"Closet Indexing" Misconstrued

Friday's Wall Street Journal featured an article about two Yale professors who have developed a measure for determining how much "active" management is present in mutual funds. The term they use for those funds which do not exhibit sufficient active management is a "closet index."

I won't go into their methodology. From what I read, it sounds feasible. They essentially compare the weight of a security in a portfolio with its weight in an index like the S&P500. The greater the summed (and normalized?) differences in the weights, the more "actively" they judge the fund to be managed.

My issue with this is their characterization of a mutual fund, or any managed fund or account, as a "closet index" simply because it does not meet, according to their criteria, a test of how securities are weighted in the portfolio.

To me, as an equity portfolio manager, I have a performance measure to attain, at some acceptable or pre-determined risk level. How I do it is up to me, within the bounds of my stated investment policies and strategies.

Therefore, although this is not what I do, if I were able to outperform the S&P by selecting the right 3 stocks to weight differentially, at the right times, why should that matter to a client?

What a client pays for, or should pay for, ultimately, is performance. Not style. If he could do better than me, he'd be doing it. If the S&P could consistently outperform my strategy over the long term, I'd be doing something else.

Why should the simple weighting of a portfolio incur the connotation that it is not "actively" managed? Granted, the authors further investigated the relationship between their measure of active management, and fund returns. Those with higher active management scores, as defined by the authors, had higher returns. Sort of.

The WSJ article states,

"According to the study, active-share percentages are a good predictor of performance. Funds registering the highest active share beat their benchmark index by an average of 1.30 percentage points per year, while those in the lowest active-share group produced returns that, on average, fell short of their benchmark by 1.41 percentage points."

While directionally comforting, I don't know if this result is actually statistically significant. The article is silent on that point. However, I doubt many clients would be impressed by a performance spread of less than 3 percentage points per year around a benchmark as descriptive of good and bad portfolio strategies. We don't know what the average annual standard deviation is, but it's a good bet that it would be higher than this difference of mean performances relative to a benchmark.

Which is to say, isn't performance, at a variance or volatility level, really the measure of a fund or account? How much real-time information does anyone have about mutual fund portfolios? Not much, actually.

So while the Yale authors' approach may segment management styles into active and less-active, it isn't clear that the resulting performance differences are sufficiently great to do much with this. If they had found a statistically significant difference from the benchmarks, and that average difference was more in the 6+ average annual percentage point range, I think I'd be much more impressed.