The Halloween Effect and Nordic Equity Markets: Fact or Fiction?

Abstract

The Halloween effect as described by Bouman & Jacobsen (2002) means that stock returns of the year are significantly higher during the November to April (winter) period than during the May to October (summer) period. In the context of efficient market theory the effect should not exist.

This study researches the existence of the Halloween effect seasonal stock market anomaly in Nordic stock markets (Denmark, Finland, Iceland, Norway and Sweden) and the pan regional OMX40 index by running OLS regression on stock returns data. The purpose of the study is to see if the Halloween effect first formally investigated by Bouman & Jacobsen (2002) holds in the five Nordic stock markets both at country index levels and in factor portfolios.

The main contribution of the study shows how the Halloween effect manifests itself in Fama & French factor portfolios using OLS regression on their portfolio returns. These portfolios are constructed and divided into value and growth portfolios based on other well documented anomalies: book to market, earnings to price, cash flow to price and dividend to price ratios. Therefore, this research paper looks at a stock market anomaly inside other anomalies.

The study finds a statistically significant Halloween effect in 5/5 Nordic stock markets but does not find evidence for the effect in the OMX40 pan regional index. The results for the Fama & French factor portfolios are found in general slightly more significant than those of the individual stock market indices. A statistically significant Halloween effect is found in 3/4 countries for all value portfolios. For the majority of growth portfolios a statistically significant Halloween effect is also found in these three countries.