A performance analysis of long-short exchange traded funds

Abstract

Exchange traded funds (ETFs) have seen their popularity increase over the past fifteen years. This increase in popularity encouraged the creation of new kinds of ETF such as leveraged ETFs, inverse ETFs and more recently long-short ETFs. Long-short ETFs are designed to mimic the long-short strategy typical of hedge funds. In this master thesis I propose a performance analysis of long-short ETFs under the form of a comparison with hedge funds. I study the returns of 27 U.S. long-short ETFs from the 02-03-15 to the 22-02-2017. I focus on the premium of long-short ETFs relative to their net asset value (NAV), their beta and alpha when compared to the HFRX Equity Hedge Index and their market timing ability. I show that long-short ETFs are trading at a relatively high premium and that this premium has an impact on the performance modelling of long-short ETFs. The explanatory power of the regression of the long-short ETF market price returns on the S&P 500 index is very low but it increases when I take into account the NAV returns instead of the market price returns. This premium seems to come from the illiquidity of the underlying basket of stocks, the liquidity of the ETFs themselves and stale pricing. Using the regression of the long-short ETF NAV returns on the HFRX Equity Hedge index, I show that long-short ETFs propose a good replication of the hedge funds long-short strategy. But even though they seem to well replicate the long-short strategy, long-short ETFs underperform the HFRX Equity Hedge index. I also show that long-short ETFs do not seem to be correlated with common risk factors. Using the Treynor and Mazuy framework (1966) I show that long-short ETFs exhibit market timing ability but are still less nonlinear than hedge funds.