The Index Is Dead. Long Live The Index.

June 24, 2013

The two most important costs of indexing are turnover and tracking error. Turnover is obvious; irrespective of cause, turnover has a direct dollar cost and eats into returns. The bulk of turnover occurs at reconstitution, when constituents move in or out of an index. To track the index, the indexer must access capital markets. Capital-market access comes with price impact, especially when the volume of a transaction is relatively large.

As one could imagine, turnover and price impact are different. An added dimension of our analysis looked at, for lack of a better name, "bad turnover." Bad turnover is that which demands transacting in volumes that are difficult for the market to meet and are thus expected to have greater price impact cost.

The CRSP Indexes also address turnover costs associated with front-running. Reconstitution is a fairly predictable process, and several leading indexes see front-running by active managers looking to trade securities transitioning into or out of the index ahead of passive managers who must make these trades. We randomize the pricing date during reconstitution, which alleviates front-running by making the inclusion/exclusion of marginal securities in a specific index harder to predict. The randomization itself follows a transparent algorithm—it should preclude manipulation of index membership without introducing any methodological opacity.

Tracking error, too, comes at a cost, albeit in a slightly less salient "risk" dimension. This can be thought of as the potential for variance in returns versus the instantaneous opportunity set. Active, benchmarked investors call this "active" risk, as it represents a decision to deviate from a naive position in the asset. Measuring the quality of investing decisions then becomes a question of the return-to-risk ratio.

In some sense, index providers make similar decisions to those made by active managers. The index obviously deviates from the instantaneous "true" opportunity set; the goal of the index provider is to strike a balance between the cost of turnover and risk from tracking error. Developing transparent, mechanical rules for banding and migration precludes a simple functional approach, but the idea pays homage to standard mean-variance optimization. To determine the appropriate mechanics, CRSP ran a large number of experiments that studied problems ranging from the symmetry of bands at breakpoints to the ideal band width to the use of thresholds versus continuous transitions to transitional packet size. We evaluated more than 40 different approaches before settling on our breakpoints, threshold bands and 50 percent packeting. We compared all approaches on several metrics: tracking errors versus appropriate Lipper and Morningstar manager indexes, tracking error versus "pure" (i.e., no migration strategy) indexes, aggregate turnover and bad turnover (Figures 6a and 6b).

Tying It All Together
We believe the CRSP Indexes represent a meaningful improvement in index usability. True to the cardinal rule, the indexes should prove to be a cost-effective approximation of the actual investing process.

For purely passive investors, the cost of implementing the CRSP Indexes is low. From an active investor perspective, CRSP's efforts to keep the indexes current and remove ad hoc constraints mean managers will no longer expend their risk budgets by allocating to securities that have recently changed category or by simply following their investment style.


Find your next ETF

Reset All