Measuring the Value of LP Fund-Selection Skill: A Fair Comparison Framework
Using the IAS fair comparison framework, this publication explores the impact of choices an investor in private markets must make.
Institutional investors not only care about the final investment result, but also its path. A bumpy route implies some painful moments, making it challenging to “stay the course.” This concern also applies to active equity management. Active managers deviate from the benchmark (i.e., take active risk) to outperform, raising the possibility of occasional unfavorable results. Is the realized active return “worth” the pain of tracking error risk? Using the information ratio (IR) as a measure of “risk-adjusted performance,” do high active risk (AR) equity portfolio managers realize a higher (or lower) IR compared to low AR managers?
We examine three equity investment mandates: Large-cap (US), developed markets excluding North America (EAFE) and emerging markets (EM). We use the Russell 1000, MSCI EAFE and MSCI EM indices, respectively, as benchmarks for the three mandates. From the eVestment database, we collect information of 515 US, 258 EAFE and 474 EM managers who reported performance any time between January 1999 and December 2019 and with a continuous performance reporting length of at least 5y. We record each manager’s mandate, investment approach (fundamental or quantitative) and monthly performance (gross of fees).
We estimate a manager’s ex ante active risk, which is not observable, by their realized portfolio tracking error volatility (TEV) versus the benchmark. Although the ex ante AR and ex post TEV are unlikely to be equal due to estimation error and unexpected shocks, we expect them to be closely related. In other words, a manager budgeting higher active risk tends to realize higher TEV relative to peers with lower active risk. Consequently, we estimate and identify a manager’s ex ante active risk level (which we argue is an intrinsic characteristic) based on their long-term realized TEV rank relative to peers.
To accommodate CIOs who may or may not separate managers with different investment approaches, we consider two scenarios: Cross-approach,where the CIO does not control for investment approach when selecting managers, and within-approach, in which case the CIO first classifies managers by investment approach and then identifies the High/Low AR groups. In each scenario, we simulate many paths of investment performance by assuming the CIO randomly chooses managers from the identified group.
The results from the cross-approach show that there is no universal relationship between manager AR and IR. High AR managers generated higher IR in developed markets (especially EAFE) but did not do as well as Low AR managers in EM, although High AR managers achieved higher alpha across all three mandates. We argue that the negative relationship between IR and AR arises from the volatility uncertainty of EM measured by the difference between the ex ante forecasting volatility and ex post realized volatility (Figure 1). It is most difficult to forecast the EM volatility. Periods of unexpectedly high volatility produces unexpectedly high TEV which drags down the IR of High AR managers. Consequently, setting an aggressive active risk target in EM is a challenging endeavor though there might be better alpha opportunities for such managers.
The within-approach, where the CIO first selects the investment approach and then groups managers by AR, offers another perspective. A CIO hiring a US fundamental manager would have been better hiring a High AR manager; but if the CIO wanted to hire a US quantitative manager, it would have been better to hire a Low AR manager. For EAFE, the High AR fundamental portfolio outperformed the High AR quantitative manager, while the results were reversed for EM where the Low AR quantitative manager outperformed the Low AR fundamental manager. We suspect, though open to other explanations, that since risk in EAFE can be estimated reasonably well, the High AR fundamental manager is able to exercise skill aggressively with a low chance of an unexpected (high) TEV outcome, thus generating both a higher IR and alpha. In contrast, risk in EM is difficult to forecast, so even a skillful fundamental manager ends up, at times, being penalized in IR terms for targeting an ambitious active risk level despite having a considerably higher alpha. US managers, quantitative and fundamental, find themselves somewhere in between with a good ability to forecast risk but limited alpha opportunities.
This study suggests the CIO should think structurally when evaluating the AR-IR relationship. Certain types of managers may be more suitable to the CIO’s investment goal, and comparing managers with different investment approaches, even in the same mandate, may be misleading. Instead, comparing managers with a similar investment approach can help the CIO focus on manager idiosyncratic security-selection skills and avoid, if any, systematic bias. Some “unexpected” results (e.g., EM High AR fundamental managers having higher alpha but lower IR compared to Low AR peers, and EM Low AR quantitative managers having lower alpha but higher IR compared to High AR peers) imply an opportunity for the CIO to assemble a portfolio of managers with distinct strategies to improve portfolio IR while meeting their active return target. We offer a more comprehensive discussion of this issue in an earlier research report: What is the Optimal Number of Equity Managers? A CIO Toolkit for Manager Allocation.