Item set: Wexcombe Capital: explaining P/E ratios
Noah Feldberg is an equity analyst at Wexcombe Capital. To screen for mispriced stocks, he regresses the trailing price-to-earnings ratio of 45 consumer staples companies on three independent variables: the dividend payout ratio (as a decimal), the consensus expected EPS growth rate over the next five years (in percent) and the stock's beta. He uses ordinary least squares on one year's cross-section.
Exhibit 1. Regression of P/E on company characteristics
Intercept: coefficient 6.40; standard error 2.10; t-statistic 3.05
Dividend payout ratio: coefficient 8.20; standard error 2.90; t-statistic 2.83
Expected EPS growth (%): coefficient 0.95; standard error 0.21; t-statistic 4.52
Beta: coefficient -2.30; standard error 1.40; t-statistic -1.64
Exhibit 2. ANOVA table
Regression: df 3; sum of squares 352.0; mean square 117.33
Residual: df 41; sum of squares 287.0; mean square 7.0000
Total: df 44; sum of squares 639.0
F-statistic: 16.76
Exhibit 3. Critical values, 41 degrees of freedom where relevant
F-distribution, 3 and 41 degrees of freedom, 5% significance: 2.83
t-distribution, 5% significance, two-tailed: 2.020
t-distribution, 5% significance, one-tailed: 1.683
Feldberg's supervisor, Ines Moreau, CFA, asks him to test whether the model as a whole has explanatory power at the 5% level and to report a 95% confidence interval for the growth coefficient.
Moreau then points out that the model leaves out financial leverage. Within the sample, more highly leveraged companies tend to have higher expected EPS growth, and leverage is believed to affect P/E ratios directly.
Finally, Feldberg plots the regression residuals against expected EPS growth. The residuals are mostly positive for companies with very low and very high growth and mostly negative for companies with moderate growth, forming a U shape. Their spread looks similar across the range of growth rates.