As factor-based investing grows in popularity, especially in the form of strategic (or “smart”) beta exchange-traded funds, an important question to ask is: How long might a factor-based portfolio underperform? After all, the studies that have established the outperformance of factor-based portfolios look at decades-long data. But during a long period in which a given factor-based portfolio outperforms, there are always subperiods in which it underperforms. The question is, how long might that be?
To answer this question, Maciej Kowara and I took an approach that we developed to measure how long active mutual funds might underperform over a given period and applied it to factor-based portfolios.
How we analyzed factor-based portfolios
To apply this approach to factor-based portfolios, we collected data that went back to the 1920s on five pairs of factor-based indices, each pair for each of five factors to form a comparison as shown in the following table:
The 5 factors we used in our analysis of factor-based portfolios
We used professor Kenneth French’s data library for five factors that go back to the 1920s; our data went through March 2018. Each of those factors is associated with a premium investors hope to earn by being exposed to them. To measure these premiums, we form two portfolios for each factor and compare their relative performance. Here are the factors in question:
- Market: The first is the equity market as a whole. Since we expect the equity market to outperform cash, we compared a total return equity index to a cash index. We found that for the nearly 19-year period, November 1963 to July 1982, the equity index underperformed the cash index. This is not to say that during those 19 years, there weren’t subperiods during which the equity index outperformed (there were), but rather, that had you held the equity index over this period, at the end of the period, you would have been better off holding cash, even though by a small amount.
- Value: For the value factor, we compared cheap to expensive stocks, as measured by their price-to-book ratios. According to the value effect, cheap stocks should outperform expensive stocks, which they do over long periods of time. However, we found that over the nearly 17½-period, July 1926 through December 1943, they underperformed by a small amount.
- Dividend: Similar to the value factor effect, the dividend factor effect is that stocks with high dividend yields should outperform stocks with low dividend yields. While this is the case over long periods, we found that over the 24½-year period, December 1975 through June 2000, low yield stocks did a little better.
- Size: Size effect is one of the best-known factor effects. Over long periods, small-cap (that is stocks with low market capitalization) outperform large-cap stocks. But, we found that over the 90+ year period that we looked at, small-cap stocks underperformed over a nearly 54-year period.
- Momentum: To examine the momentum factor, we compare stocks with high momentum to those with low momentum. By high momentum, we mean stocks with the best performance over the past year, relative to other stocks. Similarly, by low momentum, we mean stocks with the worst relative performance. According to the momentum effect, which is perhaps less fully understood than the other factor effects we discussed, high momentum stocks outperform low momentum stocks over the long run, which they do. But we found that over the 12½ -year period May 1932 through November 1944, low momentum stocks did a little better than high momentum stocks.
Patience might be key to factor-based portfolios
While many of these results are for periods in the distant past, we believe that they are informative as to how long a factor-based portfolio could potentially underperform. These results shouldn’t be interpreted as to mean that these factors do not pay off. For reasons that I will explain in a future blog post, I believe that these factors do pay off.
The lesson here is that a lot of patience may to be needed to reap their benefits.