An examination of the Halloween effect in the Finnish stock market
Väänänen, Olli (2013)
Väänänen, Olli
2013
Kuvaus
Opinnäytetyö kokotekstinä PDF-muodossa.
Tiivistelmä
This study focuses on a seasonal anomaly called the Halloween effect. The anomaly is based on empirical observations that for some reason, stock returns tend to be significantly higher during winter than summer. The summer period lasts from May through October, and the name of the anomaly comes from the well-known celebration which takes place on the last day of October.
This paper adds to international evidence by examining the effect in the Finnish stock market. The studied time period is from May 1991 through April 2013 and the effect is tested with monthly returns of both the market index (OMXH CAP growth index) and sector indices. Main testing method is regression analysis with seasonal dummy variables. In addition to Ordinary Least Squares, also GARCH(1,1) modeling and the Huber M-estimator are applied in order to find out if the Halloween effect is robust to alternative regression methods.
The results indicate that the Halloween effect exists in Finland, and that it is both statistically and economically highly significant. Depending on the method used, monthly mean winter return outperforms summer mean return by 2,2 – 2,7 %. The effect seems to appear stronger during bear markets than bull markets, but the difference is not statistically significant. Examination of sector returns reveals that good winter performance is achieved by industrials, health care, consumer goods and financials. By exploiting the Halloween effect and differences between sector returns, it was possible to create investment strategies that were able to outperform the buy-and-hold strategy with a clear marginal, even after transaction costs and adjusting for risk.
The study considers several possible causes for the anomaly, including the January effect, outliers in the data, interest rate variation and methodological issues. None of them seem to be the source of the Halloween effect and the reason for this seasonal stock return pattern remains unknown.
This paper adds to international evidence by examining the effect in the Finnish stock market. The studied time period is from May 1991 through April 2013 and the effect is tested with monthly returns of both the market index (OMXH CAP growth index) and sector indices. Main testing method is regression analysis with seasonal dummy variables. In addition to Ordinary Least Squares, also GARCH(1,1) modeling and the Huber M-estimator are applied in order to find out if the Halloween effect is robust to alternative regression methods.
The results indicate that the Halloween effect exists in Finland, and that it is both statistically and economically highly significant. Depending on the method used, monthly mean winter return outperforms summer mean return by 2,2 – 2,7 %. The effect seems to appear stronger during bear markets than bull markets, but the difference is not statistically significant. Examination of sector returns reveals that good winter performance is achieved by industrials, health care, consumer goods and financials. By exploiting the Halloween effect and differences between sector returns, it was possible to create investment strategies that were able to outperform the buy-and-hold strategy with a clear marginal, even after transaction costs and adjusting for risk.
The study considers several possible causes for the anomaly, including the January effect, outliers in the data, interest rate variation and methodological issues. None of them seem to be the source of the Halloween effect and the reason for this seasonal stock return pattern remains unknown.