When they go low, we go high (to Short Vol)

Malt Liquidity 91

The Mismatch will Always Exist

The goal of an ETF is purportedly to provide easy-to-access, liquid exposure to an underlying basket that would otherwise be too painstaking to construct. A good chunk of the time, the ETF is more liquid than its underlying:

…it’s intuitive that a sector ETF with a certain % exposure to each stock could have deeper liquidity than in the underlying — while AAPL is certainly the most liquid stock in the market, it is inarguable that the exposure SPY gives to stock #392 in the index is much more liquid than #392 itself.

This liquidity mismatch is more pronounced the less liquid/accessible the underlying is — much like a Nate Diaz/Khamzat Chimaev matchup, the liquidity in, say, corporate bonds on an intraday-trading basis is not comparable to a bond ETF. On a regular day, this is not a big deal — liquidity provision is plentiful when you don’t really need it, spreads will be tight, and the creation/redemption process will flow smoothly. However, if you demand liquidity, you have to pay for it, and most notably, volatility and spread tightness are inversely correlated. Quote providers are not the ones getting hit when real flow demands liquidity and book depth falters (which is why you always want to be careful about leaving resting orders.) Who absorbs the cost of liquidity when the underlying is trading in a much more illiquid fashion than the shares of the ETF that mimics the exposure? Well, we have an article exploring this:

…academics from a trio of US business schools suggest that ETFs’ role is not always benign, and during market dislocations can actually worsen the state of the underlying market. The potential problem stems from the creation and redemption baskets that ETF issuers trade with market makers known as authorised participants (APs) in order to handle inflows to, or outflows from, their ETFs. Unlike equity ETFs, bond funds’ creation and redemptions baskets typically do not include every bond in the index they are tracking, as this can encapsulate hundreds, or even thousands, of separate issues…

…during a crisis, when many investors are rushing for the door, redemptions hugely outweigh creations. In that scenario, if a bond is included in the basket, the APs “may then become reluctant to purchase more of the same bonds, reducing their liquidity”, the paper said.

Certainly, this is related to a core issue of the ETF due to the disclosure/predictable nature of the prospectus, which inevitably leads to trading opportunities in volatile environments when they’re predictably trading the underlying to replicate the product. Take the case of USO, which should have knocked out due to the frontrunning of its rollover schedule, but which was essentially allowed to totally violate its prospectus. In times of actual volatility, these ETFs necessarily change into shrouded (read:active) rebalancing protocols:

USO delivers its exposure to oil using near-term futures. USO gets exposure to oil using derivatives, like several oil ETPs. The fund predominately holds near-month-futures contracts on WTI, rolling into future contracts every month. This method is particularly sensitive to short-term changes in spot prices. USO held front month contracts until April 17, 2020, at which time following leeway in the prospectus, USO changed the exposure from holding specifically front-month contracts to holding predominantly front-month contracts, 30% next month and 15% contracts with further expiry.

Hmm, I wonder what happened the week of April 17, 2020?

U.S. crude oil futures collapsed below $0 on Monday for the first time in history, amid a coronavirus-induced supply glut, ending the day at a stunning minus $37.63 a barrel as desperate traders paid to get rid of oil.

Of course, the more infamous frontrunning of an ETP mimicking exposure was XIV, but catch me in a bar to rant about that one.

Intuitively, an ETF’s occasional diverging from its NAV makes sense — while in theory a basket is priced by the weighting of its components, in practice, an ETF can only stay in line relative to its no-arbitrage bounds. Put simply, there has to be sufficient liquidity in each side of the trade for the arbitrage to be profitable enough to maintain. In this sense, I think the academics get it backwards — it’s not the ETF dislocating the underlying, but rather the lack of liquidity in the underlying leaving the designated market makers/authorized participants with the choice of either unwanted exposure or tightening liquidity in the underlying to maintain the NAV. This is pointed out by the actual market makers quoted in the article:

…Izzo argued that the causality ran in the opposite direction — it is because some bonds are illiquid that they increasingly feature in redemption baskets as sell-offs intensify, not vice versa.

“The traditional approach across the entire fixed-income industry to a bond crisis is to do nothing. The default behaviour is that everyone — outside the ETF creates and redeems — just turns their chair, looks away and says ‘we will wait it out’. “If the bonds don’t trade then you don’t have to write them down. For illiquid bonds, you can’t even find a bid or an offer. “ETFs have brought liquidity into the market at times of crisis where it has never existed before,” added Izzo, who said that during the Covid crash GHCO bought bond ETFs at a 20 per cent markdown and held them on its books, a strategy that ultimately proved highly profitable.

Yeah, this is disingenuous — the ETF isn’t bringing liquidity to the underlying, but rather the short vol trade on the underlying. By holding the ETF without hedging exposure/taking the arbitrage by trading the illiquid underlying as the ETF drifts away from its NAV, you’re essentially betting on reversion when quotes come in again. While the direct vol on the bond is not shortable, this is essentially a spread collection trade between the underlying and the ETF’s rebalancing, which is short vol. If, in an actual crisis, the liquidity in the underlying was demanded at a very high rate, this trade would be blown out. I don’t think there is an issue with ETF tracking in general — I think the issue arises more when a majority of money becomes passively invested in ETFs rather than allocated to the underlying.

A simple thought experiment — what happens if the market is entirely “passively invested” (reminder that index ETFs are not passive) and liquidity is demanded in the underlying? Liquidity is not created out of thin air — if you want it, you must pay for it. As long as there is both sufficient liquidity and a properly risk-adjusted reward, someone will tighten the spreads and bring the products in line. If the risk is too great, however, the quotes will never come in due to risk of ruin — in the scenario presented above, everything would go to zero. You need real flow on both sides, else what exactly are you tracking? It’s the core problem with mark-to-market pricing in private portfolios — they simply don’t reflect either upside or downside vol properly due to lack of actual trading volumes. A theoretical price is only worth what you can actually get filled at.