International tracking error and portfolio diversification using mutual funds and ETFs

Current practice and research in portfolio asset allocation employs necessary assumptions about which securities and asset classes are held in particular proportions in order to arrive at an “optimal” mix of investments.   These studies typically employ the average performance in the input process, yet it is unlikely that investors hold average funds, and that the specific funds they own exhibit tracking error. 

We examine the influence of tracking error on performance and diversification, for a representative calendar year, 2013.  We employ a new version of the Morningstar Mutual Fund Database.  Although the primary contribution of the study is an examination of relative diversification potential of mutual funds and ETFs, we also find striking differences in performance and risk of ETFs vis-à-vis traditional mutual funds.  Mutual funds outperformed ETFs by over three hundred basis points on average, and exhibited a much lower standard deviation.  When examining common subcategories and more sophisticated performance metrics such as alpha, Sharpe ratio, and Sortino ratio, the results hold.  As expected, ETF expenses are about half of that for traditional mutual funds.  But it takes roughly double the number of ETFs to reduce tracking error to the same level as mutual funds.  Finally, when we analyze the determinants of tracking error for traditional funds and ETFs, we find mostly intuitive results – that is, tracking error is driven by turnover, holdings, leverage, fund size, and whether it is an enhanced index fund.  We also find that tracking error is determined by manager experience, especially for ETFs when tracking error is measured more carefully.  To our knowledge, this is the first study that provides a comprehensive comparative analysis of returns, risk, and diversification potential of traditional mutual funds vs. ETFs.