Instead of managing assets in-house, institutions annually delegated an average of $36 trillion over the period 2000-2012 to asset managers. This represented 29 percent of worldwide investable capital in fixed income and public equities. Institutions predominantly delegated this capital through active investment vehicles set up by asset managers.
Joseph Gerakos, Juhani Linnainmaa, and Adair Morse investigate these sparsely studied investment vehicles using an aggregate dataset obtained from a worldwide investment consultant. The dataset contains greater coverage of asset manager holdings and returns than has heretofore been available to researchers. It covers $18 trillion of annual asset holdings, and 3,272 asset managers, over the period 2000-2012. It reports information on the specific investment strategy that asset managers are using for each institutional client.
First, they estimate that asset managers charged the average institutional-delegated dollar in fixed income and public equities a fee of 45 basis points. In aggregate, these fees averaged $162 billion per year, or roughly twice the aggregate amount paid by retail mutual fund investors over the same period.
Second, they document that strategy-benchmarked tracking errors are comparable to the tracking errors of active mutual funds. Third, they find that the average annual gross-of-fees risk-adjusted return ("alpha") is 95 basis points, and the net-of-fees return is 49 basis points net of fees, when judged against strategy-level benchmarks.
Fourth, the data allow them to infer how asset managers achieve their positive net alphas. Implementing a multi-factor model based on William Sharpe's 1992 study of "Asset Allocation: Management Style and Performance Measurement," they form dynamic mimicking portfolios by estimating fund-level factor loadings, and implement the model by limiting the factors to be tradable indexes and the weights to be long-only and to sum to one. When they compare the returns of active managers with the returns on these mimicking portfolios, they find no excess return. This suggests that asset managers provide institutional clients with profitable systematic deviations from benchmarks, or so-called "smart beta."
They explore whether institutions could have performed as well as asset managers over the sample period by implementing dynamic, long-only mean-variance efficient portfolios using the tradable factors, and find that institutions, if they had the knowledge and skill to implement mean-variance efficient portfolios, would have obtained Sharpe ratios similar to those of asset managers, after allowing for trading and administrative costs. They therefore conclude that during the sample period, at the margin, asset managers earned their fees. Their estimates also imply that the introduction of liquid, low-cost-factor ETFs may erode the comparative advantage of asset manager funds.
Finally, when the researchers compare asset manager returns to market returns simply by subtracting index returns, they find that asset managers earn an annual market-adjusted gross alpha of 129 basis points over the 2000-2012 period, or $432 billion per year, with $260 billion accruing to institutions and $162 billion to asset managers. The positive gross alpha over the market together, with the adding-up constraint that requires that the gross alpha earned by all market participants must be zero, implies that the market-adjusted gross alpha of all other investors must be negative.
National Bureau of Economic Research, 1050 Massachusetts Ave.,
Cambridge, MA 02138; 617-868-3900; email: email@example.com