`Institutions are herding animals. We watch the same indicators and listen to the same prognostications. Like lemmings, we tend to move in the same direction at the same time. And that, naturally, exacerbates price movements’. [Wall Street Journal (October 17, 1989)]
As long-term investors, we should expect pension funds to focus on their long-term investment strategy. Pension funds also have predictable cash outflows and hence are unlikely to face substantial unanticipated short-term liquidity needs. They are therefore in a position to provide liquidity to financial markets at times when it is needed, for example, by investing in illiquid assets during financial crises, thereby helping stabilise financial markets and earning a liquidity premium in return.
However, pension fund managers tend to have similar benchmarks. This, in turn, might create a fear of relative underperformance compared with the peer group of fund managers and hence an incentive for them to herd, that is, to buy and sell the same asset at the same time; this type of herding is often termed “reputational” herding. If pension fund herding results in procyclical or positive-feedback investment strategies – buying assets in a rising market, selling in a falling market – this could have a destabilising effect on financial markets.
Our study – The market for lemmings: Is the investment behaviour of pension funds stabilising or destabilising?* – examines whether pension funds herd in and out of different asset classes. It also investigates whether herding is more predominant in subgroups, consistent with reputational herding. We classify pension funds into subgroups according to their size and sponsor type. Our analysis is based on a unique data set provided to us by State Street Investment Analytics, covering one third (by value) of UK pension funds’ asset holdings. The data covers UK private-sector and public-sector defined benefit pension funds’ monthly asset allocations over the past 25 years.
Our evidence provides support for the hypothesis that pension funds herd together in the very short term. There are a number of possible reasons for this. In addition to reputational herding, there is habit investing and momentum (that is, positive feedback) trading. Our evidence comes down in favour of reputational herding, since we show that pension funds herd in subgroups. We double sort funds by sponsor type (private-sector and public-sector) and by size (small, medium and large). We find that public sector funds follow other public sector funds of similar size. Similarly, large private sector funds strongly follow other large private sector funds.This could clearly be destabilizing if pension funds were also to follow the same positive-feedback strategies, such as momentum trading.
Our findings also indicate a strong short-term portfolio rebalancing by pension funds if short-term valuation changes drive the weights away from the asset mix specified in their investment mandate. Further, although we do not have any data on the pension funds’ liabilities, we can draw inferences about the changing maturity of their liabilities from the longer term dynamic asset allocation strategies pursued by the pension funds over the course of the sample period. The average pension fund – as represented by the peer-group benchmark – appears to rebalance its long-run portfolio in a way that is liability matching. As the maturity of pension fund liabilities has increased, large private-sector pension funds, in particular, have systematically switched from UK equities to conventional and index-linked bonds.
We then address more directly the question whether pension fund investment behaviour is stabilising or destabilising for asset prices. A common view in the financial markets is that pension fund investment behaviour is destabilising only to the extent that it moves market prices away from, rather than towards, their equilibrium values. So if pension funds were momentum traders, but better informed than other investors, then their trading might still be stabilising. A fundamental question therefore is whether the trading of the average pension fund reflects its superior information.
The answer to this question is no, since we find that the systematic changes in the strategic asset allocation of the average fund reflects its changing liability structure, rather than changes in the expected returns or risks on the assets (which are the signals to which informed active managers would respond).
We also find that the short-term rebalancing of the average fund largely corrects for deviations from the long-term strategic allocation caused by relative valuation changes in the different asset classes. As a result, the average pension fund’s investment behaviour can be destabilising, since it does not respond to the release of new information, with the risk that market prices can be moved away from their fundamental values.
We also investigate the market exposure of the average pension fund in our sample. We find that the peer-group benchmark returns match very closely the returns on the relevant external asset class market index (for example in equities and bonds). This result, coupled with the evidence on herding, supports anecdotal evidence that pension fund managers herd around the average fund manager who generates the peer-group average return and who is, in turn, a “closet index matcher”.
Finally, although we find evidence of pension fund herding, pension funds might differ in their ability to earn a liquidity premium by deviating from the peer-group allocation. Individual funds, for example, differ in the maturity structure of their liabilities and in the strength of their sponsor covenant. Another important factor is fund size. On the one hand, smaller funds might be able to earn a larger liquidity premium, since they are able to react more quickly to liquidity shocks than larger funds where “size is the anchor to performance”.
On the other hand, larger funds might have the scale not enjoyed by smaller funds to invest in illiquid assets such as property and infrastructure. We find no evidence of a liquidity premium being earned by pension funds. Indeed, funds that are less exposed to illiquidity generate a higher return than more exposed funds.
In summary, we find that the short-term objective of pension fund managers is to automatically rebalance their portfolios when valuation changes violate short-term investment mandate restrictions, while their long-term objective is to systematically switch from equities to bonds as their liabilities mature.
As a result, the average pension fund’s investment behaviour can be destabilising, since it does not respond to the release of new information. This mechanical rebalancing risks driving prices away from, rather than towards, equilibrium prices. The good news is that pension funds are not momentum investors, and hence, do not contribute to asset price volatility. The bad news is that the herding behaviour of pension funds, combined with their automatic rebalancing, which is driven by their liabilities rather than by expected risk and return, can prevent asset prices reaching their fundamental values. We find no evidence of a positive liquidity premium in the pension funds’ total return in excess of the peer-group return, which itself was similar to the market return as measured by external indices.
The bottom line is that, although they are long-term investors, UK pension funds have not earned a positive long-run liquidity premium on their investments. This is because their behaviour is driven by different incentives. Pension fund managers fear relative underperformance against their peer group, which encourages them in the short-term to herd around the average fund manager, who turns out be a closet index matcher.
Professor David Blake is director of the Pensions Institute at Cass Business School
Gabriele Zinna is an economist in the directorate general for economics, statistics and research at the Bank of Italy
Lucio Sarno is professor of finance at Cass Business School
* More information at www.pensions-institute.org/workingpapers/wp1408.pdf
Key points
There is an incentive for pension funds to herd, or to buy and sell the same asset at the same time.
Pension funds herd together in the very short term.
The average pension fund’s investment behaviour can be destabilising, since it does not respond to the release of new information.
