Estimating Exchange Rate Exposures: Issues in Model Structure

We show that both return measurement horizon and model specification have noticeable impacts on estimates of exposure from equity prices for U.S. firms. While increases in the return horizon leads to increases in the precision of the estimates, this effect is less significant than the impact of model structure. We demonstrate that the inclusion and of a market return variable and its particular construction has a dramatic influence on the sign and size of the exposures due to a strong relation between firm size and exposure for U.S. firms. We propose using CRSP cap-based portfolios as the control for market factors and show that this produces exposures with stronger relation to foreign cash flows and correlations with firm size.