Asset pricing puzzles: Evidence from options markets

Abstract

This paper proposes and implements a consumption-based pricing kernel (stochastic discount factor) testing methodology that focuses on the covariance between the pricing kernel and asset squared excess returns. This covariance has an intuitive economic interpretation as a risk-neutral variance risk-premium, i.e. the difference between the risk-neutral return variance and the objective return variance. In the same way that an asset riskpremium puzzle is due to a failure of the pricing kernel to adequately covary with asset excess returns, a riskneutral variance puzzle is due to a failure of the pricing kernel to adequately covary with asset squared excess returns. This paper tests a consumption-based pricing kernel specification that is compatible with habit formation, consumption durability, and constant relative risk-aversion over a range of plausible preference parameter values. The difference between consumption-based and semi-parametric option-based estimates of unconditional risk-neutral S&P500 return variance is used as a pricing kernel specification test statistic. Evidence is found of a risk-neutral S&P500 return variance puzzle if constant relative risk-aversion is assumed. The puzzle is resolved when the pricing kernel is allowed to exhibit habit formation. The acceptable habit pricing kernels exhibit higher habit levels, higher utility function concavity, and lower rates of timepreference than estimates in related papers. When the full history of consumption data is used, the preference parameter estimates are more similar to those of related papers.