From Chris Metli of Morgan Stanley
It’s easy to become numb to the low volatility environment and the risks it presents. While trying to pick a trough in vol has been a fool’s errand, focusing on the risks resulting from vol being so low is not. Low volatility has produced a regime where the risks are asymmetric and negatively convex, so being prepared for an unwind is critical. This is not a call that vol is about to spike, but you need a plan if it does.
This note details how a short vol unwind might develop. A violent rise in volatility could be driven by just a 3% to 4% one-day S&P 500 selloff. Right now the risk is greatest in the VIX complex, and demand for VIX futures from three main sources could result in 100,000 contracts ($100mm vega) to buy in a down 3.5% SPX move. For context VIX futures ADV over the last year is 230,000 (although has risen to as high as 700,000 in big selloffs).
It’s important to note that this only happens if there is a large 1-day move lower in equities starting when VIX is very low – a slower drawdown, or a selloff from higher starting levels of vol, would not create as much demand. The biggest S&P 500 selloff when VIX was less than 12 was 3.5% (Feb 2007), so this type of move would be on par with the worst-case historical move for a low vol environment.
Why highlight this now? Simply because as volatility goes lower, these risks rise. In April and May QDS acknowledged that the short vol base was large, but viewed the risk as manageable (‘Keep Calm and Carry On’). In June the team’s stance on volatility turned neutral. And since then volatility levels have only gone lower.
What happens if the S&P 500 were to fall 3.5% today?
1) First, the VIX could rise as much as 12 points. When volatility is low it tends to move a lot for a given change in the S&P 500. That effect is likely to be exacerbated now because a) skew is steep (and VIX rolls up the skew in a selloff) and b) many players in the VIX market are short. Taking these dynamics into account QDS estimates VIX could rise ~12 points for a 3.5% 1-day decline in SPX.
If VIX rises 12 points, 1-month VIX futures are likely up 5.5 points, a ~50% increase. The 1-day percentage change is a big deal in the VIX complex because the levered and inverse VIX ETFs and ETNs rebalance daily based on the percentage change, and some of the thresholds for forced unwinds are based on the percentage change. This is why lower vol creates higher risk.
2) In a 50% increase in VIX futures, the levered and inverse VIX ETFs and ETNs need to buy ~70,000 VIX futures to rebalance their portfolios and maintain target exposures (this estimate is net of redemptions – long vol ETPs are generally sold by their holders as vol rises, offsetting the levered rebalance). While these flows likely occur near the close, the dynamic is well known, and many traders will bring forward those flows to the middle of the day.
3) A VIX futures level in the high teens (up from 11 – 12 now) means dealers get short VIX call gamma. There has been considerable buying of VIX calls and call spreads, with much of the hedging flow in the last month focused on VIX (instead of SPX). As VIX futures rise, dealers will get more and more short delta, which needs to be hedged by buying VIX futures. In a 3.5% SPX selloff QDS estimates there could be 25,000 VIX futures to buy from dealers hedging.
4) If VIX futures approach +100% in a single day, there is a risk that the providers of inverse VIX ETPs cover the VIX futures that they sold to hedge the products. This is because there is a mismatch in the hedge if VIX futures rise more than 100% – the inverse ETPs can’t go below zero (-100%) but the loss on a short VIX futures position can be more than -100%.
There are two inverse ETPs that sell the front of the VIX futures curve – XIV (an ETN) and SVXY (an ETF). For XIV (holding ~73,000 contracts short) the prospectus indicates that it will unwind if the NAV falls more than 80% intraday, with investors receiving the end of day value. Given this is a known threshold, anything close to a +80% move in VIX futures would likely trigger buying (by the ETN provider and/or market participants) in anticipation of the unwind. Note that because XIV is an ETN, investors receive the theoretical value of the index based on its rules, not what the provider actually trades.
SVXY (holding ~37,000 contracts short) does not have a set threshold to unwind according to its prospectus. That said VIX futures currently have a margin requirement of ~45% of notional for the average of the front two contracts, and any decline in value of the inverse ETPs to those levels could trigger a rapid forced unwind. Note that SVXY is an ETF, so the NAV is based on the actual holdings of the fund at the end of the day.
5) The 2nd derivative impacts are likely large. An overnight gap higher that doesn’t give investors the opportunity to hedge is the worst case. Consider if there is an overnight gap in VIX futures of +150% (VIX futures to ~29, VIX to 35+):
- The holders of the inverse ETPs lose the $1.4bn as the AUM of inverse ETPs goes to zero.
- The providers (hedge counterparties / clearers) of the ETPs lose $600mm due to the mismatched hedge if VIX futures more than double.
- Investors that sold long vol ETPs against short vol ETPs (a somewhat common carry trade) have the same unhedged gap risk in a +100% VIX futures move as the ETP providers. Assuming they are 20% of the shorts in the inverse ETPs (a guess) – they lose $250mm.
- Dealers who can’t hedge their delta on the way up could lose $500mm on our estimates.
- Hedge funds who are short VIX futures ($250mm vega on just the short leg per CFTC) playing the rolldown trade lose over $4bn.
- Investors who are wrong way in VXX, SVXY, and UVXY options could lose hundreds of millions – estimating loss here is hard, but assuming 20% of the open interest is wrong way, the loss would be ~$1bn.
- Investors who have sold vol in other forms (options, variance, etc.) would take losses and likely look to cover as well.
With a buyer for every seller someone is making this money too, and some of the above could be hedged as well. But the point is that when there are losses, ‘sell what you can’ will take over and drive further supply. While the point of max pain in volatility would likely be the first day of the spike, the knock on effects could mean equity markets take longer to recover.
6) Adding to the pain – on days after the initial shock – would be the flow from annuity and risk parity deleveraging. Both of those investors are slow by comparison to the VIX market – annuities will sell over several days, starting the day after a selloff. Risk parity funds are more discretionary, and the supply could come over a matter of weeks. But given high leverage resulting from the low vol environment, their potential supply is large and could prolong any downturn.
Investors have been crying wolf about the VIX complex for years, and have been wrong so far. And it’s important to note that the odds are still heavily stacked against the above scenario playing out and the most likely scenario is still a graceful unwind of the short vol trade:
- If volatility is just a little bit higher, the unwind potential is much less – there needs to be a shock when volatility starts at these very low levels
- The unwind in VIX only happens in a 1-day gap lower in stocks – a slow bleed would not create as much supply
- History suggests a gap from low vol levels is unlikely: the biggest selloff in S&P 500 when VIX was less than 12 was -3.5%, and -2.2% when VIX was less than 11, not enough to trigger this type of unwind. That -2.2% selloff occurred on Feb 4th 1994 when the Fed raised interest rates – bond volatility remains the major risk factor.
- Investors are still not all-in on stocks, with exposures moderate and many hiding out in defensives and Tech – raising the bar for a big selloff in stocks
- Active manager performance this year has been strong, meaning funds are less likely to become forced sellers of positions, which helps keeps volatility tame and can limit the speed of a selloff
- Correlation remains low due to both fundamentals and positioning, and for the index to sell off sharply it would need to rise
The point is simply that if there is an external market shock that nobody is prepared for (and this likely coincides with a selloff across asset classes), the risks of a quick unwind are higher than in the past. QDS favors staying long equities, but does not view the risk / reward on simply selling volatility as attractive anymore. Instead consider:
- Replacing long stock with S&P 500 upside calls that look very cheap given low volatility – buy the SPX Dec 2550 call (30^) for ~1% (sub-9% implied vol)
- Buying VIX puts instead of selling VIX futures to collect rolldown – buy the VIX Sept 10.5 put for $0.25, which offers attractive leverage if futures roll down to current spot levels of VIX with a 9 handle.
- Hedging this potential tail event with OTM VIX calls – buy the Sept 20 calls (17^) for $0.45. VIX calls are not cheap by any measure, but they are reasonably priced given these potential risks, and for those that see a shock occurring in the next few months VIX calls are the best hedge.