A new research paper from the Stanford University Graduate School of Business discusses the rationality (or otherwise) of the return expectations of institutional investors, with especial reference to public pension funds.
It finds that the returns that pension fund managers expect from their investments are extrapolated from the returns those investments have made in the past. That is a normal human tendency, to presuppose inertia, but it is not ideally rational.
The report is co-authored by Aleksandar Andonov, of Erasmus University, in Rotterdam, and Joshua D. Rauh, of the Stanford GSB, the Hoover Institution, and the NBER.
Their report begins with the observation that public pension funds in the United States now have assets under management of approximately $4 trillion.
This report’s data base comes about because the Government Accounting Standards Board, Statement 67, says that the managers of this $4 trillion must report long-term expected rates of return by asset class as part of the plan’s overall long-term rate of returns assumption. This is the only setting that Andonov and Rauh are aware of where “a large sample of institutional investors” expresses “their expected returns by asset class, alongside their targeted asset allocation.” The asset class breakdown generally involves public equity, fixed income, private equity, hedge funds, etc.
Goosing the Inflation Numbers
One finding of the Andonov/Rauh study is that portfolio expected return (ER) does not generally match the pension discount rate (DR). There is no consistency in direction of the mismatch: sometimes ER is above DR, sometimes below. The mismatches come about because there is more variation in the ER than in the DR.
In looking more closely at the ER, the authors break it down into its components, the inflation assumption and the real rate of return assumption. They find that the positive effect of past performance in expected return is mostly a positive effect on real return expectations.
Also (and perhaps disturbingly if one has hopes for the notion of fiscal responsibility) unfunded pension liabilities also have a positive impact on ER. But this operates differently – it operates through the inflationary component of ER. That is: state pension plans make aggressive assumptions about inflation so as to justify their high nominal return assumptions and in return reach non-alarmist conclusions about their unfunded liabilities.
Public and Private Equity
Andonov and Rauh acknowledge that the extrapolation of past return to future (real return) expectations could be justified if there were a significant amount of long-term persistence of the assets in pension fund portfolios. Unfortunately, the evidence of persistence is spotted. In particular there is no evidence of persistence in the public equity asset class.
There is, on the other hand, “stronger evidence in the literature of persistence in alternative assets, like private equity, where we also observe extrapolation.”
In private equity funds, there is evidence of persistence in the performance of consecutive funds managed by the same general partner. Skilled pension fund managers can “select and maintain long-term relations with the general partners in private equity by reinvesting in follow-on funds,” so that in regard to this asset class, the reliance on past returns in not irrational.
On the whole, though, the paper says that the evidence suggests “that pension plans excessively extrapolate past performance when formulating return expectations.”
Hedge Funds and Real Assets
There is also strong evidence of extrapolation, and excessive extrapolation, of future returns from past performance in both hedge funds and real assets. As to real assets, there is some partial justification for this extrapolation., A recent study (Andonov, Eichholtz, and Kok 2015) showed persistence in pension fund performance in real estate, so banking on that persistence isn’t itself irrational.. Still, in the “other components of real assets (natural resources and infrastructure” these authors are “not aware of evidence showing persistence.”
As to hedge funds, the evidence of persistence is “mixed” but these authors are “not aware of any evidence that public pension funds that have performed well in hedge funds over a multi-year period in the past continue to do so in the future.”
Indeed, as these authors observe, some hedge funds generate alpha by assuming tail risk, so if the multi-year period doesn’t include a sufficient number of tail events, it doesn’t justify confidence in persistence.
Rauh specializes in empirical studies of corporate investment. Andonov completed his Ph.D. in Finance at Maastricht University in May 2014.