Morningstar's Active/Passive Barometer and other studies have demonstrated that most active managers in the U.S. have struggled to beat their index counterparts over the long term. But active managers have had greater success in some periods than others. A theory known as Dunn's Law suggests that this pattern can be explained by stylistic differences between active and index portfolios1. The idea is that indexes tend to be more style pure than their active counterparts, so they should be more difficult to beat when the style they represent has strong relative performance. Conversely, they should be easier to beat when their style has weak relative performance.
We conducted a study to put Dunn's theory to the test, using data from the Morningstar Active/Passive Barometer for 12 Morningstar Categories. The study found:
- Dunn's Law has some merit, though the results didn't always align with the theory's predictions.
- Although the relationships we examined didn't always match the predictions of Dunn's Law, the returns to various investment styles and risk factors explained much of the variation in the success rates observed from September 2002 through December 2017.
- Active managers' success rates are noisy and tough to predict. Attempting to time exposure between index and active managers is akin to timing just about anything else in the market and not advisable.
This study starts with the results of Morningstar's Active/Passive Barometer, which compares the performance of U.S. active managers against actual index funds in 12 Morningstar Categories.
The Active/Passive Barometer defines success rates as the percentage of U.S. active managers in each category that both survived and outperformed a composite of all index mutual funds and index-tracking exchange-traded funds in their category over a given period. Success rates tend to be more volatile over shorter horizons, so we focused this analysis on the quarterly rolling one-, three-, and five-year success rates from September 2002 through December 2017.
We built a regression model to explain the variation in active managers' success rates within each category over time. For the nine U.S. equity categories, the explanatory variables in the model included the returns to market risk, small size, value, and international relative to U.S. stocks. We calculated each of these explanatory variables (or factors) using index return spreads.
We set up similar regression models for the foreign large-blend and diversified emerging-markets categories, using region-specific indexes to construct the explanatory factors. Similar to the U.S. model, we included market risk, small size, and value factors in these regressions. To capture potential differences in geographic exposure between active and index funds, we added a U.S. and emerging-markets factor to the foreign large-blend model, and a developed-markets factor to the model for diversified emerging-markets funds. We also added a currency risk factor to both models.
The model for the intermediate-term bond category included market, credit, and interest-rate risk factors.
These regression models measure how closely the variation in active managers' success rates is linked to the payoff of potential style differences between active and index managers. For example, if active managers in the large-value category have greater exposure to mid-cap value stocks than their index peers, their success rates should improve when smaller stocks beat larger stocks. The regression would detect that relationship.
However, these models don't capture every way in which active funds differ from their index peers. For example, they ignore differences in security selection and sector exposure unrelated to the factors, which may limit the models' explanatory power.
With this caveat in mind, the models allow us to test the predictions of Dunn's Law. If Dunn's Law is accurate:
- The success rates in all categories should be negatively related to the performance of the market. This is because index funds are always fully invested, while active managers tend to carry larger cash balances. So, it should be harder for active managers to beat their index peers during strong rallies, and less difficult during downturns.
- The success rates in the value categories should be negatively related to the performance of the value factor, while this relationship should be positive in the growth categories. The rationale is that active value managers likely have greater exposure to growth stocks than their index counterparts, since they are less constrained. Similarly, active growth managers likely have greater exposure to value stocks than growth indexes.
- The success rates in the large-cap (and diversified emerging-markets) categories should be positively related to the performance of the small size factor, since many index funds are market-cap-weighted, pulling them toward the largest stocks. This relationship should be negative in the small-cap categories, since many active managers have the flexibility to venture further into mid-cap territory than their index peers.
- The success rates in all U.S. categories should be positively linked to the performance of international stocks, since many active managers have some exposure to these stocks, while their index counterparts generally do not.
Success rates in the intermediate-term bond category should be positively related to the payoff to:
- Credit risk, based on the expectation that active managers have tended to take greater credit risk than their index peers.
- Interest-rate risk, based on the idea that the average active manager took greater interest-rate risk than index funds over much of the sample period (though that is probably no longer the case).
In part 2 of this article, we will take a look at the results.
1) Bernstein, W. 1999. "When Indexing Fails." 1999. Efficient Frontier. http://www.efficientfrontier.com/ef/499/indexing.htm
29 March 2018