Sell in May and go away revisited.
The dictionary defines the word "anomaly" as something that is irregular or abnormal. It is a popular term among investment writers who wish to describe an event or strategy that contradicts the idea that stock price changes occur randomly. While most anomalies prove to be time-specific at best, the rule of liquidating stocks in May and reentering the market in November seems to be an exception. Bouman and Jacobsen (2002) attributed a much larger annualized return (17.1 percent) for the November-April holding period than for its May-October counterpart (6.0 percent) over the 1970-1998 years to the summer's slower trading activity--specifically, the timing and length of vacations. Lucey and Zhao (2006) suggested that greater returns during the November-April months are more than coincidental to the allegedly anomalous returns in January--the "January effect".
Kochman and Badarinathi (2008) examined returns from S&P 500 equities during the 1926-2004 period. Annualized returns for the November-April and May-October horizons were 15.57 percent and 9.07 percent, respectively. More recently, November-April dominated May-October by a margin of nearly 14 percent (19.5% versus 5.9 percent) over the 1970-2004 span. Unlike past researchers urging that investors should swap their stocks for T-bills in May, Kochman and Badarinathi recommended staying fully invested while substituting low-beta for high-beta stocks and selling call options. Most recently, Andrade et al. (2013) reported a return for November-April that was 10 percent greater than its May-October counterpart over the 1998-2012 years.
Our results are presented in Table 1. Not unlike Andrade et al. who replicated Bauman and Jacobsen's earlier sell-in-May study using an out-of-sample holding period, we wanted to test the seasonal effect over a period that did not overlap Kochman and Badarinathi. Our hypothesis is that the eight November-April periods between 2004 and 2012 will produce significantly greater returns than those generated by the eight May-October blocks. Facilitating that comparison will be an annualized return for each set of six-month periods. An annualized return will also be computed for the eight unbroken years to engender additional comparisons. All annualized returns will be accompanied by standard deviations to test for differences in volatility. The source of our data is the online website www .bigcharts.com.
The eight years immediately following Kochman and Badarinathi furnished more proof that selling in May and going away is a prudent strategy. The eight November-April holding periods produced an average six-month return of 4.92 percent--or 10.08 percent when annualized. By contrast, the May-October spans suffered an average six-month loss of 0.59 percent--or 1.18 percent when annualized. While the difference between means for the two regimes is not statistically significant per a simple f-test, it can be judged significant in terms of wealth creation for investors vis-a-vis wealth destruction.
Surprisingly, the larger annualized return for the November-April horizon was not burdened by the larger standard deviation. While the November-April months produced a standard deviation of 14.43 percent, May-October experienced a standard deviation of 19.89 percent. For the unbroken 2004-2012 period, the annualized return and standard deviation were 4.46 percent and 17.29 percent, respectively.
In sum, the so-called "Halloween effect" continues to haunt those investors who fail to differentiate between the November-April and May-October holding periods. But earning 10.08 percent annually vis-a-vis losing 1.18 percent should get their attention. Beyond the usual explanations such as summer vacations and slower efforts to arbitrage away inefficiencies, we suggest that the sell-in-May effect persists as a self-fulfilling prophecy. A sufficient number of investors believe that stocks will perform better in the November-April period and simply apply the appropriate buying and selling pressure.
Andrade, S., V. Chhaochharia, and M. Fuerst. 2013. "Sell in May and Go Away Just Won't Go Away." Financial Analysts Journal, 69(4): 94-105.
Bauman, S., and B. Jacobsen. 2002. "The Halloween Indicator, Sell in May and Go Away: Another Puzzle." American Economic Review, 92(5): 1618-1635.
Kochman, L., and R. Badarinathi. 2008. "The Halloween Effect: An Enduring Market Anomaly." Finance India, 22(2): 443.
Lucey, B., and S. Zhao. 2006. "Halloween or January? Yet Another Puzzle." International Review of Financial Analysis, 17(5): 1055.
by Ladd Kochman, Coles College of Business, Kennesaw State University. E-mail: email@example.com
Ravij Badarinathi, Cameron School of Business, University of North Carolina-Wilmington. E-mail: firstname.lastname@example.org
David Bray, Coles College of Business, Kennesaw State University. E-mail: email@example.com
TABLE 1. Comparative returns and standard deviations (November 2004-October 2012) Year November- May- April October 2005 2.36% 4.34% 2006 8.58% 5.14% 2007 7.58% 4.52% 2008 -10.57% -30.08% 2009 -9.90% 18.72% 2010 14.52% -0.29% 2011 15.24% -8.09% 2012 11.54% 1.02% Sum 39.35% -4.72% Arithmetic average 4.92% -0.59% Annualized return 10.08% -1.18% Annualized std. deviation 14.43% 19.89% * [[(1 + semiannual average).sup.2] - 1] ** semiannual standard deviation times sq. root of 2
|Printer friendly Cite/link Email Feedback|
|Author:||Kochman, Ladd; Badarinathi, Ravij; Bray, David|
|Article Type:||Author abstract|
|Date:||Mar 22, 2014|
|Previous Article:||The snob effect of red wine: estimating consumer bias in experimental blind wine tastings.|
|Next Article:||The Great Escape: Health, Wealth, and the Origins of Inequality.|