Tracking Risk of Exchange Traded Funds Revisited - A Multivariate Regression Approach

This empirical study investigates the ability of exchange-traded funds (ETFs) to replicate the risk-return characteristics of their respective benchmarks accurately. By decomposing ex-post tracking performance, this study finds that the commonly used measure, tracking error, rarely sufficiently explains the deviation from the benchmark and hence has very limited predicting power for assessing the tracking quality of an ETF. The results presented here clearly indicate that in many cases a linear metric is a more reliable predictor for future return deviations and that, therefore, the total cost of administrating an ETF provides a fairly good estimate of its tracking quality