Scare stories about ETFs tend to be more qualitative than quantitative, whether warning about bubbles or Marxism.
But if there were to be quantitatively alarming articles about the growth of ETFs, they would probably focus on one of two trends.
One would be ETF failures-to-deliver: when a market participant does not settle a trade within the agreed timeframe. In 2016, ETFs accounted for 78% of all failures-to-deliver among US securities by dollar volume, despite the ETF market being only a fraction of the size of the equity or bond markets. That is also up from 29% in 2006, while common stock failures-to-deliver have declined in both absolute and relative terms.
The other frightening statistic would be that the aggregate dollar value of ETF short interest was higher than $150 billion at the end of 2016; that was equivalent to approximately 20% of all US equity short interest by market capitalization, even though ETFs comprise less than 10% of the full US equity market.
So, a lot of ETF transactions are not being settled on time and a lot of people are betting against ETFs. Could all this mean that the ETF industry is poised to implode?
A recently published study suggests that it is not, and that the rise in both failures-to-deliver and short interest may be linked in a way that is not systemically threatening.
The paper’s authors – Richard Evans of the University of Virginia and Villanova University’s Rabih Moussawi, Michael Pagano, and John Sedunov – note that ETF failures-to-deliver are associated with how their market makers create their shares.
Essentially, ETF market makers receiving heavy order flows for new shares seem to be selling the shares before they create them. This is not necessarily nefarious, and relates more to how the market maker manages liquidity. They are technically failures-to-deliver if the market maker does not create the shares within the standard three days, but they typically are created later; there is no industry-wide Ponzi scheme.
One side effect of this process, however, is that the shares that have been sold but not yet created are generally recorded as being short. This thus inflates the reported short interest. Evans, Moussawi, Pagano, and Sedunov refer to this practice as operational shorting.
The academics connect these two phenomena by demonstrating that there is a strong statistically and economically significant positive correlation between operational shorting and failures-to-deliver in ETFs.
‘Operational shorting is one of the most significant drivers of the ETF short interest, which suggests that – unlike common stock short interest – ETF short interest has multiple components, including directional shorting, hedging components, and operational shorting components that have different determinants and perhaps diverging implications,’ the researchers summarised.
‘This is in contrast to the alternative hypothesis that these short sales are directionally driven by informed speculators that think the ETF is overvalued and thus are betting on a drop in the ETF’s value.’
Indeed, Evans, Moussawi, Pagano, and Sedunov found that operational shorting – unlike other measures of short-selling demand for an ETF, such as genuine short interest or lending fees, which were negatively related to future returns – was unrelated to ETFs’ future performance.
Should any ETF sceptics point to failures-to-deliver or short interest as danger signals, then, investors should remember that they are simply a factor of the market’s normal operation.
In fact, they may even be positive for the markets because operational shorting and technical failures-to-deliver permit market makers to meet ETF demand and provide liquidity without affecting the trading of the underlying stocks.
‘Our evidence suggests that operational shorting serves as a buffer that hinders the transmission of the liquidity shocks that continuously hit ETF markets because of the higher-frequency clientele that are attracted to ETFs,’ argued Evans, Moussawi, Pagano, and Sedunov.