Every spring the old Wall Street trading adage “Sell in May and go away” returns, essentially warning investors to sell stocks in May and come back in the fall (it is unclear whether they should come back in September or November). The effect has been attributed to everything from seasonal and biological effects of changing seasons to horse-racing to Treasury liquidity.
While it is doubtful if investors actually follow the maxim in a serious way, we were curious as to whether this seasonal effect actually exists. In other words, did it make sense to sell equities in May and buy them back after the summer, say in September?
We looked at total returns for the S&P 500 Index (including dividends) between 1978 and 2017 and separated out returns in each month of the year. This forty year period was selected because an investor could theoretically have implemented a strategy that sold the index in May and bought it back in September of each year during this time – for example, they could potentially have used the Vanguard S&P 500 index fund (launched in August 1976) and later the SPDR S&P 500 ETF (launched in January 1993).
The average annual return during this forty period was 13.09% – note that this is different from the compounded annual return of 11.83%, which is what a buy-and-hold investor would actually see. In contrast, the average return between May and August during this same period was 3.10%.
This does suggest that there may be some truth to the conventional wisdom that equities typically underperform during the summer months, with the bulk of gains coming outside of it. Yet the fact that returns are positive suggests that an investor who switched away from equities during the summer would have significantly underperformed an investor who did not. In fact, average returns across each of the four summer months are positive over the forty year period.
However, summer underperformance raises the question as to whether avoiding equities in these months, and moving to cash, provides a benefit in terms of risk-adjusted returns, i.e. after you account for volatility. Volatility can be a serious drag on actual compounded performance. For example, the arithmetic average return for an asset that rises 10% in one period and falls -10% in the next is 0%, but its compounded return is -1%.
To test this we compare a S&P 500 “Buy and hold” strategy to one that sells the S&P500 at the end of April and buys back in at the end of August. The latter strategy sits in cash during the four summer months – we use returns on the Bank of America 1-3 month treasury bill index to proxy cash returns.
A dollar invested in the S&P 500 index at the beginning of 1978 would have grown to $87.51 by the end of 2017 if you just held on – translating to a compounded annual return of 11.83% and overall return of 8,751% (before fees)! Of course, you would have had to sit through drawdowns as large as -43.75% (March 2000 – September 2002) and -50.95% (October 2007 – March 2009).
As the exhibit below illustrates, the “Sell in May, come back in September” strategy underperforms significantly. If you followed this switching strategy, a dollar would have grown to $65.24 over the same forty year period i.e. a compounded annual return of 11.01% and overall return of 6,524%. This is before fees and any transaction costs involved in switching back and forth.
While the summer switching strategy shaved off some of the drawdown in the 2000-2002 period (-25.41%), you would still have had to sit through a -47.00% drawdown during the Great Financial Crisis.
The “Sell in May, come back in September” strategy did experience lower overall volatility thanks to the strategy being in cash for four out of twelve months each year (a third of the period): 12.57% annualized versus 14.81% for “Buy and hold”. However, the latter’s higher volatility was more than compensated for by its higher returns.
The Sharpe ratio, which is the risk-adjusted return per unit of volatility, is 0.43 for the”Sell in May, come back in September” strategy, compared to 0.44 for “Buy-and-hold”. So the summer switching strategy fails to shine even after accounting for volatility.
Has switching worked more recently?
The switching strategy did outperform between 2000 and 2012, with an annual return of 4.23% compared to just 1.66% for “Buy and hold” over this thirteen year period. While the summers of 2001, 2002 and 2008 saw negative returns for the S&P 500, when the economy was in recession, 2010 and 2011 also saw significant summer drawdowns, courtesy of the European debt crisis. Though you would have needed a crystal ball to perfectly time these events.
However, you would have missed out on gains if you decided that the switching strategy was teh way to go after 2012.
A dollar invested in “Buy and hold” at the beginning of 2013 would have grown to $2.08 by the end of 2017, for an annual return of 15.79%. In contrast, the “Sell in May, come back in September” strategy would have grown that same dollar to $1.80, for an annual return of 12.52%.
In other words, switching may work in certain periods but it is extremely hard (or perhaps impossible?) to time these perfectly.
To summarize, equity markets generally underperform during the summer period between May and August. However, since the return during this period is still positive, and has historically out-performed cash, there appears to be no benefit to switching away from equities in the summer. This is true even when you account for volatility and look at risk-adjusted returns.