Gaming the Index
CRSP sets out a benchmark against front-running.
December 2, 2010
Money managers might well be afflicted with index fatigue. Yes, there are the usual job hazards of tracking errors and information ratios. But measured against what? What benchmark counts, and which benchmarks are to be blended together to provide a standard against which what a portfolio manager actually does can be judged appropriately?
A Canadian manager might match a portfolio against a blend of the S&P 500 and the S&P/TSX 60. However, if energy and resource stocks dominate the portfolio, it’s hardly a fair presentation. Better to have the benchmark as a blend of the individual GICS sectors (which is not to suggest that money managers are violating their GIPS obligations).
That’s probably too granular for trustees on investment committees.
Of course, if you want an index fund that represents a benchmark, you can always buy one. Or a futures contract, absent any reinvested dividends, of course. Marginal tracking error, assuming the index followed is representative of all the economic and investment opportunities. But perhaps the index isn’t representative, let alone the derivatives that track it.
Tracking errors can be the source of outsize profit. Wharton School finance professor Jeremy Siegel found that the original S&P 500 firms of 1957 outperformed the later versions of the index, reconstituted with that’s year’s flash in the pan.
That’s a datum worthy of further exploration, that index rejects contrarian opportunities. That may not, however, suit the bent of all institutional investors, who wish mostly to be assured that they are at least in the middle of the trend, quarterly or annually, rather than having the abject duty of explaining the outliers.
But there’s a problem which is hinted at above. Indexes group average results – and everyone wants to be above average. Most indexes can be gamed, however. They can be gamed by including stocks that are not on the index, but once were. That involves second-guessing the index selection committee, whose focus is on liquidity, market capitalization and representativeness. So money managers may be able to shift around just below the index cutoffs, where they see value on the basis of fundamentals or momentum or something else. They could be poised to capture off-index returns because they have lower criteria for liquidity, market cap and the like. Or not.
Or they could front-run the index. The Russell 2000 is reconstituted yearly, for example, and some small-cap stocks are handed a cap, gown and diploma and sent to make their fortune in the large-cap Russell 1000. The S&P500/600 and the S&P/TSX 60/Composite have similar, but more frequent, reconstitution ceremonies.
Given the public nature of the liquidity, market cap and representativeness criteria of the big-cap indexes, money managers are not making a random bet when they buy ahead of a company being included in a higher-level index. It’s not a cinch, but a calculated risk. And it takes money away from dedicated index investors.
The Center for Research in Security Pricing at the University of Chicago’s Booth School of Business is proposing a set of indexes that mitigates that front-running opportunity.
“The new CRSP Indexes blend advancements in academic research with current commercial practice in a fundamentally sound manner under the premise that an index must reflect the way that money managers actually invest.”
That sounds like a replication of money-manager alpha, or at least the reduction of a cheap structural beta opportunity.
What it involves is this: “CRSP Indexes encompass listed equity securities, including common stocks, REITs and business development companies (BDCs), of US-incorporated or US-headquartered companies with primary listings on NYSE, AMEX, NASDAQ or ARCA. CRSP assigns companies within the eligible US equity universe to portfolios based on cumulative market capitalization. There is no limit to the number of companies that may comprise an index.”
The numerical limit is key. Whether it will limit the arbitrage between the S&P 500 and the S&P600, or the Russell 1000 and the Russell 2000, depends, of course, on how widely institutional investors adopt the new indexes.
Still, some are already applauding, says The Wall Street Journal’s Jason Zweig:
“Each June, roughly 200 stocks are replaced in the Russell 2000 index of small companies. This June 25, the stocks that were added to the Russell 2000 outperformed those that were deleted by 2.3%, according to Investment Technology Group. Over the long run, sharp traders getting out in front of these forced portfolio changes have poached at least 0.38 percentage point of annual return away from Russell 2000 index funds, estimates a new study in the Journal of Empirical Finance.
“CRSP’s new family of indexes will tackle the poaching problem in several ways, says Lubos Pastor, a University of Chicago finance professor who helped design them. The boundaries between small, medium and large stocks will be set as proportions of the value of the total market, rather than as fixed dollar amounts or as an unchanging number of companies. Stocks will be ‘partially weighted,’ or shared, across different size indexes—so, as a company grows or shrinks, it doesn’t have to be added or eliminated in one fell swoop. Any additions or deletions will be made in ‘packets,’ or gradual steps over time, and the days on which the substitutions take effect will be randomized.”
An efficiency apartment, with fewer doors.