The cross-section of January effect

We examine the cross-sectional January effect among portfolios that long sentiment-prone and difficult-to-arbitrage stocks and short sentiment-insensitive and easy-to-arbitrage stocks. These long-short portfolios on average earn over 20 times higher returns in January than in a non-January month. 85% of the cross-sectional January effect comes from its long legs, consistent with a sentiment-driven mispricing explanation. The cross-sectional January effect persists over time and remains significant after accounting for common risk factors and time-varying factor loadings.