What Is A Better Explanation For The Halloween Effect: Seasonal Affective Disorder (SAD) Or Vacations

Introduction

In this paper, we will talk about a some causes of a certain calendar anomaly. We will try to compare them to each other and see which of them would make more sense. In financial research, these calendar anomalies are among the most popular topics. Although often still not convincingly explainable, a lot of papers are aimed at addressing this phenomenon. One of the most well-known calendar anomalies is the so called Halloween effect, also known as the “Sell in May”-effect. According to its theory, stocks underperform in the periods from May until November. The strategy follows that stocks should be sold in May, but should be bought again at the beginning of November, after Halloween.

A lot of possible explanations have been given by one of the first researchers to publish a paper about this effect, Bouman and Jacobsen (2002). They gave possible explanations like data mining, vacations, risk, trading volume and interest rates, but also gave reasons for why these explanations could also not be the real cause behind the anomaly. In this paper we will give information about the explanation of vacations being the cause of this strange behavior of the stock market. We will bundle this cause with another possible cause, this being the festivities in the holiday season. This is the so called holiday effect which also affects the stock market. We will address this with the holiday argument.

Other researchers like Kamstra, Kramer & Levi (2003) suggested namely that the calendar anomaly could be caused by a change in risk taking, which in turn is caused by SAD, also known as Seasonal Affective Disorder. SAD is a disorder that causes some depression in the periods of fall, winter and the begin of spring. According to a study from Zuckerman (1984) a higher level of depression comes with a lower level of sensation-seeking. This lower level of sensation-seeking means a higher level of risk aversion, which could be the cause of a better performance of stocks during the fall, winter and the begin of spring.

In this paper, we provide an overview of the existing literature on the SAD hypothesis as well as the vacation hypothesis. We will discuss arguments from both sides, both studies against as well as studies in favor of the hypotheses, hoping to provide the reader with a comprehensive and objective summary of the current debate. We discuss the two hypotheses separately. First we will summarize the most important literature on the vacation hypothesis, after which we will provide a similar overview of the debate on the SAD hypothesis. For the last part, we will conclude which of the hypothesis is a better explanation for the anomaly.

Vacation hypothesis

The idea that vacations might cause the Halloween effect is not a new one. For as long as academics have been trying to solve the “Sell in May and Go Away”-puzzle, the idea that the length and timing of vacations might be the leading cause of the annual summer dip that has been persistent over so many years has been around. Bouman and Jacobsen (2002) were very quick to kill a lot of plausible explanations, discarding [... all of that ... ]. There was only one explanation for the “Sell in May”-effect that that the two authors found to be plausible: vacations. Approximating the length of vacations by looking at paid annual leave and the timing of vacations by looking at per month outbound travel, Bouman and Jacobsen were able to establish a strong link between the length of vacations in a country and the size of the Halloween effect in that same country. They also found that during popular vacation months (that is, months with a high level of outward bound travel), stock returns were significantly lower, at least in the May-October period they tested for.

In line with the findings of Bouman and Jacobsen (2002), Hong and Yu (2009) find a significant dip in trading activity during the summer holiday months (defined as July to September in Northern Hemisphere countries, January to March in Southern Hemisphere countries. Do note that Bouman and Jacobsen (2002) did not find the “Sell in May”-effect to be reversed in Southern Hemisphere countries, Bouman and Jacobsen explicitly mention Southern Hemisphere countries to also have higher stock market returns in the winter and in the spring (presenting it as evidence against the vacation hypothesis). Weirdly, Hong and Yu (2009) do find turnover and share return to drop in the Southern Hemisphere summer). They find that both big, as well as small investors, trade less during summer months, at the very least contributing to lower stock prices and returns. On top of this, after analyzing data from 51 countries, they find that in 38 of these countries, nationwide average turnover, as well as average share return, drops significantly in summer vacation months. They find the drop to be stronger in European and in North American countries than in the rest of the world, which,lines up nicely with the vacation hypothesis, as these two regions have the strongest summer vacation tradition.

Though the methodology between the two studies differs, with Hong and Yu (2009) focussing only on the months July, August and September while Bouman and Jacobsen (2002) analyze the entire May-October period, for example, their findings are similar and the studies mostly support each other.

Save your time!
We can take care of your essay

• Proper editing and formatting
• Free revision, title page, and bibliography
• Flexible prices and money-back guarantee

Place Order

Perhaps the most convincing evidence for the vacation hypothesis was delivered by Rantapuska and Kaustia (2015). Unlike Bouman and Jacobsen (2002) and Hong and Yu (2009), who looked at marketwide statistics, Rantapuska and Kaustria (2015) used account level stock data from Finnish investors in their analysis. While the study was focussed on testing the effect of mood on investor behaviour through analysis of the weather in Finland, it also touched upon the vacation hypothesis.

The paper finds that investors tend to sell of their stocks before vacations to fund their increased expenses during the summer vacation, “people sell stocks before and during their summer holidays [...] and then buy stocks after these periods.” (Rantapuska and Kaustia, 2015, p18). This is very much able to at least partially explain the returning seasonal stock price dip and makes for a strong argument for the vacation hypothesis.. Interestingly, they also find that investors tend to sell off their stocks before the weekends (essentially a weekly “vacation”), this leads them to also draw forward “mental closure” as a possible explanation for the pre-summer vacation sell-offs, along with funding summer-vacation consumption. Just like Hong and Yu (2009), Rantapuska and Kaustia (2016) found decreased trading activity during vacation periods (“when schools are out” p22), which once again supports the vacation hypothesis. The study finds that seasonal patterns in stock price data can best be explained by holiday seasons, which is, of course, exactly what the vacation hypothesis is about.

The SAD hypothesis

It did not take long for Kamstra, Kramer & Levi (2003) to come with a competing explanation for the Halloween effect. The hypothesis with which they came up with was one related to depression. The goal of their study was to research if seasonal depression, caused by differences in the amount of average length of day. would have any effect on stock market returns. The effect of SAD (Seasonal Affective Disorder) on behaviour is well documented, with psychological studies consistently showing that in a depressed state, investors are less willing to take risk. A higher level of risk aversion in the market would lead to lower returns on stocks, as investors flee the stock markets and move their capital to low-risk fixed income securities instead. This suggests that stock returns would be decreasing before and increasing after the winter solstice (the shortest day). After the short fall days leading to lower stock prices, the stock market would boom in the winter when length of day started increasing again and investors are once again willing to buy stocks.

By analyzing daily stock market data from nine different stock markets over the world, controlling for certain environmental factors like rainfall (‘precipitation’), cloudiness and temperature, Kamstra, Kramer & Levi (2003) concluded that there was a significant negative correlation between day length (and, by implication, SAD) and daily stock market returns. Kamstra and Kramer could not leave it at this. With varying partners, they have continued to write papers on the subject, building further and further on their SAD hypothesis and adding more components to it, rushing to defend it every time it is even remotely challenged.

Resistance to the SAD hypothesis starts quickly after it was first proposed, when Jacobsen & Marquering (2008) (Jacobsen being the same Jacobsen as from Bouman & Jacobsen (2002)) fail to reject or to confirm the SAD hypothesis, note that the first version of this paper was published in 2004. In their eyes, all Kramer, Kamstra and Levi (2003) did was confirm that there was a strong seasonal effect on stock market returns. What they failed to do however, is proof that it is SAD causing these changes, and not some other factor varying seasonally. To illustrate this, Jacobsen & Marquering (2008) ‘explain’ variability in stock returns by linking it to ice cream consumption (in an earlier version of the paper, not the 2008-version) the exact same way Kamstra, Kramer & Levi (2003) link it to seasonal affective disorder. Like they do, Kamstra, Kramer & Levi are quick to defend their hypothesis in a response paper published in 2009, directly examining the econometrics models used by Jacobsen & Marquering (2008). In a response to this response, Jacobsen & Marquering (2009), accuse Kamstra, Kramer & Levi (2009) of missing the point of their earlier paper, once again they do the econometrics and point out that seasonal differences in stock returns can just as well be attributed to other seasonally varying factors as they can to SAD. Their point is clear: “correlation does not mean causation” (Jacobsen & Marquering, 2009).

Opposition to the SAD hypothesis does not end here. Actually, it is rather widespread, though,as of yet, not very conclusive. For example, Kelly & Meschke (2010) show themselves rather sceptical, challenging both the psychological as well as the statistical/econometric base that Kamstra, Kramer & Levi (2003) provides. They go as far as calling the SAD effect “mechanically-induced” by the econometric model Kamstra, Kramer & levi use to ‘prove’ their SAD hypothesis. Kamstra, Kramer & Levi, of course, disagree, publishing a paper in 2012 firmly criticizing Kelly & Meschke (2010), accusing the duo of “making errors of commission and omission”, misinterpreting empirical results and ignoring statistical patterns in favor of the SAD effect. Kamstra, Kramer & Levi (2012) go as far as analyzing the data used by Kelly & Meschke (2010) by themselves and ‘proving’ the SAD effect using it.

SAD continues to divide. Somewhat recently, for example, Kaustia & Rantapuska (2015) investigate the extent to which variation in mood (specifically, variation in mood caused by variation in length of day (and variation in weather, which is not interesting for our purposes)) affect investor’s trading decision. They investigate this by analyzing direct account data of all Finnish investors over a seven year period, trying to find correlations to environmental factors (Finland is particularly suitable for this investigation, as length of day in Finland varies greatly from North- to South-Finland (on the same day) as well as over the year, as it is a high latitude country). They “find little evidence of seasonal affective disorder (SAD) affecting the tendency to buy versus sell” (Kaustia & Rantapuska 2015, p24), which is a pretty strong blow in the face for the SAD hypothesis. In this the case, the empirical data does not go its way.

Conclusion

From the literature, it very much seems that the vacation hypothesis is a more plausible explanation for the Halloween effect than the SAD hypothesis. On the one hand, we have an explanation around which it is rather quiet (which speaks in its favor), with the studies that are interested in it for the most part supporting the hypothesis. On the other hand, we have an explanation surrounded with heavy debate, decades of back-and-forths between academics and apparent cracks in the very foundation of the hypothesis. Judging by the literature examined, the vacation hypothesis seems a lot more reliable of an explanation than the SAD hypothesis. However, let it be noted that the vacation hypothesis is by no means definitive. As Jacobsen & Marquering (2008, 2009) mentioned in their papers reviewing the SAD hypothesis, a lot of things vary with the seasons and can thus be shown to “explain” the Sell-In-May-effect, even ice cream consumption. It might well be the case that vacations ‘just align better’ and that there is some other, previously unexplored factor at work, causing this mysterious calendar anomaly.

Literature list

1. Bouman, S., & Jacobsen, B. (2002). The Halloween indicator,' Sell in May and go away': Another puzzle. American Economic Review, 92(5), 1618-1635.
2. Hong, H., & Yu, J. (2009). Gone fishin’: Seasonality in trading activity and asset prices. Journal of Financial Markets, 12(4), 672-702.
3. Jacobsen, B., & Marquering, W. (2008). Is it the weather?. Journal of Banking & Finance, 32(4), 526-540.
4. Jacobsen, B., & Marquering, W. (2009). Is it the weather? Response. Journal of Banking & Finance, 33(3), 583-587.
5. Kamstra, M. J., Kramer, L. A., & Levi, M. D. (2003). Winter blues: A SAD stock market cycle. American Economic Review, 93(1), 324-343.
6. Kamstra, M. J., Kramer, L. A., & Levi, M. D. (2009). Is it the weather? Comment. Journal of Banking & Finance, 33(3), 578-582.
7. Kamstra, M. J., Kramer, L. A., & Levi, M. D. (2012). A careful re-examination of seasonality in international stock markets: Comment on sentiment and stock returns. Journal of Banking & Finance, 36(4), 934-956.
8. Kaustia, M., & Rantapuska, E. (2015). Does mood affect trading behavior?. Journal of Financial Markets, 29, 1-26.
9. Kelly, P. J., & Meschke, F. (2010). Sentiment and stock returns: The SAD anomaly revisited. Journal of Banking & Finance, 34(6), 1308-1326.
10. Kim, C. W., & Park, J. (1994). Holiday effects and stock returns: Further evidence. Journal of Financial and Quantitative Analysis, 29(1), 145-157.
11. Zuckerman, Marvin. (1984) “Sensation Seeking: A Comparative Approach to a Human Trait.”Behavioral and Brain Sciences, 1984,7(3),pp. 413–71.