The concept of active share was introduced several years ago as a measure of the degree to which a portfolio of stocks differs from its benchmark. One of the intriguing results of the initial research on active share was that high active share managers seemed more likely to outperform than low active share managers. This led, predictably enough, to a widespread belief that if active management wasn’t “working,” the solution was to be more aggressive, or as it was often expressed, to invest with more conviction.
At a basic level, there’s some obvious truth in this claim. A manager with very low active share makes very small deviations from his benchmark index, and so can hardly be expected to generate large excess returns. Similarly, a manager with high active share makes large deviations from his benchmark, which might lead to large differences in performance. The difficulty is that those differences are no more likely to be positive than negative.
Consider: can an underperforming manager (of which there were plenty last year) improve his results by randomly selling half of his names, thus holding a more concentrated portfolio? Doing so will increase active share for sure. Is there any reason to believe that it will improve performance? Of course not — which means that logically, it cannot be true that raising active share will enhance performance. In that sense, high active share may be necessary, but it is clearly not sufficient.
Researchers at AQR Capital Management have recently argued that high active share is not necessary to produce attractive results. Moreover, they show that the initial suggestion of a relationship between high active share and benchmark outperformance is due to a quirk in the data. (High active share managers tended to be small cap managers, and small cap benchmarks had negative alphas relative to the entire equity market. The high active share managers looked good because their benchmarks looked bad.) Properly adjusted, AQR concludes that “there is no evidence you’re more likely to be right just because you have a high conviction.”
This in no sense eliminates the usefulness of the active share concept — it’s a handy cross-sectional measure of a manager’s aggressiveness. It can help us frame reasonable expectations about differential performance. What it can’t do is tell us whether those performance differentials will be positive or negative. In the search for alpha, high active share may not be necessary, and is definitely not sufficient.