Abstract

We revisit the role of liquidity risk. We successfully replicate Pastor and Stambaugh’s (2003) gamma liquidity risk index, and within their time period, concur with their risk premium estimate. An out-of-their-time-period analysis finds post-time-period returns that are higher and pre-time-period returns that are lower than in-time-period returns. Modest variations to the index that are intended to improve power—such as value weighting, including zero volume days, including all stock price levels, and a modification intended to reduce estimation error—all cast doubt on whether the gamma premium is compensation for liquidity risk. We create five alternative liquidity risk indices from various popular liquidity proxies. Using time-series that start in either 1932 or 1968, none of the 10 specifications produce statistically significant risk premia.