Estimating Exchange Rate Exposures: Some "Weighty" Issues

From a sample of 910 U.S. firms over the period 1977 1996, we find that structure of the empirical model has significant impacts on resulting estimates of exchange rate exposures from equity returns. While lengthening the return horizon has minimal impact on exposure estimates, the inclusion of a market portfolio in the specification results in significant changes to the exposure estimates. We further demonstrate that different definitions of the market portfolio result in important differences in the overall distribution of exposure estimates and the interpretations of the sign, size, and significance of many firms' exposures. The source of the exposure differences across market portfolios is due to a strong size-exposure relation for U.S. firms. Surprisingly, this size-exposure relation does not appear to be driven by an underlying correlation between size and foreign cash flow position of the firms. An alternative model specification using matched CRSP capital-based size portfolios as controls for market movements in the exposure model produces firm-level exposures with a stronger relation to foreign cash flows and less of a correlation with firm size.