On the Explanatory Power of Dummy Variable Regressions