Portfolio Selection Based Upon P/E Ratios: Diversification, Risk Decomposition and Implications

A study was performed to extend the examination of extra-market covariation to groups of stocks selected in accordance with the popular criteria of price-earnings (P-E) ratio level. A sample of 2,208 stocks with positive P-E ratios of up to 50 was segmented into deciles. The time period studied was from January 1971 through December 1980. From each decile, 80 randomly chosen portfolios of 2-40 securities were created. Their variances were calculated, then segmented into: 1. total systematic risk, 2. total unsystematic risk, 3. the ordinary least squares estimated group effect, 4. the pure unsystematic risk, 5. the pure group effect, and 6. the beta nonstationarity effect. Findings indicated that both low and high P-E ratio portfolios contained abnormally high levels of unsystematic risk without regard to portfolio size and that this risk was the result of both a beta nonstationarity effect and a pure group effect.