Gus Sauter, Vanguard's recently retired CIO, argues that theory provides little guidance as to how investing, beyond holding the total market, is actually done (Sauter, 2002). His point, that managers define the opportunity set, is best articulated in his cardinal rule of indexing: An index must reflect the way that money managers actually invest.
Building the Ideal Index
The Center for Research in Security Prices (CRSP) at the University of Chicago Booth School of Business sought, with its new index products, to combine current academic thought and practice with Sauter's cardinal rule, while paying considerable attention to the material constraints faced by investors. The result is a family of indexes that are both theoretically justifiable and a practical representation of those securities in which a manager, subject to a related mandate, could invest.
Other authors have done a good job of describing important features of benchmarks: completeness, objectivity, investability, etc. The CRSP Indexes have all of these features; however, this article delves deeper. CRSP seeks to explain our index design process and present the indexes' mechanics in the context in which the solutions arose. We believe our process can be most easily digested by understanding our theoretically guided and empirically validated approach and the balances struck that make CRSP's indexes valuable.
CRSP's approach to index construction directly combines theory and empiricism:
Theoretically And Logically Guided
The CRSP Indexes aim to be "current." Company performance and valuation fluctuate with economic conditions, firm decisions and investor expectations. The result: Companies that looked cheap/expensive/small/big/U.S.-domiciled/liquid at one point likely will not remain so indefinitely. We could simply state that an index should be as close to current as possible, but that would have clear drawbacks in terms of turnover. A common industry compromise between being current and limiting turnover is to reconstitute indexes semiannually or annually. The CRSP Indexes' quarterly reconstitution places a relative premium on being current; we married reconstitution to a novel migration strategy that limits turnover. The resulting indexes reflect changes to the investment opportunity set quickly while keeping turnover low.
Free float is another example of a constraint derived theoretically/logically rather than empirically. It has become widely adopted by index providers because it makes sense. Shares that are not available for trading cannot possibly lie in the investment opportunity set. While a departure from pure cap-weighted indexes, free-float-adjusted indexes are a more appropriate representation of those assets that investment managers should consider in their decisions.
A robust way to set breakpoints is another theoretical problem. Breakpoints reflect the ability to discriminate between the opportunity set contained in one index and the next. Ideally, breakpoints would arise naturally along some dimension of investor concern. However, the dimension of investor concern itself may not be obvious.
We begin with a simple example: market cap. Academics and practitioners have long noted that equities with different market capitalizations display significant differences in average returns. In addition, small stocks and large stocks, as groups, have tended to move together. Domestic index providers have historically made the decision on cap-index membership in an ad hoc fashion by using counts of securities as proxies for market capitalization. For example, the S&P 500 Large-Cap Index contains 500 stocks, while the Russell 1000 Large Cap Index contains 1,000 stocks when reconstituted. However, most of us scratch our heads when asked why an investor should care whether a security is ranked 999 or 1,001 by size; it would seem hard to argue that such arbitrary cutoffs reflect genuine investor concerns. Furthermore, count-based indexes remain anchored at a point in time and always reflect the relationship that count had with market cap when the index was conceived. Since the number of listed securities fluctuates dramatically over time, as shown in Figure 1a, the economic significance of a given count-based index changes, too. Figure 1b shows that today's Russell 1000 Index represents a substantially larger proportion of the total market than it did 15 years ago. In contrast, CRSP's capitalization-based indexes use cumulative capitalization breakpoints, a solution common among other providers in the international space. Cumulative cap maps closely to more reasonable investor concerns and has the advantage of keeping the indexes current from an economic perspective.