Is faith in past winners justified?
Oliver Kunkel and Hansruedi Scherer weigh up the risks involved in selecting asset managers based on historical investment performance
Rankings of top performers are publicly available for retail funds. However, data-bases for institutional mandates (segregated accounts and institutional funds) are still incomplete and rare. For our historical simulation of picking past winners we use a set of track records that were submitted by asset managers as part of public and non-public mandate tenders for institutional clients in Europe. These track records must be representative for the mandate to be awarded, calculated according to performance presentation standards and are all gross of management fees.
Figure 1 summarises the number of track records (no duplicates) included in our analysis for different investment categories. As becomes apparent from the table, investors can choose from a broad range of asset managers.
For our simulation of picking past winners, we assume that an investor randomly awards a mandate to an asset manager which belongs to the 25% best performing managers over a time period of three years. Further, the investor measures the success of the manager selection by comparing the manager’s performance two years after the mandate has been awarded to the complete set of managers shown in figure 1.
By repeating this procedure for a large number of investors, we get the results for the manager selection strategy of picking past winners (figure 2).
Apparently, the returns of the top 25% performing managers were much better in the three years before selection than in the two years afterwards. The past winners for balanced mandates, for example, underperformed their benchmark on average in the two years following the hiring. In almost all asset classes, the past winners had at least one year of underperformance. More importantly, the returns of the managers selected based on the picking-past-winners strategy were, on average, not significantly better than all managers included in the simulation. For the two subsets with the largest number of managers – emerging market bonds and global equities – the selected managers were, on average, statistically worse than all managers included in the simulation.
These results suggest that picking past winners does not lead to ‘picking future winners’, at least not in the short term of two years after awarding a mandate. In this respect, our simulations are similar to the findings of Goyal and Wahal (2008), who analysed the selection of asset managers by institutional investors in the US between 1994 and 2003.
The observation that picking past winners is a questionable strategy is closely related to the lack of persistence in asset management returns. Empirical research in this field can mainly be found for retail equity funds, likely because of the better data availability. Studies by Hendricks, Patel and Zeckhauser (1993), Brown and Goetzmann (1995) and Blake, Elton and Gruber (1996) found that asset managers displayed no short-term persistence on their relative returns. Studies by Wermers (1996), Carhart (1997) and Barras, Scaillet and Wermers (2010) included risk-adjusted measures, and found little evidence for persistence in risk-adjusted performances of active managers.
“These results suggest that picking past winners does not lead to ‘picking future winners’, at least not in the short term of two years after awarding a mandate”
On the other hand, some studies, such as Bollen and Busse (2005) or Kosowski, Timmermann, Wermers and White (2006), found some evidence for short-term and long-term persistence in equity manager performances. Studies of bond portfolios, such as Blake, Elton and Gruber (1993) or Philpot, Hearth, Rimbey, and Schulman (1998) found no evidence for persistence in manager performance. Various other studies have shown some evidence for persistent investment returns over short periods of time, for example, Polwitoon and Tawatnuntachai (2006).
Do the findings for retail funds also apply to asset managers of institutional investors? It can be argued that persistently poor asset managers may be less frequent, since institutional investors, in contrast to private investors, apply a more systematic monitoring of their investments. Lakonishok, Shleifer and Vishny (1992), for example, found persistence in absolute returns of institutional equity funds over periods of two to three years. A later study by Busse, Goyal and Wahal (2010) failed to demonstrate persistence among asset managers that manage funds for institutional investors.
In summary, the empirical findings on performance persistence both for private and for institutional investors are heterogeneous.
It is noticeable that studies that take into account an adjustment for risk, such as the capital asset pricing model (CAPM) or factor models, appear to show even less evidence for persistence of manager performance. In evaluating manager returns, an analysis should therefore include risk measures in an attempt to separate manager alpha from systematic exposure to risk factors, such as credit risk, small-cap risk, and others.
However, the lack of empirical evidence of performance persistence rather supports the case that picking past winners is a questionable strategy for the selection of asset managers. Therefore, asset managers should be evaluated by a broader range of criteria derived from experience, as well as academic findings such as the investment approach, the portfolio-construction process, the characteristics of the product team, and of costs.
Oliver Kunkel is head of asset manager selection and Hansruedi Scherer is a partner at Swiss investment consulting firm PPCmetrics