A Comparison of FX Exposure Estimates with Different Control Variables

We compare the FX exposure estimates of four empirical models that differ only in the choice of control variable. We use a large sample of U.S. equities (19,100) over a long time span (1980-2011). We find a much higher percentage of statistically significant FX exposure estimates with a bond return control variable than with a broad equity index. We also find that the FX exposure estimates with no control variable are close to those for the bond return control variable, and the estimates with Fama-French factor control variables are close to those with the equity index