JultikaUniversity of Oulu repository

Exchange Traded Funds (ETFs) should trade at a price equal to their fundamental Net Asset Value (NAV). However, ETFs’ can occasionally pose economically significant premiums/discounts to their NAV prices, i.e. arbitrage opportunities. The theoretical part focuses on ETF arbitrage and explains why this arbitrage trading is attractive to high-frequency traders (HFTs). In the empirical part, we introduce HFT activity proxies to a factor model explaining the observed SPDR trust (SPY) premiums during 2.1.2002–15.1.2013. A range of statistical and econometrical tools are then employed to study the detailed relationship between these factors and the SPY premiums. In addition, we replicate a popular method used to study HFTs’ effects on stock markets, and apply it to analyze HFTs’ effects on ETF pricing process. By utilizing an exogenous technology shock (implementation of Regulation National Market System) which improved U.S. market infrastructure, we should be able to dissect the effects caused by heightened HFT activity. The absolute size of SPY premium is significantly related to endogenous ETF factors. The exogenous factors serving as proxies for available arbitrage capital improve the explanatory power. The Reg. NMS implementation fails to serve as an exhaustive structural break point in ETF pricing dynamics. Although, the post-Reg. NMS era has very low ETF premiums and higher trading volumes. This can indicate that Reg. NMS made markets more suitable for HFT, and therefore high-frequency ETF arbitrage might have been more efficient during the post-Reg. NMS era. Simple implications are: ETF premiums can be significant in relation to their annual expense ratios and investors can improve their trade execution by understanding the drivers behind ETF premiums. ETF premium volatility modeling can also be useful in risk management and in investment decision making. Understanding HFTs’ role in ETF mispricing adds to our incomplete knowledge on the effects of individual HFT strategies.