مواجهه با نرخ ارز خطی و غیر خطی

This paper presents a new methodological approach to examine exchange rate exposure which takes account of the role of the market portfolio and macroeconomic variables in exposure regressions, exchange rate regimes based on periods of depreciation and appreciation, and nonlinear exposure. Within each regime we show that the stock market's own exposure to exchange rates should be taken into account before considering industry exposure. In addition, we adjust the exchange rate and the stock market for common economy-wide factors that are unrelated to exchange rates. Within this framework we show that exposure to bilateral exchange rates is statistically and economically important and that industries with extensive international trade are more often exposed than industries with low levels of international trade. The signs of exposure coefficients in each regime are consistent with the extent to which an industry exports. We also show that nonlinear exposure is often statistically and economically significant. Interestingly, there is little evidence that industries are exposed to a currency basket.

Empirical evidence indicates that there are deviations from purchasing power parity that national economies are becoming increasingly integrated, product and financial markets are becoming globalized, and corporate profits are affected by currency movements. In the light of this, it is puzzling that the economic and statistical relationship between exchange rate changes and US stock returns is marginal at best. The majority of empirical studies test for a constant linear relationship between stock returns and exchange rate changes.1 In contrast, the theoretical literature on the relationship between the value of a firm and the exchange rate generally posits a nonlinear relationship.2 Furthermore, there are a number of theoretical models that illustrate how firm behavior will be different when the currency is depreciating relative to it appreciating.3 Therefore, measuring exposure could be further complicated by the fact that exposure may depend on the exchange rate regime. These factors mean that the traditional method of measuring linear exposure may be inappropriate. Consider Fig. 1 which illustrates both linear and nonlinear exposure for an exporter. The long straight line depicts the case where exposure is estimated as one linear relationship for the whole period and the hyperbola does the same for nonlinear exposure. We also plot two straight lines joined at zero but with different slopes where linear exposure is shown separately for appreciation and depreciation periods.Empirical evidence indicates that there are deviations from purchasing power parity that national economies are becoming increasingly integrated, product and financial markets are becoming globalized, and corporate profits are affected by currency movements. In the light of this, it is puzzling that the economic and statistical relationship between exchange rate changes and US stock returns is marginal at best. The majority of empirical studies test for a constant linear relationship between stock returns and exchange rate changes.1 In contrast, the theoretical literature on the relationship between the value of a firm and the exchange rate generally posits a nonlinear relationship.2 Furthermore, there are a number of theoretical models that illustrate how firm behavior will be different when the currency is depreciating relative to it appreciating.3 Therefore, measuring exposure could be further complicated by the fact that exposure may depend on the exchange rate regime. These factors mean that the traditional method of measuring linear exposure may be inappropriate. Consider Fig. 1 which illustrates both linear and nonlinear exposure for an exporter. The long straight line depicts the case where exposure is estimated as one linear relationship for the whole period and the hyperbola does the same for nonlinear exposure. We also plot two straight lines joined at zero but with different slopes where linear exposure is shown separately for appreciation and depreciation periods.

This paper has uncovered evidence that US industries are exposed to changes in exchange rates. Previous empirical evidence has concluded that such exposures are negligible. We uncover these new findings by considering the effect of the dollar's regimes on exposure estimates. In addition, we show that exchange rate exposure must be estimated relative to individual currencies, not a currency basket. Simultaneously to the above two points, the overall currency exposures of the stock market and common economy-wide macroeconomic factors that affect the stock market and currencies should be accounted for in the empirical estimates of industry exposures. The results show that, in general, industries with extensive international trade have greater incidence of statistically significant exposure and that this exposure is more important economically. Exposure is regime specific and is not uncovered when considering the whole sample period. Similar to the extant literature, we find that exposure coefficients change sign over different regimes which is consistent with changes in terms of trade, tariffs and quotas, imported costs, international competition, hedging policies, and firm behavior. We are able to show that as firms export more they benefit from a dollar depreciation and suffer from a dollar appreciation, exactly as predicted in theory. Thus, although the sign of the exposure coefficient could be negative or positive, an increase in exports always increases stock returns when the dollar depreciates and decreases stock returns when the dollar appreciates. Nonlinear exposure effects are also found to be statistically and economically important. They improve the explanation of returns over and above that of the linear exposure coefficients. Nonlinear exposure appears to be just as important in industries with low levels of international trade as it is in industries extensive international trade. This finding is perhaps not too surprising since the nonlinear term captures the effects of large changes in exchange rates. Industries that previously have had only small amounts of international trade might find it profitable to increase imports and, or, exports as changes in the exchange rate make international trade profitable relative to the costs of undertaking it. Using non-orthogonalized regressions fails to uncover exchange rate exposure, either in the regimes or over the whole sample. We investigate the robustness of our results and show that the exposure estimates are not materially affected by a large list of additional explanatory variables. However, we do show that using an orthogonalized currency basket instead of orthogonalized individual currencies drastically affects the results. With a currency basket we find very little evidence that any industry, irrespective of the extent of international trade and the regime considered, is exposed to the exchange rate.