The Cross-Section of Stock Returns before CRSP

Abstract

This study examines the cross-section of stock returns out-of-sample using a novel and independent database of U.S. stocks between 1866 and 1926. Over this ‘pre-CRSP’ era (i) the relationship between market beta and returns is flat, (ii) value, momentum, low-risk, and seasonality factor premia are sizable and significant, and (iii) size, short-term reversal, and long-term reversal premia are weak. On average, the studied equity factor premia do not materially decay out-of-sample when unaffected by post-publication arbitrage. Additionally, we provide novel insights into economic explanations of factor premia over the combined preCRSP and CRSP ‘super’ sample covering 159 years of cross-sectional stock data.

Keywords: empirical asset pricing, return anomalies, factor premia, p-hacking, momentum, value, beta, low-risk, size, reversal, machine learning. JEL classification: G10, G11, N21, N22, quant investing

JEL Classification: G10, G11, N21, N22