اثرات اقتصاد کلان در رژیمهای نرخ ارز اسمی: بینش جدید برای پویایی نقش قیمت

This paper analyzes the effects of pegged and floating exchange rates using a two-country dynamic general equilibrium model that is calibrated to the US and a European aggregate. The model assumes shocks to money, productivity and the interest rate parity condition. It captures the fact that the sharp increase in nominal exchange rate volatility after the end of the Bretton Woods (BW) system was accompanied by a commensurate rise in real exchange rate volatility, but had no pronounced effect on the volatility of GDP. This holds irrespective of whether flexible or sticky prices are assumed—which casts doubt on the widespread view that the roughly equal (post-BW) rise in nominal and real exchange rate volatility reflects price stickiness. A flex-prices variant of the model captures better the fact that the correlation between US and European GDP has been higher in the post-BW era than under BW.

Much research has been devoted to explaining the macroeconomic effects of exchange rate regimes. After the end of the Bretton Woods (BW) pegged-exchange rate system, the volatility of nominal and real exchange rates between the major currency blocs (US, Europe, Japan) rose sharply. By contrast, the volatility of real GDP showed little change after the end of BW, but the cross-region correlation of GDP increased markedly. For example, the standard deviation of Hodrick–Prescott filtered log quarterly nominal and real exchange rates between the US and an aggregate of the three largest continental European economies (EU3: Germany, France, Italy) rose from less than 1% under BW to about 8% in the post-BW era. The standard deviation of US and EU3 GDP was between 1% and 2%, in both eras; the US–EU3 GDP correlation rose from −0.18 (BW) to 0.48 (post-BW). This paper analyzes these facts using a quantitative two-country dynamic general equilibrium (DGE) model. Interest centers on the relevance of these facts for a central and controversial issue: the role of price stickiness in (international) macroeconomic models. The simultaneous rise in nominal and real exchange rate volatility after the end of the BW system, is widely viewed as reflecting price stickiness—and used to justify (Keynesian) sticky-prices models, see; e.g., Mussa, 1986, Mussa, 1990, Dornbusch and Giovannini, 1990 and Caves et al., 1993, and Obstfeld and Rogoff (1996). The results presented in this paper cast doubt on this view. A flexible-prices variant of the model here–that features shocks to money supply, productivity and to the uncovered interest rate parity (UIP) condition–can capture the stylized facts described in the first paragraph. A sticky-prices variant accounts for the post-BW rise in nominal and real exchange rate volatility, but fails to explain the rise in the cross-country GDP correlation. Thus, the simultaneous rise in nominal and real exchange rate volatility after the BW era cannot be interpreted as evidence for price stickiness (flex- and sticky-prices variants both capture this phenomenon). The widespread view described above seems to be based on the assumption that money shocks are the main source of real exchange rate fluctuations—standard theory predicts that money shocks have no effect on the real exchange rate under price flexibility, but induce real exchange rate movements that closely track the nominal exchange rate when prices are (sufficiently) sticky. However, econometric attempts to predict post-BW short-run exchange rate movements from changes in money and other macroeconomic fundamentals (productivity, fiscal policy) have failed (Rogoff (2000)). Also, structural models driven only by these fundamentals generate insufficient exchange rate volatility. This applies both to flex- and to sticky-prices models.1 In order to generate more realistic exchange rate volatility, this paper allows for UIP shocks; these shocks can be interpreted as reflecting transitory biases in households' exchange rate forecasts. Variants of the model with a pegged and with a floating exchange rate are calibrated to the US and the EU3. I document that UIP shocks were much larger in the post-BW era than under BW. Estimates of the time series process of UIP shocks in the BW era [post-BW era] are used to calibrate the pegged-rate [floating-rate] variant of the model. A flex-prices version of the model and a sticky-prices version are compared. The latter assumes Calvo (1983) staggered price setting; the mean lag between price changes is set at 4 quarters, as often assumed in Keynesian models. Simulations of the floating-rate variant suggest that UIP shocks are a powerful source of nominal and real exchange rate fluctuations—much more than money and productivity shocks. Predicted real exchange rate volatility is markedly higher under the float than under the peg. The floating-rate variant (with post-BW UIP shocks) captures about 80% of the standard deviations of post-BW nominal and real US–EU3 exchange rates. In the model, nominal exchange rate movements induced by UIP shocks have only a limited effect on national price levels, due to the small volume of US–EU3 trade (about 1% of GDP); thus, these movements are accompanied by (roughly) equiproportional variations of the real exchange rate; also, these exchange rate movements only have a weak effect on GDP. The model thus captures the fact that the sharp rise in exchange rate volatility after the end of BW did not greatly affect the volatility of US and EU3 GDP. These results hold irrespective of whether sticky or flexible prices are assumed. By contrast, flex- and sticky-prices model versions yield sharply differing predictions regarding the effect of the exchange rate regime on the cross-country GDP correlation. Monetary policy affects output under sticky prices, but is neutral under flexible prices. As a peg requires international synchronization of monetary policy, the sticky-prices version predicts that the cross-country GDP correlation is higher under a peg than under a float. That prediction is inconsistent with the finding that the US–EU3 GDP correlation was lower under BW. Flex-prices variants of the model, by contrast, capture that finding, provided the calibration takes into account the fact that US and EU3 productivity innovations were negatively correlated under BW, but positively correlated after BW. The work here is related to the Keynesian literature of the 1960s to 1980s that analyzed exchange rate pegs and floats (e.g., Mundell, 1968); this literature predicted that the exchange rate regime affects real variables, but provided only limited quantitative results. It also lacked the micro-foundations that characterize modern DGE macro models. The recent open economy DGE literature typically assumes a floating exchange rate. Exceptions include Bacchetta and van Wincoop, 2000 and Obstfeld and Rogoff, 2000 and Devereux and Engel (2003) who compare the welfare effects of pegs and floats, using highly stylized sticky-prices models (with closed form solutions) that generate insufficient exchange rate volatility. In contrast, the paper here presents a positive analysis of the BW/post-BW regimes, based on a quantitative business cycle model with realistic exchange rate volatility.2