BE/ME and E/P work better than ME/BE or P/E in regressions

► Ratios produce outliers when the denominator is close to zero. ► Because book equity and earnings can be near zero, they are poor choices for a denominator in a ratio. ► Regressions using market equity/book equity or price/earnings are less powerful than tests using their reciprocals. ► Trimming market equity/book equity or price/earnings improves power, but they remain inferior to their reciprocals.