شرایط تجارت و رژیم نرخ ارز در کشورهای در حال توسعه

Since Friedman [Essays in Positive Economics, University of Chicago Press, Chicago (1953) 157–203] an advantage often attributed to flexible exchange rate regimes over fixed regimes is their ability to insulate more effectively the economy against real shocks. I use a post-Bretton Woods sample (1973–96) of 75 developing countries to assess whether the responses of real GDP, real exchange rates, and prices to terms-of-trade shocks differ systematically across exchange rate regimes. I find that responses are significantly different across regimes in a way that supports Friedman’s hypothesis. The paper also examines the importance of terms-of-trade shocks in explaining the overall variance of output and prices in developing countries.

In the early 1950s, Milton Friedman made his case in favor of flexible exchange rate regimes, based on the fact that, in a world with sticky prices, the nominal exchange rate could be used to insulate the economy against real shocks. Since then, a number of theories have confirmed his original intuition and it has become one of the least disputed arguments in favor of flexible exchange rate regimes.1 An empirical implication of this set of theories is that the short run response to real shocks should differ across exchange rate regimes. In particular, regimes that allow for a larger movement in relative prices should have smoother adjustment of output to real shocks. The aim of this paper is to test and quantify Friedman’s hypothesis. The reason why the exchange rate regime may matter is the presence of some kind of price stickiness. Friedman argued that when economies are hit by real shocks the countries that can change relative prices more quickly have smoother adjustment in terms of quantities. In particular, he noticed that in a world with sticky prices the speed at which relative prices adjust depends crucially on the exchange rate regime. Under a flexible regime, relative prices can adjust immediately through changes in the nominal exchange rate, while under fixed regimes the changes happen at the rate permitted by the nominal stickiness, which is usually much slower. Therefore, flexible regimes should have smoother quantity responses and quicker relative price adjustments to real shocks than do fixed regimes.2 Once the nominal price stickiness is relaxed, differences across regimes should vanish. Given the prominent role played by exchange rate regimes in developing countries, it is perhaps surprising that there is scant empirical work addressing the validity of Friedman’s hypothesis. Most of the empirical literature on exchange rate regimes presents no direct test of the hypothesis addressed in this paper because it makes no distinction between nominal and real shocks. For instance, Baxter and Stockman (1989), Flood and Rose (1995), and Ghosh et al. (1997) examine output and real exchange rate volatility across exchange rate regimes but do not distinguish between the contribution of real and nominal shocks in the volatility of those variables.3 Aside from the greater variability of real exchange rates in countries with flexible regimes, they find little evidence of systematic differences in the behavior of other macroeconomic variables across regimes. By contrast, Bayoumi and Eichengreen (1994) find that output and inflation in G-7 countries have responded differently to aggregate demand shocks under the Bretton Woods system and the regime of flexible rates that has prevailed subsequently. The focus of their paper, however, is different from the mechanism underlying Friedman’s theory. In this paper, in order to focus on Friedman’s hypothesis, the analysis is restricted to a single real shock given by the terms of trade of a country (the ratio between export prices and import prices in the same currency). Evidence is presented suggesting that the terms-of-trade series can be treated as exogenous for the sample of developing countries examined. The exogeneity of the terms of trade helps to identify the response of real GDP, real exchange rate, and consumer prices to terms-of-trade changes across different regimes, eliminating the need for complex identification strategies and interpretations of estimated residuals. The sample used includes data for seventy-five developing countries from 1973 to 1996. The findings of this paper provide ample empirical support for Friedman’s hypothesis. The following results are obtained: (a) the short-run real GDP response to terms-of-trade changes is significantly smaller in countries with flexible exchange rate regimes (floats) than in those with fixed regimes (pegs). Differences are mostly driven by the response to negative shocks. After 2 years, a 10% fall in the terms of trade reduced real GDP by 1.9% in pegs and 0.2% in floats; (b) after a negative shock, the real exchange rate is slow to depreciate in pegs, while it depreciates immediately and significantly in floats. Two years after a 10% negative shock, the real exchange rate has only depreciated by 1.3% in pegs and by 5.1% in floats. The response of the real exchange rate to positive shocks is not significantly different across regimes. In particular, there is no sharp appreciation in floats after positive shocks; (c) in pegs, the delayed and small real depreciation comes from a fall in domestic prices. Overall, the evidence is consistent with the view that countries with flexible regimes are able to buffer real shocks better than those with fixed regimes. The empirical methodology is also used to estimate the importance of terms-of-trade shocks in explaining the overall variance of real GDP, the real exchange rate and the price level in developing countries. The paper finds that 30% of the real GDP fluctuations in developing countries with a fixed regime can be explained by terms-of-trade disturbances. In floats, by contrast, the contribution of terms of trade to the variance of real GDP is approximately 10%. These contributions are substantially smaller from those found by Kose (2002) and Mendoza (1995) which are based in calibrations of real business-cycle models. In the case of the real exchange rate volatility, terms-of-trade shocks explain 13% in pegs and 31% in floats. That is, despite the well-documented larger real exchange rate volatility that exists in floats relative to pegs, terms-of-trade disturbances can explain a larger share of this volatility in floats than in pegs. The importance of terms-of-trade shocks in explaining the variance of real GDP, real exchange rates and prices varies considerably across time periods but less so across regions. The sensitivity analysis performed in the paper highlights several important findings. The pattern of dynamic responses to terms-of-trade shocks described above is present in all decades and regions, though it is most significant in the 1980s and in Africa. The qualitative findings are not altered by using a purely de-facto exchange rate regime classification. The response to large and small terms-of-trade shocks follows the same pattern, though the responses to large shocks are statistically more significant. Highly dollarized countries with flexible regimes have similar short-run responses to those of the average float. Finally, the difference in level of financial development that exists across regimes is not the driving force of the results. The paper proceeds as follows. Section 2 describes the classification of exchange rate regimes and the data used, and examines the exogeneity assumption of terms of trade. Section 3 introduces the empirical specification used and the dynamic response functions generated. Section 4 reports a series of robustness checks that include whether the response to a shock varies with the magnitude of the shock, with different samples, or over the various periods. Section 5 presents conclusions.

The fixed versus flexible debate remains highly contentious. In the search for clearer answers, the theoretical arguments involved ought to be tested and quantified. The main contribution of this paper is to isolate one specific type of real shock, the terms of trade of a country, and examine how countries with different regimes cope with this shock. Thus, the paper is able to scrutinize Friedman’s hypothesis and identify significant differences in the responses to terms-of-trade shocks across exchange rate regimes. The paper presents substantial evidence in favor of Friedman’s predictions that short-run real GDP responses to real shocks are significantly smoother in floats than in pegs. Furthermore, the real exchange rate response is consistent with the assertion that the exchange rate regime plays a key role in explaining the different observed real GDP responses. In response to a fall in the terms of trade, the small and slow real depreciation observed in pegs is due to the fall in domestic prices while the large and immediate real depreciation in floats reflects a large nominal depreciation. This pattern is present in all of the time periods and regions examined, though it is most pronounced in the 1980s and in Africa. The paper also finds that responses are symmetric to shocks of different signs in pegs and assymetric in floats. After a negative shock, countries with flexible exchange rate regimes have larger real exchange rate changes and smaller real GDP changes than after positive shocks. Terms-of-trade disturbances are found to explain 30% of real GDP fluctuations in fixed regimes and approximately 30% of real exchange rate fluctuations in countries with flexible regimes.