Thursday 30 December 2010

The January Effect v QE v Market Internals

December US equity markets have been typical in the shrinking of traded volume which makers them prone to being squeezed. At this point I hesitate calling it squeezing (hurting the shorts by mark-ups which forces closures (buying) of positions) as there have been 18 up-days out of 20 trading days in the MTD. Short squeezes are short-term phenomena. Although December began with a powerful rally, there has been anaemic follow-through as we near the end of 2010. As is common with the Christmas period, trade has been light, and price action muted. Based on market internals, institutional involvement has been clearly absent from the market over the past two weeks. In the last week or two of trading the broad market has exhibited indecision, and market internals (advancing volume to declining volume, for example) have differed between exchanges and indices. In addition there has been a noticeable flow of money into sectors that have significantly lagged the market over the past year (oil-related, homebuilders, banking, pharmaceuticals, and real estate). This is often a leading indicator that a rally may be nearing exhaustion. It is too early to definitively suggest that the market is about to change trend, but when laggards become leaders, caution is warranted.

There will likely be institutional buying early in the New Year as the whole staff comes back to work and markets are still hitting minor new highs. But will this persist? At this point the influence of QE on markets should be more important than calendar effects, but is the impact waning? Evidence from the bond markets suggests it may be.

As for the calendar effects I have reproduced below a long term study on the January effect. It comes from www.cxoadvisory.com/calendar-effects/



Does long term data support belief in exceptionally strong performance by the U.S. stock market during the month of January? Could this conventional wisdom be an artifact of data snooping or a victim of market adaptation? Robert Shiller's long run sample, which calculates monthly levels of the S&P Composite Stock Index since 1871 as average daily closes during calendar months, offers data for testing. Using monthly levels of the S&P Composite Stock Index for January 1871 through November 2010 (nearly 140 years) and monthly closes of the S&P 500 Index for January 1950 through November 2010 (nearly 61 years), we find that:

The following chart shows the average return by calendar month for the S&P Composite Stock Index over the entire sample period, with one standard deviation variability ranges. The average return for all 1,678 months in the sample is 0.42%. At 1.51%, January has the highest average return of all months. January has the lowest standard deviation of returns (2.86%), so this high return is not compensation for high variability.

October is the only month with a negative average return (-0.39%).

Is this apparent January effect consistent across subsamples?

The next chart compares the average return by calendar month for the S&P Composite Stock Index over the entire sample period and three approximately equal subperiods (46-47 years each). The performance of the stock market is consistently strong on average during January, and January is the best month for two of three subperiods. However, there is generally substantial variation in average returns by calendar month over the three subperiods.

For greater granularity and trend analysis, we examine relative performance during January by decade.

The next chart shows the outperformance of the average return for January relative to the average monthly return by decade over the entire sample period, along with a best-fit linear trend line. The trend line indicates that the magnitude of any January effect is declining, but the sample size in terms of number of decades (14) is small.

For even greater granularity, we examine the effect by year.

The next chart shows the outperformance of the return for January relative to the average monthly return by year over the entire sample period, along with a best-fit linear trend line. The trend line again indicates that the magnitude of any January effect is declining. Outperformance appears to disappear, or even reverse, during the past two decades.

A plausible interpretation of the above results is that there used to be a somewhat reliable January effect, but the market has adapted to extinguish it.

Since the Shiller data calculates monthly index levels as average daily closes during months (perhaps representing typical investor experience) rather than monthly closes, we compare the above results to those for monthly closes of the S&P 500 Index.

The next chart shows the average return by calendar month for the S&P 500 Index during 1950-2010, with one standard deviation variability ranges. For this calculation, we approximate the January 1950 return using the opening level for that month (since the December 1949 close is not available). The average return for all 731 months in the sample is 0.69%. At 1.10%, January has the fifth highest average return of all months, behind December, November, April and March. January has the second highest standard deviation of returns (4.84%), trailing only October.

Is performance during January consistent across sub-samples?

The next chart compares the average return by calendar month for the S&P 500 Index during 1950-2010 and two approximately equal subperiods (30-31 years each). During the first (second) subperiod, the performance during January is tied for third (seventh) place among the 12 calendar months.

For greater granularity and trend analysis, we examine relative performance during January by year.

The final chart shows the outperformance of the return for January relative to the average monthly return by year during 1950-2010, along with a best-fit linear trend line. The trend line indicates that any outperformance during January disappears or reverses during the past two decades.


In summary, evidence from long run data suggests that the conventional wisdom regarding outperformance of the U.S. stock market during the month of January derives either from snooping of an insufficient sample of older data or a real effect that the market has recognized and extinguished. Recent January returns are relatively weak.

No comments:

Post a Comment