Historically, stocks tend to outperform in November, stall in December and January, and then resume higher in February and March. Stock markets have seasonal effects and calendar anomalies, which challenge the Efficient Market Hypothesis (EMH).
The Halloween effect conveys investors’ belief that the average equity returns of November through April are significantly stronger than the remaining months of the year. In response to such belief, stock traders sell at the start of May, holding the proceeds in the form of risk-free assets, and then buy again in November, i.e., around Halloween.
Causes of the Halloween Effect
Current research on the Halloween effect is not sufficient to uncover the true major causes of this stock market anomaly. Although academics have suggested a couple of possible explanations:
- The “agricultural hypothesis” implies that farmers typically take on credit during late spring and early summer to buy sowing seed. Their higher demand for credit then leads to an increase in interest rates, and a decrease in market liquidity. Consequently, these two factors combined will then drive the market down. However, Bouman and Jacobsen (2002) rejects such assumption by providing evidence that the Halloween effect cannot be explained by either changes in interest rates or trading volume.
- An alternative explanation for this anomaly could be that investors feel financially constrained after their vacation. In this case, they may demand a higher liquidity premium during the winter months.
The Studies of 2002 through 2013
Sven Bouman and Ben Jacobsen’s “The Halloween Indicator, ‘Sell in May and Go Away’: Another Puzzle” pioneered the study on the existence of the Halloween effect. They use stock market time series of the developed markets that begin in 1964 and several emerging markets that start at the end of 1988. Their data set for the study contains 344 monthly returns. The approach is a basic regression and it incorporates a seasonal dummy variable. Bouman and Jacobsen (2002) finds that there is indeed a substantial difference between the returns of November through April and the reminder months, and the returns of this interval are significantly higher. The Halloween effect is present in 36 out of the 37 countries in their sample.
Moreover, the study concludes that there is a positive and significant relation between its three proxies for the length and timing of summer vacations, and the impact of vacation on trading activities. However, there is no evidence that shows this calendar anomaly can be explained by factors such as risk, interest rates, cross correlation across markets, or the January effect, whether in developed markets or emerging markets.
If investors take a closer look at Bouman and Jacobsen’s calculation process, it is not hard to find that there are at least two outliers in their documentation for the time series of U.S. equity. The first one is the October of 1987, when the famous “Black Monday” happened on October 19, 1987. The crash began in Hong Kong and spread west to Europe, and ultimately hit the U.S. The Dow Jones Industrial Average (DJIA) dropped by 508 points to 1738.74, or 22.61%. The second outlier is the August of 1998, when the Greenwich-based hedge fund management firm Long-Term Capital Management L.P. experienced a $4.6 billion loss accentuated through the Russian financial crisis.
To address the “two outliers” issue, Edwin D. Maberly and Raylene M. re-examine Bouman and Jacobsen’s results and extend the analysis with the S&P 5oo futures. The new set of data covers the period between April 1982 and April 2003. To control for the outliers, they modify the original regression by inserting a second dummy variable. Interestingly, their evidence argues against the existence of the Halloween effect. However, it is important to note that their analyses fail to consider the other 35 foreign markets, and therefore, their conclusion cannot persuasively reject the validity of the Halloween effect.
In April 2012, “The Halloween Effect during Quiet and Turbulent Times” by Ramona Dumitriu and Razvan Stefanescu investigates the Halloween effect with the data of 28 countries containing monthly series of January 2000 through December 2011. Specifically, they use the daily closing values of the stock market indices for two separate periods: 1) From January 2000 to December 2006, or the “quiet evolution;” 2) from January 2007 to December 2011, or the “turbulent evolution.”
- For the period of 2000 through 2006, Dumitriu and Stefanescu (2012) identifies the presence of the Halloween effect on nine stock markets and a reversal of this effect for one market.
- For the period of 2007 through 2011, they find a single Halloween effect, with negative returns, for the stock market from Greece, which was heavily influenced by the crisis during the time.
The study suggests that geographical position has a major impact on the intensity of the Halloween effect – there are some differences between emerging markets and developed markets. In addition, the results reveal that there are major changes between the two periods of time, for most stock markets.
Later in the October of 2012, Ben Jacobsen and Cherry Yi Zhang extend this effort to include 108 stock markets using 55,425 monthly observations. Precisely, the results show that the returns of November through April are on average 4.52% higher than those of the remainder of the year, furthermore, over the past 50 years, the average difference between the returns of the two periods are as high as 6.25%. According to this finding, the Halloween effect could be a successful trading strategy in beating the market.
If investors are interested, they can try combining regression models with a bootstrap-based simulation setup, with an aim to test whether stock markets are indeed so inefficient. In line with the EMH, should investors still believe the existence of the Halloween effect? Entering November, investors can again watch the stock markets closely to test the validity of such trading strategy.
- “Stock Market Efficiency Withstands another Challenge: Solving the ‘Sell in May/Buy after Halloween’ Puzzle” by Edwin D. Maberly and Raylene M. Pierce
- The Halloween Indicator, “Sell in May and Go Away”: Another Puzzle by Sven Bouman and Ben Jacobsen 2002
- The Halloween Effect During Quiet and Turbulent Times by Ramona Dumitriu and Razvan Stefanescu April 22, 2012
- The Halloween Indicator: Everywhere and All the Time by Jacobsen and Cherry Zhang October 2012
 The Halloween Indicator, “Sell in May and Go Away”: Another Puzzle by Sven Bouman and Ben Jacobsen (2002)
 “The January effect is a seasonal anomaly in the financial market where stock prices increase in the month of January more than in any other month.” Definition by Wikipedia.org
 “Exorcising Ghosts of Octobers Past,” The Wall Street Journal (Dow Jones & Company). pp. C1–C2. Retrieved 2007-10-15.