Why ETF Tracking Error Occurs (And How To Avoid It)

by on March 22, 2011 | ETFs Mentioned:

As the ETF industry has grown by leaps and bounds in recent years, advisors are finding themselves with more options than ever before. While choices for achieving certain types of exposure are limited–VNM, for example, is the only pure play Vietnam ETF on the market–most asset classes present investors with various options. In many cases, there are multiple ETFs linked to the same benchmark. There are now more than 30 ETFs in the Large Cap Blend Equities ETFdb Category, three of which are linked to the S&P 500 Index.

Once the universe of 1,100+ ETFs has been narrowed down to those aligning with a specific investment thesis, advisors evaluate potential ETF investments on a number of different criteria to identify the most attractive option [see the Free ETF Screener]. Expenses are one common comparative metric, and weighting methodologies, level of diversification, and fund structure are other points that can be compared head-to-head. Another characteristic of ETFs is tracking error, the delta between the performance of an ETF and the change in the underlying index. While replicating the performance of an index may seem like a relatively simple process, there are a number of ways for tracking error to creep into ETFs. While avoiding tracking error completely is a tough task, there are a number of steps that can be taken to avoid this phenomenon [for more ETF insights, sign up for our free ETF newsletter]:

Expenses

Perhaps the simplest source of tracking error is the expense ratio of an ETF; funds operating at full efficiency can be expected to trail their underlying index by a margin equal to the bottom line fees charged. Indexes are hypothetical measures of performance that are not directly investable–and as such there is no associated management fee or other expense components.

This component of tracking error is also the easiest to avoid–or at least to minimize. All else being equal, the lower the expense ratio of an ETF, the lower its downside tracking error will be. If EEM and VWO were to perfectly replicate the MSCI Emerging Markets Index at all times, the Vanguard fund would be expected to show annual tracking error of about 22 basis points while the iShares fund would differ from the benchmark by 69 basis points [see 25 Cheapest ETFs].

Sampling Strategies

Some investors assume that in order to replicate an underlying index, ETFs buy each index component in a proportion equal to its weighting in the benchmark. While this “full replication” strategy is used by many ETFs, others employ “sampling” techniques that involve constructing a portfolio that is a sub-set of the index but maintains a comparable risk/return profile. Disconnects between the composition of the underlying index and the ETF portfolio can obviously result in tracking error, especially it the returns between funds excluded from the ETF and those represented vary considerably.

It isn’t possible to invest directly in an index, but for all practical purposes the ETF boom has transformed indexes from hypothetical benchmarks to investable assets. And it’s important to keep in mind that many indexes weren’t designed to be investable assets, but rather measures of asset performance maintained electronically. As a result full replication of certain broad-based benchmarks may present logistical and expense challenges for ETF issuers, and sampling strategies are often needed to construct a liquid and cost-efficient product [ETF Investors Are Embracing Low Cost Options... Or Are They?].

The Barclays Capital U.S. Aggregate Bond Index, which serves as the basis for a handful of funds in the Total Bond Market ETFdb Category, is a great example. The index consists of more than 8,000 individual securities, many of which trade infrequently. AGG has about 700 individual holdings, while State Street’s LAG holds about 400 bonds. Vanguard’s BND has more than 4,700 individual holdings–or about 60% of the number of bonds in the underlying index.

In the fixed income space, sampling strategies may be more effective because managers are able to match effective durations and credit qualities of a large pool of bonds using just a portion of the underlying securities. Sampling can be more challenging in the equity arena where the price drivers of individual stocks aren’t nearly as uniform [Study: ETFs Are Honing In On Their Benchmarks].

Rebalancing

While replicating an index is generally a passive process, all ETFs require manager oversight at certain points. Indexes are reconstituted from time to time, with new securities coming in and others going out. Because the index itself is a hypothetical basket of securities, reconstituting or rebalancing is a simple flip of the switch. ETFs, on the other hand, must actually go out into the market and complete trades. To the extent that a fund isn’t able to execute at the identical time and at the same price as the index does, tracking error may arise [see Myths About ETF Liquidity].

Rebalancing can be the result of a number of occurrences. Last year Israel was upgraded from “emerging” to “developed” status by MSCI, forcing funds like EEM to sell Israeli stocks and those such as VEA and EFA to buy them. Other examples may include a reshuffling of the components of the Dow Jones Industrial Average or S&P 500 or a need to reallocate weightings as a result of divergent performances from individual securities (more on this below).

Avoiding tracking error associated with portfolio rebalancing can be tough to accomplish with absolute certainty, since this source depends on the performance of the fund manager throughout the course of the year. But there are a couple of options for investors looking to reduce the likelihood of rebalancing-related tracking error.

1. Embrace Market Cap Weighting

The first is to focus on ETFs linked to indexes that require minimal rebalancing–specifically, market capitalization-weighted products. Because cap-weighted indexes give the largest weighting to the most valuable companies, the fund should closely approximate the index even as individual securities change in value.

Contrast that to an equal-weighted index that will regularly rebalance to give an equal allocation to each component security. Because certain individual components will perform better than others, a fund will deviate from its equal-weighted baseline strategy almost immediately after it is established. There are certain potential disadvantages to market capitalization weighting–the strategy has a tendency to overweight overvalued stocks and underweight undervalued stocks–but one advantage is the simplicity of the rebalancing process [Why Weighting Methodologies Matter When Choosing An ETF].

2. Use ETNs

Another tool for eliminating tracking error comes in the form of an exchange-traded note. Like ETFs, ETNs offer exposure to an index. But unlike ETFs, ETNs are debt securities that have no underlying basket of securities. Rather, the indicative value of the ETN is based on the performance of a specified benchmark. So much like an index, an ETN simply flips a switch when underlying securities are bought, sold, or rebalanced; there is no trading activity and no commissions incurred in the process [see The Tax Advantages Of ETNs].

3. UIT vs. ETF

For investors who don’t feel comfortable with the credit risks ETNs entail, there are subtle nuances between the structures of various exchange-traded products that can impact tracking error as well. The Unit Investment Trust (UIT) structure was common during the early days of the ETF industry, and is more restrictive in certain respects than true 1940 Act ETFs. UITs such as SPY must fully replicate their underlying index, and are restricted from lending out securities that make up their portfolio. That means that tracking error will be limited compared to products that have the latitude to reinvest dividends or lend out securities to make additional income [see SPY vs. IVV: S&P 500 ETF Options].

As such, UITs may have appeal to investors implementing somewhat complex strategies that include options, as the replication of the related benchmark will be exact [the ETF Screener features the ability to screen ETPs by structure type].

Disclosure: No positions at time of writing.