This paper reconciles the relative pricing controversy between DeBondt and Thaler (1985, 1987), Chan (1988), and Ball and Kothari (1989). The negative autocorrelation in long-horizon index returns, along with the selection criterion of the contrarian strategy, can explain the positive covariance between time-varying betas and risk premiums. However, test-period beta estimates reflect the reversal of earnings expectations associated with underlying factors. The controversy thus reduces to the debate of Fama and French (1988) and Poterba and Summers (1988) over the source of the temporary price components in the market index. Rational changes in expected returns and cash flows explain most of the cross-sectional variation in returns