Martijn Cremers is an associate professor of finance at the Yale School of Management. He and his Yale colleague Antti Petajisto have conducted research that focuses on mutual fund management, and their paper, How Active is Your Fund Manager? A New Measure That Predicts Performance, has been widely cited for its role in identifying successful active managers. His work has been the focus of two articles in Advisor Perspectives, From Yale University: New Research Confirms the Value of Active Management and Compelling Evidence That Active Management Really Works.
Dan Richards interviewed Professor Cremers on January 4 at the annual meeting of the American Economic Association in Atlanta, GA. This interview is one of a series that Dan conducted at that conference, and we will provide links to videos of his other interviews. Dan is president of Toronto-based Strategic Imperatives.
Let’s start by talking about what exactly active share is.
The active share of mutual funds distinguishes those fund managers who are active managers and measures how active they really are. It then attempts to correlate that with their performance.
We set aside the passive funds. We look at the funds that claim to be active, and we try to look what they are holding in their portfolio to see how active they really are.
The previous metrics, before our active share measure, did not look inside their portfolios.
As part of our research, we compare the holdings of the supposedly actively managed portfolio to what we think is its benchmark.
What led to your research?
An article in the Wall Street Journal piqued my interest. It looked at the Fidelity Magellan fund. It measured the R-squared obtained by regressing the returns of the fund against the S&P 500. It was very close to 1.0, suggesting that Magellan was not doing very much other than benchmarking the S&P 500
This is sometimes referred to as hugging the index.
Yes. That result suggested that it was hugging the index.
But another possibility was that it was not hugging the index at all, but it was taking diversifying positions away from the index, and it was so well diversified that its portfolio still ended up with an R-squared close to 1.0, even though the holdings were still different from the S&P 500.
When you started looked at the funds that were active and figured out the extent to which they were really active, what did you find?
We looked at a very broad cross-section of funds over a long period of time. We had 1,000 funds and over 20 years of data.
We basically found that until about 1990, almost all funds were active. Their active share was very high. After 1990, we saw a clear shift toward less active management – toward what we call closet indexing. Closet indexers have low active share and low tracking error volatility.
Tracking error volatility is a traditional measure of active management that compares the returns of a mutual fund to its benchmark. It looks at the residuals in the difference in the returns and calculates the volatility in that series.
That is a purely returns-based measure. Our measure is a holdings-based measure.
You look inside the fund and compare it to the benchmark, to see if it has the same weighting at a company or sector level.
We directly compare the weights in mutual funds to the weights in the benchmark.
Why do you think that after 1990 you saw a significant shift away from active management?
Today, we find that approximately 30% of funds are not that active and would be classified as closet indexers. Why did this shift happen? I’m not quite sure. I have some possible explanations for this. One is that the whole mutual fund industry has become much more focused on the benchmark. Relative performance evaluation became more important in the last 10-15 years.
Another possibility is that the very best managers with the very best ideas and with the highest active share have left the mutual fund industry for the hedge fund industry.
The really important insight for investors is that you’ve identified a new measure with which to evaluate fund managers. What outcome, in terms of returns, have you observed for those funds that have high active share?
Let me start with the funds that we argue are closet indexers – ones with low active share and low tracking error volatility. Because their holdings are similar to those of the benchmark, their gross returns (before trading costs and expenses) are similar to those of the benchmark. But their net returns, after trading costs and expense, are significantly lower.
Basically, an investor in a closet index will do worse than the benchmark by 100-200 basis points per year over a 15-year period for the 1,000 funds we looked at.
The message is very simple: You want to avoid closet index funds. It doesn’t mean that all closet funds do worse than their benchmark. It is just that if you look at all of them over a long period of time, they underperform by 100 to 200 basis points per year.
What about the ones with a high active share?
That result is probably the most controversial. We find significant evidence, in our view, that a lot of managers actually do have some skill. We focus on the high active share managers with a good track record – meaning that last year they did well – and that are not too big. We look at the funds below median size and they seem to do quite well versus their benchmark.
Is there persistence? Do managers that outperform in one year sustain their outperformance?
That is a very important question. Previous research has shown that.much of the persistence that seemed to be there goes away once you adjust for momentum. We looked at that in great detail.
First, the active share measure itself is quite persistent over time. If you sort funds according to their active share, you see that active share is very stable and doesn’t change much over time.
If the active share measure itself is persistent, then we argue that it more likely that fund performance is persistent. We find direct evidence for this. If we sort funds based on their performance and active share, we find the best outperformance in the next year for funds that were both very active and had a good track record last year. Momentum takes out some of that but leaves enough behind to say that we do find evidence of persistence.
What you’re suggesting flies in the face of the traditional view among many academics. Many academics suggest that the market is very efficient and that the best way to participate in the market is to buy index funds with the lowest possible cost. Are you suggesting that is not the case?
I am saying that to some extent, but I don’t necessarily disagree with the main views. We find other evidence that is consistent with the main view.
The largest funds, no matter how active they are, don’t show any outperformance. If the large funds don’t outperform, that suggests it’s not that easy to spot an active fund that will outperform. It’s really only those funds that have high active share and are not too large that have the outperformance
Those funds may not be easy to identify. And as those funds become larger, it becomes hard for them, even if they are very active, to outperform their benchmarks.
Read more articles by Dan Richards
