Are OLS Estimates of the Return to Schooling Biased Downward? Another Look