In Good and in Bad Times? The Relation between Anomaly Returns and Market States

Abstract

We evaluate the relation between the size of 138 return anomalies and market states using a sample of 56 countries from 1981 to 2019. We find that the vast majority of anomalies (51 of 138 statistically significant at the 5% level) perform better if the country’s stock market index trades below its 200-day moving average, our definition of a bad market state; 10 anomalies perform significantly better in good market states. On average, the value-weighted four-factor alpha amounts to 46.7 (31.2) bps per anomaly-month in bad (good) times. In relative terms, abnormal anomaly returns are 49.8% higher in bad times. Our findings are consistent across regions and different anomaly classifications. They are robust to alternative market state classifications and additional controls for investor sentiment. The evidence suggests that risk or data-mining cannot entirely explain anomaly returns.