Measuring Investor Overcrowding
Are institutional investors mostly investing in the same securities at the same time? And can a better understanding of such crowding help investors?
With the recent news that Apple, Inc. has become the most widely held stock among both growth and value hedge funds, as well as growth mutual funds, it’s an opportune time to revisit the topic of investor crowding — and to explore some leading research published by Harvard Business School’s Kenneth A. Froot, based on one of the largest databases of buy-side investor positions.
As Froot pointed out in a 2011 appearance at the CFA Society of the U.K., researchers have built a whole variety of measures that help us understand what investors are doing. Working with a team from State Street Associates, the professor obtained access through State Street Global Advisors to custody data for about 25,000 underlying institutional portfolios. This data was then used to define concepts for measuring investor behavior.
For example, the team measured not only the flow of investor holdings but also the flow in relation to benchmarks and respective overweightings and underweightings, searching for what they call “measures of agreement,” or shared attributes. Figure 1 below shows the statistically significant levels of overcrowding in stocks that have relatively high commodity exposure, as well as undercrowding in inflation-exposed stocks, over the time period studied.
Figure 1: Overcrowding in Commodity Stocks and Undercrowding in Inflation Exposed Stocks 12/31/01 to 08/31/10.
Sources: State Street Associates.
Such an exercise surfaces a number of key methodological issues. For starters, evaluating such a large number of institutional portfolios and describing what is happening in the distribution involves using more subjective statistical analysis or “moments” than just simple averages or volatility.
Another point of concern is whether flows are “representative of real demand from buy-side investors in purchasing securities rather than just what is happening on the other side of the trade,” as Froot told members of CFA UK. In a secondary equity offering, for example, a company is forcing its shares into the market, which produces a supply shock that the buy side might be accommodating. “In flows, it is important to understanding the differences between a demand shock and a supply shock; both can involve frantic trading as well as varied responses to the price,” he cautioned.
Froot places research on overcrowding within the wider context of behavioral finance. Still, that may be of limited practical value to investors. Consider the housing market crash: Investors may have suspected that prices were due for a correction in the United States — and that such an event was inevitable — but could they have avoided the herding behavior?
On a more practical note, one takeaway from Froot’s research concerns investors’ responses to their own profit and losses on positions (P&Ls), which the professor and his team believe is measurable and potentially predictable. “With the huge amount of information we have on individual positions and the P&Ls that result from those positions,” Froot told the CFA UK audience, “the issue of when investors choose to take risk off or on a particular position becomes relevant.”
Past investor P&Ls, he added, should not matter according to classical economics. But a large literature about retail investor behavior suggests that past P&Ls do in fact affect current performance. The “disposition effect” explains why retail investors tend to ignore their investment accounts when markets fall. When markets go up, retail investors tend to trade more frequently and make ill-timed trading decisions. Loss aversion suggests that investors have a stronger preference for avoiding losses than acquiring gains.
“Institutional investors are naturally on the other side of such trades, and their behavior is exactly the opposite because of necessary market equilibrium,” said Froot. “If retail investors are long an asset relative to a benchmark, then institutional investors must be short relative to that benchmark.”
On average, he said, institutions are better than retail investors at cutting risks when they first experience losses. “When the losses are no longer small,” Froot explained, “then they tend to sell very aggressively.”
Thus behavioral finance can help researchers determine which P&Ls in fact matter. “Is it the P&L on the position; on the portfolio level; among associated investors or styles of investing; on the current positions; or on the positions in place two to four days ago?” said Froot, demonstrating the breadth of possibilities. “All of these considerations can be tested.”
Froot and his team are now working through their database to begin to understand more about how investors frame their trading decisions. Among the many challenges is one that stands out as key in the context of fast-moving markets: creating a framework for forecasting future risk before the information on which it is based becomes outdated.