Both regressions show that anticipated depreciations in the real exchange rate have a negative - although not highly significant - impact on output, while unanticipated changes have a positive effect. Regressions (3) and (4) show that the same conclusions remain when the current real exchange rate is replaced by its once-lagged value. These results provide, therefore, significant support for the theoretical model presented above.
Ag6nor: Devaluation and the Real Exchange Rate
35
The preceding results can be contrasted with those obtained by Edwards [1986, 1989 b]. The ad hoc real output equation estimated by Edwards [1986] has, as explanatory variables, the real exchange rate, the terms of trade, the ratio of nominal government expenditure to nominal income, and the unexpected rate of growth of the nominal money stock, in addition to a country-specific time trend. His equation is estimated on pooled annual data for 12 developing countries over the period 1965-80. His initial regressions show that unexpected money changes and the ratio of government spending to nominal income have a positive effect on output, while the terms of trade variable is not statistically significant. The real exchange rate has a negative sign, as suggested by the contractionary devaluation hypothesis. In a second step, however, Edwards [ibid.] also introduces the lagged value of the real exchange rate, in order to test the idea that the contractionary effect of devaluations are only a short-run phenomenon. He finds that the coefficient of the contemporaneous real exchange rate turns out to be negative, as before, but the coefficient of the once-lagged real exchange rate turns out to be positive and statistically significant and of the same order (with opposite sign) as the coefficient on the current real rate. His conclusion is that "although devaluations are contractionary in the first year, this effect is completely reversed in the second year.., devaluations are neutral in the medium to long run" (p. 506). For comparison purposes, the output equation (16) was also estimated by adding the once-lagged value of the real exchange rate. The results are shown as (5) and (6) in Table 1. The lagged value is negative and statistically significant in both equations, while the current value remains also significantly negative. Our regressions therefore do not substantiate the view that real depreciations are neutral in the "medium to long run", as defined - rather incorrectly - by Edwards. 14 This is also consistent with later work by Edwards [1989 b], where the long-run neutrality result is shown to be rejected for most developing countries in his sample. The above results illustrate the importance of a proper specification of the output equation to be estimated when testing for contrac14 It must be stressed that to account for the possibility of longer lags in adjustment, the output equation should be estimated with higher-order lagged values of the real exchange rate. This, however, was not feasible here because of the limited number of observations available.
36
Weltwirtschaftliches Archiv
tionary effects, is It is also important, however, to consider an adequate definition of the real exchange rate. F o r instance, as a proxy for the real exchange rate, Edwards uses the nominal exchange rate with respect to the US dollar times the ratio of the US wholesale price index relative to the domestic consumer price index. By contrast, I M F data on multilateral (effective) real exchange rates are used in this study. Now, the point is not that trade-weighted real exchange-rate indices calculated by the I M F are necessarily "better"; any index has inherent conceptual, interpretative and methodological limitations. It is rather that variations in the weights and deflators used in the construction of indices often yield significantly different quantitative results. 16
V. Summary, Conclusions and Extensions In this paper, the effect of real exchange-rate changes on output has been analyzed using annual data for a group of 23 countries over the period 1978-87. The empirical analysis was based on an aggregate output equation explicitly derived from a rational expectations macro-model with imported intermediate goods. The results show that an anticipated depreciation of the real exchange rate has a negative effect on economic activity, while an unanticipated depreciation has a positive impact on output. Moreover, contrary to the results produced by Edwards [1986], the contractionary effect of anticipated depreciations remains significant even after a year. In evaluating the results, however, it must be kept in mind that the methodological approach used in this paper is subject to several limitations. As Kamin [1988, p. 3] argues, time-series regression analysis of exchange-rate effects on output may not be totally appropriate for characterizing devaluation episodes. First, they do not tell us what has happened historically during a devaluation episode. Real exchange rates move more or less continuously over time; they merely show more exceptional movement during devaluations. Second, not only are devaluations typically associated with other stabilization policies, but they are large, isolated events, that occur only sporadically and t5 To a certain extent, errors in the specification of the output equation can be dealt with by using a vector autoregression approach. Kamas [1990] has recently presented preliminary evidence on this issue. 16 Variousforms of real effectiveexchange-rate indices have been used in the literature; Maciejewski [1983] provides a fairly exhaustive taxonomy of indices and their differences.
Ag6nor: Devaluation and the Real Exchange Rate
37
their influence, particularly as regards expectations, may differ qualitatively from slower, more routine exchange-rate adjustments. Subject to the above limitations, the econometric results presented in this paper seem nevertheless to provide some fairly interesting elements in the current debate on the role of devaluation in developing countries. Before firm conclusions can be reached as regards the potential contractionary effect of exchange-rate changes, however, several extensions of the present analysis should be considered. First, an even larger sample and/or estimation period should be used, in order to provide more representative results. In this paper, because of the need to consider consistently calculated time series for real effective exchange rates, it has been possible to examine only one-year (lagged) effects of real currency changes. However, this may not be the proper time-frame to examine the long-run effects of devaluations. As discussed earlier, some recent studies based on "factual" approaches have examined the long-term impact of currency changes within a threeyear period. Second, as discussed above, the real exchange rate is an endogenous variable; an important extension would be to extend the analytical model to allow for the joint determination of output and the real exchange rate [see, for instance, Edwards, 1988, 1989b]. Finally, this study has considered only one of the mechanisms that can explain a contractionary effect of devaluation; other links have been discussed in the literature, and these may be worth investigating.
Data Appendix All data are taken from the International Financial Statistics tapes, prepared by the International Monetary Fund. The countries considered - mainly on the basis of data availability - are Bolivia, Burundi, Cameroon, Chile, Colombia, Costa Rica, Cyprus, the Dominican Republic, Ecuador, Gabon, Malawi, Malaysia, Malta, Morocco, Nigeria, Paraguay, the Philippines, Sierra Leone, Togo, Uruguay, Venezuela, Zaire, and Zambia. Annual data are used and cover the period 1978-87, the longest period for which data on real exchange rates (defined below) could be collected for all selected countries. Real output is measured by GDP at 1980 prices (IFS line 99bp). The price level is measured by the consumer price index (IFS line 64). Broad money is used to measure the stock of money balances (IFS line 35L). World inflation is measured by the rate of growth of consumer prices in the industrial countries, while world economic activity is measured
Weltwirtschaftliches Archiv
38
by the index of industrial production in developed countries (IFS Worm Tables, lines 64 and 66, code 110). Real government spending is calculated as total government expenditure (IFS line 82) deflated by the consumer price index. The fiscal deficit is line 80 in IFS, while the monetary base is line 14. The real exchange-rate index used is a nominal effective exchange-rate index (i.e., an index o f the period average exchange rate of the currency considered to a weighted geometric average of exchange rates for the currencies of selected partner - or competitor - countries), adjusted for relative movements in domestic price or cost indicators of the home country and its partner or competitor - countries. This is line rec in the IFS. The index is defined so that an increase reflects an appreciation; in the calculations reported in the paper, the inverse value was used. References
Barbone, Luca, Francisco Rivera-Batiz, "Foreign Capital and the Contractionary Impact of Currency Devaluation, with an Application to Jamaica". Journal of Development Economics, Vol. 26, 1987, pp. 1-15.
Branson, William H., "Stabilization, Stagflation, and Investment Incentives: The Case of Kenya 1979-80". In: Sebastian Edwards, Liaqat Ahamed (Eds.), Economic Adjustment and Exchange Rates in Developing Countries. Chicago 1986, pp. 267-293.
Bruno, Michael, "Stabilization and Stagflation in a Semi-Industrialized Economy". In: Rudiger Dornbusch, Jacob A. Frenkel (Eds.), International Economic Policy: Theory and Evidence. Baltimore 1979, pp. 270-289. Buffie, Edward E, "Devaluation and Imported Inputs: the Large Economy Case". International Economic Review, Vol. 27, 1986, pp. 123-140.
Chopra, Ajai, Peter J. Montiel, "Output and Unanticipated Money with Imported Intermediate Goods and Foreign Exchange Rationing". 1MF Staff Papers, Vol. 33, 1986, pp. 697-721.
Cooper, Richard N., Currency Devaluation in Developing Countries. Essays in International Finance No. 86. Princeton 1971.
Denoon, David B., Devaluation under Pressure. India, Indonesia, and Ghana. Cambridge 1986.
Diaz-Alejandro, Carlos E, "A Note on the Impact of Devaluation and the Redistriburive Effect". The Journal of Political Economy, Vol. 71, 1963, pp. 577-580. -, Exchange Rate Devaluation in a Semi-Industrialized Country: The Experience of Argentina, 1955-61. Cambridge 1965. Donovan, Donal J., "Real Responses Associated with Exchange Rate Action in Selected Upper Credit Tranche Stabilization Programs". IMF Staff Papers, Vol. 28, 1981, pp. 698-727. -, "Macroeconomic Performance and Adjustment under Fund-Supported Programs: The Experience of the Seventies". IMF Staff Papers, Vol. 29, 1982, pp. 171-203.
Agrnor: Devaluation and the Real Exchange Rate
39
Dornbusch, Rudiger, Open Economy Macroeconomics. 2nd edition. New York 1988. Edwards, Sebastian, "Are Devaluations Contractionary?". Review of Economics and Statistics, Vol. 68, 1986, pp. 501-508. -, "Real and Monetary Determinants of Real Exchange Rate Behavior. Theory and Evidence from Developing Countries". Journal o f Development Economics, Vol. 29, 1988, pp. 311-341.
- [1989a], "Exchange Controls, Devaluations, and Real Exchange Rates: The Latin American Experience". Economic Development and Cultural Change, Vol. 37, 1989, pp. 457-494. - [1989b], Real Exchange Rates, Devaluation and Adjustment: Exchange Rate Policy in Developing Countries. Cambridge 1989.
Goidstein, Morris, Peter J, Montiel, "Evaluating Fund Stabilization Programs with Multicountry Data: Some Methodological Pitfalls". I M F Staff Papers, Vol. 33, 1986, pp. 304-344. Guiti~in, Manuel, "The Effects of Changes in the Exchange Rate on Output, Prices and the Balance of Payments". Journal of International Economics, Vol. 6, 1976, pp. 65- 74.
Gylfason, Thorvaldur, Credit Policy and Economic Activity in Developing Countries with IMF Stabilization Programs. Princeton Studies in International Finance No. 60. Princeton 1987. -, Michael Schmid, "Does Devaluation Cause Stagflation?". The Canadian Journal of Economics, Vol. 16, 1983, pp. 642-654. -, Ole Risager, "Does Devaluation Improve the Current Account?". European Economic Review, Vol. 25, 1984, pp. 37-64. -, Marian Radetzki, Does Devaluation Make Sense in the Least Developed Countries? lIES Seminar Paper No. 314, University of Stockholm. Stockholm 1985.
Hamilton, Clive, Contractionary Devaluation: A Review of the Literature. Australian National University, Faculty of Economics, Working Paper No. 158. Canberra 1988.
Hanson, James A., "Contractionary Devaluation, Substitution in Production and Consumption, and the Role of the Labor Market". Journal of International Economics, Vol. 14, 1983, pp. 179-189.
Hardouvelis, Gikas, "Optimal Wage Indexation and Monetary Policy in an Economy with Imported Raw Materials". Journal oflnternational Money and Finance, Vol. 6, 1987, pp. 419-432.
Himarios, Daniel, "Do Devaluations Improve the Trade Balance? The Evidence Revisited". Economic Inquiry, Vol. 27, 1989, pp. 143-168. Hirschman, Albert O., "Devaluation and the Trade Balance: A Note". Review of Economics and Statistics, Vol. 31, 1949, pp. 50-53. Johnson, Gove G., Jr., et al., Formulation of Exchange Rate Policies in Adjustment Programs. International Monetary Fund, Occasional Paper No. 36. Washington 1985.
Kamas, Linda, "Further Evidence on Contractionary Devaluation: Cointegration and Vector Autoregressions". Santa Clara University, Santa Clara, June 1990, mimeo.
40
Weltwirtschaftliches Archiv
Kamin, Steven B., Devaluation, External Balance, and Macroeconomie Performance: A Look at the Numbers. Princeton Studies in International Finance No. 62. Princeton 1988.
Khan, Mohsin S., "The Macroeconomic Effects of Fund-Supported Adjustment Programs". IFM Staff Papers, Vol. 37, 1990, pp. 195-231. -, Michael D. Knight, Fund-Supported Adjustment Programs and Economic Growth. Occasional Paper No. 41, International Monetary Fund. Washington 1985. Krugman, Paul, Lance Taylor, "Contractionary Effects of Devaluation". Journal o f International Economics, Vol. 8, 1978, pp. 445-456. Lizondo, J, Saul, Peter J. Montiel, "Contractionary Devaluation in Developing Countries: An Analytical Overview". IFM Staff Papers, Vol. 36, 1989, pp. 182-227.
Maciejewski, Edouard B., "'Real' Effective Rate Indices: A Re-examination of the Major Conceptual and Methodological Issues". IFM Staff Papers, Vol. 30, 1983, pp. 491-541.
Marston, Richard C., Stephen J. Turnovsky, "Imported Materials Prices, Wage Policy, and Macro-Economic Stabilization". The Canadian Journal of Economics, Vol. 18, 1985, pp. 273-284.
Mundlak, Yair, "On the Pooling of Time Series and Cross Section Data". Econometrica, Vol. 46, 1978, pp. 69-85.
Nunnenkamp, Peter, Rainer Schweickert, "Adjustment Policies and Economic Growth in Developing Countries - Is Devaluation Contractionary?". Weltwirtschaftliches Archiv, Vol. 126, 1990, pp. 474-493.
Pesaran, Hashem, The Limits to Rational Expectations. Oxford 1987. Roca, Santiago, Rodrigo Priale, "Devaluation, Inflationary Expectations and Stabilisation in Peru". Journal of Economic Studies, Vol. 14, 1987, pp. 5-33.
Schmid, Michael, "Stagflationary Effects of a Devaluation in a Monetary Model with Imported Intermediate Goods". Jahrbiicher 3~r National~konomie und Statistik, Vol. 197, 1982, pp. 107-129.
Shea, Koon-Lam, "Imported Inputs, Devaluation and Balance of Payments: A Keynesian Macro-Approach". Southern Economic Journal, Vol. 43, 1976, pp. I 106-1111.
Sheehey, Edmund J., "Unanticipated Inflation, Devaluation and Output in Latin America". Worm Development, Vol. 14, 1986, pp. 665-671. Solimano, Andres, "Contractionary Devaluation in the Southern Cone. The Case of Chile". Journal of Development Economics, Vol. 23, 1986, pp. 135-151. Wijnbergen, Sweder van, "Exchange Rate Management and Stabilization Policies in Developing Countries". Journal o f Development Economics, Vol. 23, 1986, pp. 227247.
Z u s a m m e n fa s s u n g : Produktion, Abwertung und realer Wechselkurs in Entwicklungsl/indern. - Der Verfasser untersucht, ob - wie neuerdings von den ,,neuen Strukturalisten" behauptet wird - Abwertungen in Entwicklungsl/indern kontraktive Wirkungen haben. Nach einem kurzen Oberblick fiber die Literatur werden Daten ffir
Ag6nor: Devaluation and the Real Exchange Rate
41
23 Entwicklungsl/inder aus den Jahren 1978 bis 1987 benutzt, um ein Modell zu sch/itzen, das die Entwicklung der realen Produktion explizit aus einem Gleichgewichtssystem yon aggregierten Nachfrage- und AngebotsgrSBen bei rationalen Erwartungen ableitet. Es ergibt sich, dab antizipierte Abwertungen des realen Wechselkurses eine negative Auswirkung auf die Produktion haben, w/ihrend Abwertungen, die nicht antizipiert werden, positiv wirken. AuBerdem zeigt sich, dab die kontraktiven Effekte noch nach einem Jahr andauern, was der Ansicht widerspricht, Abwertungen seien auf ,,mittlere" Sicht neutral.
R 6 s u m 6: Output, d6valuation et taux de change r6el dans les pays en voie de d6veloppement. - Cette 6tude 6value si les d6valuations dans les pays en voie de d6veloppement ont un effet de contraction comme les Nouveaux Structuralistes l'ont r6cemment soulign6. Apr6s avoir bri6vement examin6 la litt6rature, rauteur se sert de donn6es relatives ~ 23 pays en voie de d6veloppement pour la p6riode 1978 ~i 1987 afin d'estimer un mod61e de la production r6elle. Ce mod61e est explicitement d6riv6 d'un syst6me d'6quilibre de l'offre et de la demande globales avec anticipations rationnelles. Les r6sultats indiquent que des d6pr6ciations anticip6es du taux de change r6el ont un effet n6gatif sur l'activit6, tandis que des d6valuations non-anticip6es ont un effet positif. En outre, il est montrb que l'effet de contraction persiste plus qu'une ann6e ce qui contredit l'hypoth6se que les d6valuations sont neutres fi moyen terme.
R e s u m e n : Producci6n, devaluaci6n y tasa de cambio real en los paises en desarrollo. - En esta trabajo se evalfia si, como los "Nuevos Estructuralistas" han enfatizado recientemente, las devaluaciones en los paises en desarrollo son contractivas. Despu6s de pasar revista a la literatura existente, se utilizan datos de 23 paises en desarrollo para 1978-1987 para estimar un modelo de comportamiento de la producci6n real, explicitamente derivado de un modelo de equilibrio general de oferta y demanda con expectativas racionales. Los resultados indican que devaluaci6nes anticipadas de la tasa de cambio real tienen un impacto negativo sobre la producci6n, mientras que cambios no anticipados tienen un efecto positivo. Ademfis, se demuestra que el efecto contractivo persiste despu6s de un afio, contradiciendo asi la opini6n que las devaluaci6nes son neutrales en el mediano plazo.