Blog - Investing Notes

May 29, 2008 - Turn-of-the-Month Effect Trend and Interaction

One of the stock market anomalies we seek to exploit in our Strategy Test is the Turn-of-the-Month Effect, a significant concentration of excess stock market returns around the turns of calendar months. Has the strength of this effect changed over time? Does it (as asked by a reader) interact with the U.S. political cycle? Using daily closes for the S&P 500 index for the period January 3, 1950 through May 2, 2008 (700 months), we find that:

The following chart shows the raw returns for the S&P 500 index over the ten trading days centered on turns of calendar months across the entire sample period. The average return for these turn-of-the-month intervals is 0.61% with standard deviation 2.75%. The average return for all ten trading-day intervals in the sample is 0.34% with standard deviation 2.87%, confirming that turns of the month tend to have abnormally high returns. The red trend line slopes slightly downward from left to right, but the data is noisy.

For a big-picture perspective we examine average turn-of-the-month abnormal returns by calendar year.

The next chart shows the average by year of the abnormal return for the S&P 500 index over the ten trading days centered on turns of calendar months across the entire sample period (59 years, with 2008 a partial year only). To calculate these values, we:

1. Average the 12 turn-of-the-month raw returns for each calendar year.
2. Subtract from each such average the raw average return for all ten trading-day intervals during that year.

This method involves a slight overlap of daily return data from year to year. The average abnormal turn-of-the-month return so calculated across the entire sample is 0.22%. Turns of the month outperform the contemporaneous (by year) ten trading-day intervals in 41 of 59 years. However, the trend of outperformance is down, to the point of disappearance (substantially influenced by notable underperformance in 2007 and partial 2008). The combination of sample size and variability makes the trend inference iffy.

Is there any cycle to turn-of-the-month outperformance on an annual scale?

Visual inspection suggests no particular cycles in turn-of-the-month outperformance by year. One place to look for regularity is the U.S. political cycle. Grouping the results in the previous chart by presidential term year yields average turn-of-the-month outperformance of 0.33%, 0.11%, 0.29% and 0.17% compared to contemporaneous (by year) ten trading-day intervals during the first, second, third and fourth presidential term years, respectively. Subsamples are so small (14-15 years) that these differences are probably not meaningful.

In summary, the turn-of-the-month effect likely has no cycles on an annual scale since 1950 and may be disappearing.

For related research, see Blog Synthesis: Calendar Effects. See especially our blog entries of 7/20/06, 8/9/06 and 4/5/07 on the turn-of-the-month effect.