Most ETFs track an index instead of employing a professional manager. and inevitably there is a lag between the performance of the index and the funds net asset value. Part of this has to do with the fact that all ETFs charge investors an annual fee to recoup the costs of running the fund, something that is not reflected in the performance of the index. Yet there is still a degree of divergence even after the effect of expenses is discounted known as tracking error.

It would be nice to think that as the bugs get worked out of the system, tracking error would shrink, but unfortunately that is not the case: A study published this summer by Morgan Stanley found that tracking error across all U.S.-listed ETFs averaged 125 basis points in 2009, compared with about 50 basis points in 2008. Even worse, the study found a broader range and magnitude of tracking error in 2009 than 2008, with 37.5% of ETFs suffering from tracking error larger than 100 basis points compared with just 13.5% in 2008; at the same time, the number of ETFs reporting tracking error of less than 25 basis points declined from 44.6% to 22%.

There are several ways that tracking error can creep into an ETF. Some ETFs use a methodology of full replication, in which all the index constituents are owned in their proper proportion, but full replication would be too costly for other indexes, and managers opt instead to use representative sampling to achieve a similar result. While this usually works pretty well, it can lead to a divergence if the performance of index members that were left out varies from the others.

Index members also change, and it is not always possible for the portfolio to match each change simultaneously. Furthermore, there are some ETFs that track indices whose weightings conflict with the portfolio diversity requirements mandated by the Securities and Exchange Commission. (The SEC prohibits funds from owning portfolios in which one issuer accounts for more than 25% of total assets, or in which the sum of issuers with portfolio weights of more than 5% exceeds 50% of the funds assets.) For equity funds, how the fund handles dividends can lead to tracking error, as well.

As a group, international and global ETFs tend to have the largest tracking errors. the study said, because the underlying markets are more difficult to access, which encourages portfolio managers to use representative sampling rather than full replication. Emerging market ETFs were a particular problem, accounting for seven of the nine international ETFs with the largest tracking error and trailing their target index by an average of 836 basis points.

At a minimum, it is important to know whether an ETF is managed using representative sampling. and if so, how many holdings it has compared to its target index. International ETFs and fixed income ETFs that target less liquid markets like high yield bonds should receive extra scrutiny. Better still is to research the tracking error of a fund, and learn what has caused the error. And for those who cant tolerate tracking error at all, there are of course exchange-traded notes, which dont own the underlying asset at all and are simply obligations of the issuer. In essence, ETN investors drop the risk of tracking error in exchange for assuming the credit risk of the ETN issuer.

S&P Senior Financial Writer Vaughan Scully can be reached at Vaughan_Scully@standardandpoors.com. Send him your ideas for ETF story topics.