In the study "Selling Fast and Buying Slow: Heuristics and Trading Performance of Institutional Investors" the authors examine the buying and selling decisions in institutional business.  To this end, the four participating researchers analysed the trading decisions of experienced portfolio managers using a comprehensive data set containing information on the daily portfolio and the transactions carried out. In this way, a total of 783 institutional portfolios with an average value of 573 million US dollars was examined from 2000 to 2016, comprising 2.4 million buy and 2 million sell orders.
Buy vs. sell decisions
A first finding of the study is that the portfolio managers achieved clear added value in their purchasing decisions. This can be seen by comparing the transactions with alternative, purely random purchases of shares from the portfolio not traded on the respective day. Outperformance can be demonstrated by means of both simple and risk-adjusted returns.
The situation is different when the selling decisions are considered. These are also assessed by comparing the transactions with alternative, purely random sales of shares from the portfolio that were not traded on the respective day. Managers here perform up to 100 basis points worse on a one-year horizon. The underperformance is shown by both simple and risk-adjusted returns.
The authors describe an asymmetrically pronounced attention of the portfolio managers as the reason for the diverging buying and selling yields. According to this, they do significantly more research for buying decisions than for selling decisions. The latter are often based on simple heuristics such as conspicuously high or low past returns and are primarily regarded as a source of cash to implement new purchase decisions.
The fact that managers do not lack the skills to make profitable sales decisions, but only the attention necessary to do so, becomes apparent when looking at the situation in a differentiated way: The researchers separately examine those sales decisions that were made on days when companies announced their quarterly figures. On average, these sales based on actual information outperformed random sales.
The graphs below show the average returns of buy and sell decisions that took place on the day the quarterly figures were announced (red bars) and on all other days (blue bars) - relative to random buying and selling. The chart on the left shows that the buying decisions add value regardless of when they were implemented. In contrast, the right-hand chart shows that only selling decisions made on the day the figures were announced add value.
Retrospective selling decisions
The research shows that managers are proactive in their purchasing decisions and rely on real information, which enables them to add value. The authors of the study conclude that portfolio managers see their job primarily in finding the next big investment idea. However, in order to implement this idea (as quickly as possible), selling discipline is neglected by using simple heuristics.
In contrast to buying decisions, sales are therefore not forward-looking, but retrospective. The apparent advantage of this is that sales can be justified quickly in this way:
a sharp rise in prices: Upward potential exhausted, mean reversion expected
sharply fallen prices: basic investment idea no longer valid, higher fluctuation margin expected
While these arguments may be partially valid, the study's investigations show that the use of mere heuristics is a clear mistake. Sales decisions perform particularly badly when markets are in a turbulent phase and the aim is to build up (quickly) cash positions. There is also the following correlation: the weaker the overall performance of a fund manager, the poorer the returns on sales compared to random comparisons. On the buying side, however, such a correlation does not appear to exist.
The study shows that portfolio managers focus on finding undervalued stocks or giving higher weight to stocks already in the portfolio. Extensive research is often conducted for these purchase decisions and on average a clear added value is achieved. Experience shows that portfolio managers closely monitor their absolute and relative returns, but rarely or never analyse what returns they have missed out on due to poor selling decisions. This is because sales are often only seen as a starting point for implementing new purchase ideas and are therefore planned at short notice.
The greatest potential for improving the returns achieved by institutional investors therefore lies in making a comparable effort for sales as for purchases in order to sell the least attractive stocks in the portfolio in a targeted manner. Or to put it simply: to pay as much attention to selling as to buying.
Selling discipline is key
Portfolio managers achieve clear added value with their buying decisions, but make selling too easy and lose performance there.