“…In addition, the HJ distance is about the same when = 1, equal to 0.448 compared to 0.451 when is unrestricted (the HJ distance is slightly smaller when = 1 because, when is fixed, one fewer parameter is estimated, reducing the AIC penalty). Thus, the results using W=G −1 are largely supportive of the conjecture of Malloy, Moskowitz, and Vissing-Jorgensen (2009). We note, however, that if the model with = 1 is misspecified, parameter estimates can be sensitive to the objective function minimized, as we find here.…”