Abstrakt
A critical aspect of trading Exchange-Traded Funds (ETFs) is the arbitrage trading strategy taken by authorized participants (APs) to keep ETF prices in line with their net asset values (NAVs). ETF arbitrage trading is a strategy that exploits the discrepancies between an ETF price and the value of the ETF's underlying assets.
In this study, I quantitatively examine the effect of ETF arbitrage on the underlying assets of an ETF. I develop a dynamic state-space model that jointly estimates the price dynamics of an ETF and its underlying assets by explicitly incorporating the ETF arbitrage.
The model is estimated individually for the Dow Jones Industrial Average ETF (DIA) and the VanEck Vectors Semiconductor ETF (SMH). The empirical results show that ETF liquidity shocks propagate to the underlying assets via the ETF arbitrage mechanism.
These ETF liquidity shocks add a permanent layer of transitory volatility to the underlying asset prices. I find that a unit of liquidity shock to DIA brings a range of 0.1% to 0.93% of extra volatility to the underlying assets of DIA.
Similarly, a unit of liquidity shock to SMH adds a range of 0.33% to 0.95% of additional volatility to the underlying assets. In addition, I show that it takes APs longer to correct deviations between the ETF price and its NAV.
It takes approximately 4 and 10 minutes for APs to perform the ETF arbitrage for DIA and SMH, respectively. Finally, the findings suggest that an ETF arbitrage transaction speeds up the price discovery process in the ETF markets.
There are approximately 74% and 67% variations in the premiums of DIA and SMH due to price discovery, respectively.