A new paper by the Pension Institute looks at the investing behavior of UK defined-benefit plans over the past 25 years, and finds a degree of correlation that should be worrisome to plan sponsors. The sponsors might want to design an improved incentive structure “that can better motivate pension funds to move away from ‘short-termism’ and focus on long-term returns and fundamental value.”
The paper, The Market for Lemmings, is the work of David Blake, Lucio Sarno, and Gabriele Zinna. Blake and Sarno are each affiliated with the City University of London, Cass Business School, and Zinna with the Bank of Italy. Sarno is also the author of a graduate textbook, The Economics of Exchange Rates, from Cambridge University Press.
The Market for Lemmings begins with the observation that there exists a large body of scholarly work on the herding behavior of investment institutions. One of the common observations of scholars in this field is that institutions tend to know about one another’s trade, more than do individual investors. This, along with the high correlation of the signals to which they pay attention, results in correlated behavior. That much is generally accepted.
Carrying Forward the Work of LSV
But, Blake et al continue … there has been more work on institutions generally, or on the specific concerns of mutual funds and/or hedge funds, then there has been on pensions. In fact one of the few studies that have looked specifically at herding in the pension fund industry is a quarter century old – a study by Lakonishok, Shleifer, and Vishny in 1992.
That LSV study was limited to all-equity pension funds, and it failed to find evidence of herd-like movements into or out of specific stock positions. But at the same time LSV acknowledged that “there are times when money managers simultaneously move into stocks as a whole or move out of stocks as a whole,” and their database didn’t allow them to say anything more about “this type of herding.”
There was another important caveat in that LSV report. It said that “herding might be more prevalent among subgroups of pension funds than in aggregate.”
Blake et al picked up on both of those caveats of the 1992 study, and designed their own database so that it would be sensitive to both of the sorts of herding waved off by LSV. They have made three findings. First, yes, pension funds do herd in their investment, and this behavior shows up in very short-term, terms, month to month. Second, yes, their herding becomes more evident as the data is broken down into sub-groups, either by separating public from private sector pension managers or by disaggregating by fund size. Related to this, they observe that the herding isn’t the consequence of superior information, but of mechanical portfolio rebalancing.
Impact of Herding on Prices
Finally, Blake at al find that pension fund herding has an impact on the price of assets in financial markets. That would seem to be analytically true: a “herd” moving in the direction of X is by definition an increase in the demand for X, after all. What is more substantive is the issue whether this asset price move makes for healthy markets by moving prices toward fundamental valuations. Here, Blake and his associates can’t be optimistic.
To test the significance of the asset price changes, these authors looked at the cumulative absolute abnormal returns (CAAR) in the months before and after herd trades. The idea is that if the asset prices moved at such times in the direction of fundamental valuation, “the deviation of market returns from their fundamental values in the months following their trades should be smaller than the deviation observed in the months preceding their trades.” The result of this natural experiment was that the trades have no such stabilizing consequences.
Regarding the subgroups mentioned above, one might well ask: why is it important (non-arbitrary) to distinguish between small and large pension funds for the purpose of studying investment styles and herding? The answer, which appears in a footnote in this report, [though it arguably could have been displayed more prominently], is that size is “an important determinant of pension fund asset allocation. Portfolio return volatility is highly negatively correlated with fund size, possibly reflecting the fact that small funds are generally less diversified than large funds.”