Momentum Loses Its Momentum

There's an interesting new paper released last month on the Social Science Research Network: "Momentum Loses its Momentum." In simplistic terms, momentum strategies for equities go long "winners" and short "losers" (though one can build long-only portfolios). At least prior to the late 90s, momentum strategies had shown large excess returns -- more than 0.75 percent a month (at least before implementation costs).

The authors of the paper evaluated the robustness of momentum returns in the U.S. stock market over the period from 1965 to 2010. They found that momentum profits have disappeared since the late 90s, a finding consistent with an improvement in market efficiency -- once an anomaly is discovered and exploited it should shrink and eventually disappear. And since there's no systematic risk story to justify the momentum premium (as momentum is a behavioral phenomenon), the findings are exactly what you should expect from a highly efficient market. For example, the flood of money into hedge funds that use momentum strategies could explain the disappearance of the momentum premium. Before concluding that the opportunity to exploit the momentum premium has come and gone, let's take a closer look at the study.

First, 10 years is too little time to know if anything has changed in terms of long-run expected returns. To illustrate this point, consider the following. For the 63-year period 1927-89, the small-cap and value premiums were 2.6 and 4.5 percent, respectively. Over the next decade, the small-cap and value premiums were -1.5 percent and -1.7 percent, respectively. You might conclude that the premiums had disappeared. Then from 2000-10, the small-cap and value premiums were 5.6 percent and 7.0 percent, respectively. Also note that the market, size and value premiums weren't statistically significant over the period 1999-2010. And there have been other periods when the results were similar for all the risk factors, well before they were "discovered." The bottom line is that the premiums are volatile enough (as the annual standard deviations of the premiums are multiples of the premiums themselves) that you can see these kind of results over even fairly long periods even if the mean is relatively high.

Second, the excess monthly return of the momentum factor for the longest period that we have data for (January 1927 through August 2011) has been about 0.7 percent per month. The performance of momentum over the last 10 years is less than a one standard deviation event, well within the expected. Thus, it's very hard to hard to draw any conclusion from such a short period.
Third, most of the poor performance of momentum is due to 2009. In 2009, the momentum premium was -83.5 percent. The worst annual return prior to that year was 2003 (another year with a big reversal after a bear market) when the value premium was -24.4 percent. Thus, we can think of 2009 as the 100-year flood for momentum. Removing that year has a big effect on the results -- removing just the first quarter of 2009 would have done the trick. From 1999 through 2009 momentum had an annual return of 3.4 percent with Sharpe ratio of 0.15. Removing 2009, the annual return jumps to 10.2 percent and produced a Sharpe ratio of 0.44.

2009 was a year that exhibited a historic turnaround, not one that has happened every 10 years or so (which would give one cause to be concerned). What this seems to demonstrate is that any change in the momentum factor isn't really about the last 10 years. Instead, it's all about 2009. Of course, 2009 happened, and one shouldn't dismiss it. However, one should be more concerned if the poor performance had been more persistent and spread out.

When we look closer at the data, we see that there was no evidence prior to 2009 that the momentum factor had weakened, and there has been no evidence in the short period since. The authors' conclusion would have been more plausible if they showed a slow degradation in momentum profits -- as over time more and more money poured into the strategy. However, that just isn't in the data.

There's another important point to consider, one we persistently bring up -- you shouldn't consider an investment (or investment strategy like momentum) in isolation. Instead, the right way is to look how its addition impacts the risk and return of the entire portfolio. Investors dream of finding strategies that have negative correlation with other parts of their portfolio, especially if they have significant premiums associated with them. For those whose portfolios load on the value factor, momentum strategies bring an important diversification benefit, because momentum is negatively correlated with the value premium.

We can see this when we use a mean-variance optimizer. Even over the last decade, an MVO would have chosen positive weights on both value and momentum. The same would have been true for the 1990s when value did poorly. Value and momentum strategies in combination are more stable and more reliable, smoothing out returns. If you look at value and momentum decade by decade, you find that the returns are relatively stable, without a discernible trend. Supporting this point are the findings of the 2011 paper, "The Role of Shorting, Firm Size, and Time on Market Anomalies," which showed that returns to value and momentum strategies have been stable over the last century. And the same could be seen when looking at the international data, for the period it's available.

Photo courtesy of DaveFayram on Flickr.
More on MoneyWatch:
TIPS Update for October 2011 Hedge Fund Update for Third Quarter 2011 Third Quarter Update on 2011's Sure Things John Bogle's Dream Fund The Behavior of Individual Investors: To Err Is Human
Three ways I can help you become a wiser investor:

• Follow me on Twitter: http://twitter.com/larryswedroe.
• Preorder my next book Investment Mistakes Even Smart Investors Make and How to Avoid Them.
• Download the podcast of my radio show via Buckingham Asset Management's iTunes page.