Content area
The main objective of this dissertation is to analyze the tracking error of ExchangeTraded Funds (ETFs) in the United States (U.S.) and its main determinants.
We use a sample of 88 U.S. ETFs on U.S. equity indices, for the period 1996-2016, defined according to asset class, investment region, and category. The ETFs selected include several size and investment styles (Micro Cap, Small Cap, Mid Cap, Large Cap, Extended Market or Total Market).
The study of the tracking error and its determinants is an important subject because, when investing in ETFs, the tracking error is one of the most important factors, as investors expect to receive the same return as its underlying index. If this does not happen, the product does not fulfill its purpose and investors will not stay in the market. We start by analyzing the tracking error of the ETFs presented in our sample using five different methods and then we analyze the determinants of tracking errors. We conclude that ETFs track quite well their underlying indexes and that the average tracking error has been decreasing over the period. We also conclude that the variables size, expense ratio, liquidity, dividend yield, risk and the bull & bear period are statistically significant for any level of significance. We found that the variable size and the dummy variable bear produce a negative effect in the tracking error and the explanatory variables liquidity, risk and expense ratio produce a positive effect in the ETFs’ tracking error. In what concerns to the other statistically significant explanatory variable – dividend yield – we do not achieve consistent results because the sign of the parameters changes with the method used to estimate the tracking error.