Is the January Effect Real?

In a previous report, we examined the January Effect (December 25, 2008). This is the popular idea that stocks go up during the first month of the year because investors are sitting on cash accumulated as the result of tax loss selling and they redeploy that cash back into the markets in January. We found some support for this idea in that stocks do tend to show gains more often in January than in a typical month.

When considering this hypothesis, an additional test needs to be conducted. If tax losses are the rationale behind the expected rise in stock prices for January, then the first month of the year following a down year should invariably be up.

We put this idea to the test, following the same methodology in our previous report. We start by adding the results of our tests to the table originally presented to illustrate the January Effect. Using all available data, the results are summarized in Table 1.

click to enlarge

Table 1: A comparison of the percentage of the time various indexes close higher in the first week of the year to the percentage of up weeks in their history.

The NASDAQ Composite index performs just as well whether the previous year was up or down. Mid cap stocks and small cap stocks perform slightly better in the first week after a down year. Large cap stocks tend to underperform. These initial results support the idea that tax loss selling in December drives stocks higher in January.

One important consideration in this research is that the sample sizes are generally not statistically significant. The S&P 400 and Russell 2000 indexes have only three years when the conditions were met (prior to 2008). The large cap indexes show a tendency to perform slightly worse after a down year, and there are 29 years in the test for the S&P 500; 39 years for the Dow.

Monthly results are shown in Table 2.



Table 2: A comparison of the percentage of the time various indexes close higher in January to the percentage of up months in their history.

We can see in Table 2 that the January Effect does not seem to hold up over the course of the month after a down year. These results show that the January Effect may not actually exist. If it was caused by investors reinvesting proceeds from tax loss selling, we should expect to see better performance after down years, and we do not see this. The impact should be greatest in the speculative stocks, and instead we see some statistically insignificant selling in these stocks.

Based upon this research, we can conclude that the January Effect is most likely an example of relative strength, rather than a seasonal tendency. Recent performance often drives investor behavior. Gains often follow gains. Losses in December do not carry over to gains in January, despite widespread faith in the January Effect.

Disclosure: NO Positions

Comments

[Steven Hansen:]

Mike, good work.

Jan 04 07:05 PM

[wes wornom:]

You cited the January effect in your recent article. There is no predictive effect to January's performance. The problem with the "forecasting" aspect is that the performance of the first month is baked into the yearly number. I saw a study of this supposed effect in the 1980's. I think that it was in Barron's magazine inthe 1980's. It showed that you could use the first month of any 12-month period (January, February, March, etc.), and use the first month's performance in that period to "forecast" the average for 12 months. The fallacy is that the measuring period in included in the forecast period, making the "forecast" meaningless. A forecast, to be a forecast, must not include the base period.

Take a simple example. Let's say that the period was two months, not twelve months. If you know that the first month declined, the average for the two months is likely to decline, too. The same would apply if the first month rose. Isn't it clear that this could apply to any two-month period in the year?

Sincerely

Wesley Wornom

Jan 31 10:53 AM