What Professional U.S. Stock Investors Get Right

Scoring Managers "Selling Fast and Buying Slow: Heuristics and Trading Performance of Institutional Investors," a draft paper by Klakow Akepanidtaworn, Rick Di Mascio, Alex Imas, and Lawrence Schmidt, tackles an unusual question. Setting aside market-timing, professional stock investors have two jobs: 1) buying equities and 2) selling them. How well do they perform each role?

In resolving that query, the authors take conventional analysis one step further.

Researchers frequently use publicly available information to calculate an investment manager’s contribution. Compare a fund’s after-cost returns with those of an appropriate benchmark, to compute the performance difference. That amount measures the effect that the manager has on the portfolio—the “alpha,” in investment argot.

To be sure, this task is not as simple as it sounds. Selecting the appropriate benchmark is critical; the performance-measurement path is littered with the bodies of investment consultants who drew incorrect lessons because they made incorrect comparisons. Also, accidents happen. Often, what is credited or blamed on portfolio-manager decisions should instead be called “market noise.”

Eventually, though, sheer numbers suffice. It’s easy to get one fund wrong. It’s not so easy to get the entire industry wrong, over an extended time period. Enough studies have occurred so that we have a pretty good idea of how the typical U.S. equity-fund manager fares: slightly better than the benchmark before expenses are considered, slightly behind after.

The Big Dig So far, so good. What we have not understood, however, is how this "slightly better than the benchmark before expenses" performance has been achieved. Are professional stock investors: 1) good at buying securities but unskilled at selling them; 2) good at selling securities but unskilled at buying them; or 3) roughly neutral at each task?

Those answers have not been forthcoming because the data have been lacking. Understanding an investment manager’s overall outcome requires only a fund’s total returns, which are easy to gather. Determining the effect of buy and sell decisions is quite another matter. That task requires trade-by-trade information—when each stock is purchased, when each is sold, and at what price. And what outsiders have access to such data?

The authors do, that's who. Three are academics, but the fourth works for an investment-research firm called Inalytics, which collects and analyzes such material for its institutional clients. This stock-trading data cannot be found in public filings, nor in any mutual fund database, including Morningstar's. It is specialized knowledge.

(Which is my excuse for why Morningstar hasn’t already written this article.)

Armed with these figures, the authors studied investment managers’ buy and sell decisions. Each time a U.S. equity fund purchased a stock, the authors determined how that transaction affected the fund’s subsequent performance. They did the same for each security that the funds sold. In all, they evaluated 4.4 million trades, for 783 portfolios, spanning a 17-year stretch.

(Specifically, the authors compared the return for each freshly purchased equity position against the “counterfactual” possibility that the manager did not create a new position, but instead invested those assets into one of the fund’s current holdings, randomly selected. Similarly, the return for stocks that were sold was compared with the return of a stock that the fund continued to retain. Again, that security was randomly selected.)

Skilled Buyers The authors then distinguished between trades that occurred on the days when companies announced their earnings, and trades that did not. This, I confess, would not have been the first test that I would have considered when deciding how to extend the research. Perhaps it was not the authors' initial idea, either. Perhaps they first examined other possibilities. Either way, the distinction proved meaningful.

The authors’ conclusions are shown below.

To complete this column’s headline, what professional U.S. stock investors get right are their equity purchases.

They do well when they buy stocks on the day of earnings announcements; and they do well when they buy them on other days, too. Overall, the authors found that the funds’ newly purchased equities beat the control group by 60 basis points over the ensuring 12 months, and by a modestly larger amount for 24 months. The margin was similar in either case; whether earnings had been announced had little effect.

Mixed Sellers The news was otherwise for the stocks that were sold.

For one, there was a large performance difference between those securities that were sold on the day of an earnings announcement and those that were not. Those unloaded during earnings announcements subsequently trailed the control group, by roughly the margin that the buys exceeded. In other words, professional stock managers were as adept as using breaking information about equities that they immediately sold as they were at using: 1) breaking information about equities that they immediately bought, or 2) nonbreaking information about equities that they gradually decided to purchase.

What they couldn’t do well was sell stocks on days when there were no relevant earnings announcements. When investment managers were left to their own devices to sell equities, rather than being forced into the act by earnings news, they fared very badly. In fact, that was their worst decision. On average, the basis points forgone by discarding stocks under such circumstance were greater than the basis points gained from their correct judgments under other conditions.

In other words, among the best decisions that the professional stock managers could have made would have been to buy the very stocks that they chose to sell—those securities that they jettisoned at their leisure, on days without earnings announcements.

Isn’t it ironic? Friday’s column will discuss that finding in further detail. Until then, suffice it to say that investment managers appear to be smarter buyers than they are sellers.

Edit: The authors inform me that my jest was misplaced. They did not stumble upon studying the effects of earnings announcements. It was in fact their first analysis, based on prior articles that had used earnings announcements as a method of testing “attention shocks.”

They were ahead of me. No surprise, I know.

John Rekenthaler has been researching the fund industry since 1988. He is now a columnist for Morningstar.com and a member of Morningstar's investment research department. John is quick to point out that while Morningstar typically agrees with the views of the Rekenthaler Report, his views are his own.