Consistent estimation of the risk-return tradeoff in the presence of measurement error

Prominent asset pricing models imply a linear, time-invariant relation between the equity premium and its conditional variance. We propose an approach to estimating this relation that overcomes some of the limitations of the existing literature. First, we do not require any functional form assumptions about the conditional moments. Second, the GMM approach is used to overcome the endogeneity problem inherent in the regression. Third, we correct for the measurement error arising because of using a proxy for the latent variance. The empirical findings reveal significant time-variation in the relation that coincide with structural break dates in the market-wide price-dividend ratio.