A Bayesian Analysis of Lunar Effects on Stock Returns

Abstract

Biological, psychological and medical evidence widely suggests that the lunar phases may affect human behavior and mood. This suggestion motivates this study of the relationship between lunar phases and stock returns. Relevant papers indicate that lunar cycles effects do have an effect on stock returns. They indicate that the mean daily stock returns are lower near the full moon and higher near the new moon days. This paper further investigates the association between the lunar phases and daily stock returns by using a two-regime autoregressive model with a GARCH(1,1) innovation. Rather than only examining the average daily returns, the discussion will be extended in three directions: the average daily returns, the correlation between consecutive daily returns, and the GARCH volatility. The Bayesian approach will be applied to the daily stock returns of 12 countries, including the G-7 markets and five emerging markets in Asia. In general, the statistical results indicate the existence of lunar effects on daily stock returns, although different patterns are shown by the G-7 markets and some of the discussed Asian markets. In particular, the autocorrelation for consecutive daily returns is significantly different, according to both the lunar phases and the diverse structures of the various stock markets. Furthermore, for some of the G-7 markets, the volatility of the stock returns does change according to different lunar phases; higher volatility in the full moon period. In summary, the evidence is consistent, and supports the popular belief that lunar phases do affect human financial behavior.