Money managers know how to buy. What they need to do is to learn how to sell. Most of them are terrible job at it.
That is the stunning conclusion of a research paper published earlier this month. If you manage money — or have any of your own invested in managed accounts — it is required reading. It goes far beyond the usual underperformance critiques of active management in an attempt to decipher why fund managers are so bad at this critical aspect of investing.
The paper’s authors make a number of observations about institutional managers, but the most profound and alarming is this: Fund managers would be better off, and in some cases much better off, merely selling holdings completely at random.
The study found money managers do exhibit skills in picking which stocks to buy. Where the problem arises is in their decisions about what to sell:
While investors display clear skill in buying, their selling decisions underperform substantially — even relative to strategies involving no skill such as randomly selling existing positions – in terms of both benchmark-adjusted and risk-adjusted returns.
There is an asymmetry between buying and selling. Purchase decisions are forward-looking, conducive to an analytical process that seem to be consistent and quantifiable. Selling stock in a portfolio is backward-looking; the retrospective nature seems to be susceptible to the kinds of behavioral biases and cognitive errors we typically think of as common among non-professional investors. The study found that many professionals were just as likely to suffer from these behavioral errors as the amateurs.
You might think that selling stock, so important to the total return of any portfolio, was a well-understood science. But you would be wrong. Instead, asset managers often lack any sound analytical framework for selling, using crude rules of thumb or gut instinct, neither of which has a good record for yielding positive results. Indeed, the most common reason to sell seems to be to buy the next great investment idea. This effort to chase stocks or other assets that look hot has been a recipe for disaster, both for returns and the active-management business model in general.
Of course, simply selling holdings at random is a methodology that seems unlikely to be accepted by institutions or paid for by investors.
This study is intriguing because the methodology used seems both robust and clever: the data set contained the daily holdings and trades of 783 portfolios, with an average value of about $573 million. The researchers then evaluated more than 89 million trading data points and 4.4 million trades. The study encompassed the years between 2000 and 2016 — a period that included at least two major market crashes, and two recoveries, one modest and the other significant for its duration. That avoids the common issue of a cherry-picked era. Overall, the research looked at 2 million sells and 2.4 million buys made by veteran institutional portfolio managers.
The clever part comes in how the researchers evaluated performance. Rather than comparing the portfolios returns to a benchmark, the study created what the authors call a counterfactual sell portfolio. Whenever the actual portfolio manager would sell a security, the counterfactual portfolio would sell a different, randomly selected security.