The determinants of ETF liquidity: Theory and evidence from European markets

Despite the importance ETFshave recently gained, little is known about their liquidity. The conventional view on ETFliquidity is that what really matters is neither the size of the ETFnor its trading volume but the liquidity of its benchmark index. We argue that while creation/redemption effectively creates a tight link between the ETFand the index liquidity, other factors are likely to affect the former. We develop a traditional inventory 2-period model where a risk averse ETF market makerliquidates her ETFinventory. She then bears illiquidity costs when closing out her stock basket position. One first result is that ETFsreplicating illiquid indices should trade with higher spreads. Second, with risk-averse market makers, our model also predicts that ETFspreads should be positively correlated with the underlying index volatility. Third, we also show that the maker maker will act so as to minimize her expected inventory at the end of the day. This will be easier on more heavily traded ETFs, that should thus exhibit tighter spreads. We then provide empirical evidence of the determinants of the spreads in the European equity ETFmarkets from their inception in 2000 to the end of 2012. We find that our theoretical predictions are confirmed by the data as the spread appears to positively depend on variables related to the inventory risk (index volatility, low turnover, funding liquiditycost, currency risk). The stock basket spread does not affect ETFspread in the whole sample, as would be expected, but still does so for the 20% ETFswith the lowest daily trading volumes. Collectively, these findings suggest that ETFsare dealt as stocks by market makerswhen the trading volume is sufficiently high to manage the inventory with low risk. It is only when trading volume is too low that market makershave to account for the illiquidity of the underlying stock basket in their quoted spread .