Pensions’ flawed return forecasts

Modern City Concepts: intersection

We all know past performance is not indicative of future results, but a new study finds evidence that US public pensions are basing performance forecasts on their own prior experiences anyway.


US public pension plans base their return expectations for individual asset classes on their own experience with those classes, new research has found.

Stanford University’s Joshua Rauh and Erasmus University’s Aleksandar Andonov examined how institutional investors set return expectations. Their work shows evidence for the claim that pension plans “excessively extrapolate” from past performance when formulating return expectations.

Their study looks at 231 state and local government funds in the US over the period 2014-16. It states that it’s the first that looks at the relationship between beliefs and past experience for institutional investors.

By examining US public pension plans with combined assets of $4 trillion, the authors find that variation in institutional investor return expectations is influenced by the funds’ own investment histories.

Their paper, The return expectations of institutional investors, states a fund’s asset-class based expected returns, with its chosen weights, should equal, or at least approximate, its discount rate, or overall portfolio expected return. But the research finds that is often not the case.

The paper details the plans’ underlying assumptions, and their behaviours in making them, and states that the average returns experienced in the last 10 years of an asset class, and the extent of the plan’s unfunded liabilities, add “substantial explanatory power” regarding expected returns.

More specifically, each additional percentage point of past return raises the portfolio’s expected return by 36 basis points. And an unfunded liability equal to an additional year of total government revenue raises the portfolio expected return by 14 basis points.

Further, when unfunded liabilities are large, state pension plans are more likely to make aggressive predictions about inflation to justify high nominal return forecasts than to use higher real asset return assumptions for that purpose, the paper states.

Using past performance as an indicator of future performance could be justified, if performance is persistent, the authors state. But there is little evidence in most asset classes to indicate that is the case. In fact, the authors write that in public equity, skill and persistence in pension fund performance are weak or non-existent.

In private equity, however, there has been some evidence of persistence (although our recent interview with MIT’s Antoinette Schoar reveals her new study that could rebuke previous findings around this, see Private equity persistence slips), so the authors investigated this further.

They separated the private equity investments by date: those more than 13 years old and probably realised; those nine to 13 years old and most likely realised; and those 3 to 8 years old and only partially liquidated.

The results showed that the pension funds exclusively extrapolated the returns of the oldest group of investments.

The research uses pension fund data disclosed under US Governmental Accounting Standards Board Statement 67, which requires that public pension funds report long-term expected rates of return by asset class as part of a justification of the plan’s overall assumption for a long-term rate of return. The data reveals the return expectations of individual asset classes, alongside their targeted asset allocation.


To access the paper click below



Join the discussion