When it comes to identifying a successful portfolio manager, certain skills and traits appear more closely linked that others. One of the key characteristics is the ability to understand and play by the rules of probability, letting winning positions run whilst trimming and closing losing positions.
The disposition effect in behavioural finance states that investors are more likely to sell winning positions and hold on to losing positions, given the displeasure of incurring losses. Thus, portfolio managers are likely to lose money through their sell decisions than they would be expected to do by chance alone. Up to now, academic studies on the disposition effect have typically used data from individual investors across a range of asset classes. What this doesn’t tell us however, is whether this bias is also present in the decision making of professional portfolio managers.
In this paper, we use our Peer Group Database of institutional investor decisions to explore why some portfolio managers don’t run their winning positions and cut their losing positions; indeed these portfolio managers do the opposite, to the detriment of performance and alpha.
Our findings reveal that portfolio managers do have skill, but it often it is offset by a series of predictable and observable shortcomings. Ultimately, this means it fails to translate into superior performance and alpha.
Behavioural finance explains portfolio managers’ lack of skill when selling through the disposition effect and more specifically prospect theory and mental accounting, the psychological and behavioural hurdles facing investors.
Prospect theory suggests that investors are risk averse when looking at gains but tend to become risk takers when confronted with losses.
Mental accounting, meanwhile, points out that investors view each position within a portfolio as an entirely separate item, and thus treat them in an inconsistent manner.
To test these hypotheses and our own observations, we turned our Peer Group Database of institutional investor decisions. We wanted to see whether the majority of stocks being sold had been winning positions, or whether investors were disciplined by running winning positions and systematically weeding out losing positions.
Our database is substantial, so we removed any biases that may have arisen from the self-selection of portfolios by our Asset Manager clients.
This meant we only used decisions and transactions provided by our Asset Owner relationships, totalling 45,000 individual trades across a broad spread of industries, regions and benchmarks.
Through our analysis of the data, we found that poor selling negatively impacts returns by an average 94 basis points (bps) per annum (p.a.). Meanwhile, favourable buy decisions, which contribute 47bps p.a., offset this but this is clearly not sufficient to offset the losses from sell decisions.
We also examined what impact the disposition effect has on performance, using our proprietary tool, Trading P&L. In doing so, we found that the stocks sold negatively impacted on performance by a significant 94bps p.a. – which greatly increases our understanding of the skills set (or lack of) of the average portfolio manager, and provides real evidence of the impact of poor sell decisions. Whilst we found that the buy decisions added value (47bps p.a.), this was not sufficient to offset the losses from sell decisions.
The disposition effect shows that investors tend to lose more money when selling than by chance alone. This theory and our results support our consideration that portfolio managers may well have skill, but this does not result in superior performance as this skill is offset by a series of predictable and observable shortcomings – essentially, poor sell decisions.