Dumb Alpha: Sell in May and Go Away?
Every April, I am asked by clients and fellow investment professionals alike if the old adage, “Sell in May and go away,” still holds true?
One of the key advantages of the ideas I present in the Dumb Alpha series is that they allow portfolio managers to rapidly improve their work-life balance. Since I am a naturally lazy person, I am constantly looking for ways to reduce my workload without my boss — or my clients — noticing.
The sell-in-May effect, also known as the Halloween indicator, is one of the most well-known calendar effects. It holds that investors can outperform a simple buy-and-hold strategy by selling stocks at the beginning of May and buying them back at the beginning of November. If this were true, I could dramatically improve my work-life balance by going on a six-month vacation in May, just to come back in November and work for six months until the following spring.
When I proposed this idea to my boss, he wasn’t very keen on it, arguing that, in largely efficient markets, this effect should not exist after transaction costs are taken into account. In other words, it should surely be arbitraged away by professional investors once widely known. I decided to dig in and look at the scientific evidence. After all, what is a weekend of extra research if one can expect to gain a half year off if proven right?
It is indeed correct that many calendar effects do not survive increased scrutiny. Examples like the turn-of-the-month effect or the day-and-night effect require quite a lot of trading in a portfolio. If trading costs are reasonably high, many of these effects become unprofitable. Similarly, some other well-known calendar effects, like the January effect, disappeared once they were described in literature and exploited by professional investors.
One of the first rigorous analyses of the sell-in-May effect was done by Sven Bouman and Ben Jacobsen, who looked at 37 international stock markets from January 1970 to August 1998. They found that the sell-in-May effect was present in 36 out of 37 countries and was statistically significant in 20 of them. The effect is not small, either. In the United States, Bouman and Jacobsen document a return in the November-to-April time frame that is 11 percentage points higher than in the May-to-October time frame; for the United Kingdom, the return difference is 24 percentage points — and can be traced back to the year 1694! So the sell-in-May effect has been around for a very long time, and, as it requires only two trades per year, it persists even after trading costs.
Efficient market advocates were quick to reply. Edwin Maberly and Raylene Pierce pointed out that the sell-in-May effect disappears in the US stock market once the months of October 1987 and August 1998 are excluded from the data. Could it be that the effect was caused by just two months of awful performance? If the returns were that lumpy, surely it wouldn’t be possible to exploit them, because most investors would have lost their jobs or given up long before the next event materialized.
In 2013, three researchers published what I consider the final verdict on the matter in the Financial Analysts Journal. Testing the sell-in-May effect with out-of-sample data from November 1998 through April 2012, they found that in the 14 years since the publication of Bouman and Jacobsen’s original analysis, the indicator did not disappear. In fact, on average, across the 37 markets studied, the out-performance in the winter months was still about 10 percentage points higher than in the summer months. They also found that the effect does not come in lumps. It exists in three out of four years and does not depend on specific industries, countries, or months.
It seems clear that the effect is both real and persistent. What causes it is totally unknown, although several hypotheses have been proposed, tested, and rejected. Here we have a Dumb Alpha generator that defies logic and explanation. But, as a mentor of mine used to say: “Truth is what works” — and, even though the underlying causes of the effect are unknown, it does seem like a true investment anomaly.
Now, I think I need to have a chat with my boss about my next vacation.
For more from Joachim Klement, CFA, don’t miss Risk Profiling and Tolerance: Insights for the Private Wealth Manager, from the CFA Institute Research Foundation, and sign up for his regular commentary at Klement on Investing.
If you liked this post, don’t forget to subscribe to the Enterprising Investor.
All posts are the opinion of the author. As such, they should not be construed as investment advice, nor do the opinions expressed necessarily reflect the views of CFA Institute or the author’s employer.
Image credit: ©iStockphoto.com/CSA-Archive