An Analysis of ETF Tracking Errors, Liquidity of Indices, and Correlation

Buetow and Henderson performed a study of 845 exchange-traded funds (ETFs) traded on U. S. exchanges. Using a tracking error formula, they tabulated the annual index returns and ETF tracking errors. They also developed a correlation analysis. 

They showed that "ETFs tracking less-liquid indices incurred larger absolute tracking errors." Therefore, the managers of ETFs track the benchmark indices as closely as possible and "performance appeared to be related to the liquidity of the securities composing the indices." The correlation specifics indicated that "most ETFs exhibited high correlation with their benchmark index." 

When an index was composed of non-U.S. securities, the correlations between the ETFs and the indexes were lower than when only U.S. securities were involved. The authors explained that for "ETFs that track benchmarks consisting of non-U.S. assets the nonsynchronicities due to time-zone differences between the index valuation and the close of the ETF trading may result in a higher correlation between the ETF returns and U.S. equity returns." 

They found that "ETF tracking error was correlated with equity returns, implying that the benefits of diversification are less than what is implied by the benchmark indices." An investor should exercise caution when the ETFs contain non-U.S. assets because of the time-zone differences. These ETFs have less correlation with their indexes. 

The ETF industry continues to generate new ETF products patterned on arcane indexes. One expects some of these products to exhibit liquidity problems, and such problems could restrain the returns. The authors declared that generally the investor does not achieve as much diversification with the ETF as delivered by the index upon which the ETF is based. 

Caution flags are waved for non-U.S. assets within ETFs as well as for specialized ETFs which can exhibit liquidity problems.

Source: The Journal of Portfolio Management, Summer (2012), 112-127.