Our last two alternative tests examine different estimation methods. First, we reestimate model 2 by using weighted least squares. In this case, the weighting factor
is the inverse of the standard error of the exposure coefficients estimated by model
1. Weighted least squares assigns more weight to the more precise estimates and
therefore can increase the accuracy of our second-stage regression. The results,
presented in Regression 4, remain unchanged. Again, consistent with our hypothesis,
the use of currency derivatives significantly reduces a firm’s exchange-rate exposure.