Is tracking an index worth the cost?

By Brad Steiman | November 12, 2013 | Last updated on November 12, 2013
4 min read

Many see traditional index funds and ETFs as efficient vehicles for gaining diversified exposure to specific asset classes. But are indices precise representations of the underlying asset class? If not, it may not be worth incurring the cost of tracking them perfectly.

Let’s examine returns across indices that seek to represent the same asset class. The proliferation of index vendors has resulted in many indices for the same (or similar) asset classes, each with slightly different definitions and reconstitution (or rebalancing) schedules. Tables 1 and 2 (below) show compounded returns, tracking error and correlations among indices for U.S. large-cap value and U.S. small-cap value from January 1999 to June 2013. The time frame is based on data availability for the selected indices.

Is tracking an index worth the cost? charts 1 and 2

The compounded return differentials and tracking error numbers are surprisingly high. U.S. largecap value index annualized performance varies between 4.31% and 5.47%, while tracking error between the different indices typically ranges between 2% and 3%, with only two results below 2%. For U.S. small-cap value indices, the compounded return differentials and tracking error numbers are even higher. U.S. small cap value index annualized performance varies between 8.53% and 10%. The minimum tracking error is 2.48% and the maximum 6.13%.

The dispersion in index returns stems from small differences in security weights, a result of subtle differences in methodology and reconstitution dates.

Random performance fluctuations

The construction rules governing various indices do not represent any sort of forecasting or timing ability intended to improve returns. So when one index outperforms the others in the same asset class over a month, year or longer, the result is likely period-specific and purely random.

Tables 3 and 4 (below) shows the maximum over- and underperformance among these indices for each of the 12-month periods between January 1999 and June 2013. The differences can be large and would make many investors uneasy. Over time, the winners and losers tend to swap places.’

Is tracking an index worth the cost? charts 3 and 4

What are the costs?

Since each index targets the desired asset class in a similar fashion, with no evident expected return benefit attached to choosing one over another, picking an index to track is likely to be subjective. Where does this leave investors trying to capture the return of a specific asset class?

The arbitrary nature of this decision, and the fact that selecting any specific index will produce tracking error relative to the others, suggests trying to achieve perfect replication should not be an objective. In fact, there is a high cost to near-perfect tracking.

These costs are incurred in many ways. For instance, when indices are reconstituted, managers may have to trade specific securities within a narrow period of time. Trading during reconstitution periods can be especially disadvantageous. Enough liquidity has to be incentivized—typically requiring a shift in security prices—so all managers tracking that particular index can trade the same securities in the same direction at effectively the same time.

Demand for alternatives grows

Demand for alternatives to traditional equity indexing has been growing, as investors seek cost-effective ways to outperform benchmark indexes, SEI’s latest commentary finds. Lowering the volatility and periodic uncertainty of single-factor portfolio returns can enhance returns with lower tracking error, plus there’s a better chance of avoiding periods of prolonged underperformance. As a result, SEI expects the industry to trend in this direction in the coming years.

A more subtle cost is potential style drift. Indices are reconstituted or rebalanced periodically, but stock prices constantly change. In the months or weeks between reconstitutions, an index may include securities that have experienced significant price movements and no longer belong in the index based on its own specification and methodology.

Despite the continuous changes occurring in the market, index fund managers seeking to minimize tracking error must hang on to (and invest new money from cash flows into) securities that have migrated out of the index until the next reconstitution date. This constraint can cause meaningful style drift.

The objective of all of these factors— no rebalancing for months, reinvestment of cash flows in the original set of securities at specific weights, and synchronized rebalancing across managers at specific dates—is to achieve a daily return as close as possible to the published return of the chosen index.

But how does this fit with the original objective of earning the return of the desired asset class in the most efficient manner possible? Are these two objectives compatible? Early on, indices were benchmarks for evaluating a fund manager’s risk-adjusted performance. But the use of indices has shifted. In some cases, index investors are letting the arbitrary composition of a commercial benchmark drive their investment strategies.

A superior investment approach should offer consistent and diversified exposure to an asset class without incurring the potentially high cost of trading.

By removing the low tracking error requirement, this type of strategy could effectively avoid trading during reconstitution periods and focus on capturing the returns of the broad asset class. In fact, by allowing some tracking error and diversifying across many names, one could reduce costs and even seek out opportunities to add value by providing liquidity to the indexers who are forced to trade during the narrow time window of reconstitution. Focusing on reducing tracking error can be costly and there’s no evidence that incurring those costs has meaningful benefits.

Brad Steiman is director, head of Canadian Financial Advisor Services, and vice president of Dimensional Fund Advisors Canada ULC.

Brad Steiman