Why One Bank Is Worried That Credit ETFs Will Blow Up Next

Two weeks after the unprecedented, record Feb. 5 surge in the VIX, the historic move is rapidly fading into memory as "the dip gets bought" again: to those who correctly predicted it, like Fasanara Capital  and Eric Peters,  and those who profited from it, like Ibex Investors  and Peter Thiel, congratulations.

Now the question is which product, read ETF, will generate returns similar to the 94% bonanza reaped by those who were short the XIV ahead of its terminal implosion (the XIV will cease trading after Feb 20, with holders receiving a cash payment of its residual value, which is virtually nothing).

To be sure, numerous candidates have been proposed over the past two weeks, with two names frequently cited - certainly by this site among others - include the two big junk ETFs, HYG and JNK, whose existence is only assured as long as the structure and liquidity of the underlying cash junk bond market isn't seriously tested, as the ETFs would spontaneously implode once the market of underlying junk bonds freezes up, something which Howard Marks has repeatedly warned about.

What is strange - as we discussed on several occasions  over the past 2 weeks - is that credit fared relatively well in the washout from the VIXplosion, widening by markedly less than various volatility models would have suggested; according to Deutsche Bank cross asset strategists, on its own the VIX spike should have been worth 56bp to IG spreads and 156bp to HY if only on a theoretical, "model basis."

That spreads are lagging is not altogether that surprising looking at historical examples: as Deutsche notes, credit spreads widened much less than expected on short-lived increases in volatility in the past (notably in 2010, 2011, and 2015), but the primary risk is that the longer vol remains elevated, the more likely it is that credit will have to reprice.

Indeed spreads have started to widen somewhat, and the longer that vol remains elevated, the more they should continue to do so, and in the event of another leg higher in volatility, credit would appear to be the logical nexus of risk.

* * *

Which brings us to the key question: what could catalyze a collapse of credit ETF? In responding to this inquiry, Deutsche Bank's Dominic Konstam writes the following:

Our colleagues in equity derivatives had discussed the amount of vega that short vol funds would have to buy following an increase in VIX, which effectively helped to magnify a justifiable increase in volatility. With the washout in those funds, the next nexus of risk in our minds is credit space, particularly given the relative under-reaction thus far.

So going back to Deutsche Bank's volatility model, which maps the impact of VIX on IG and HY spreads, the bank estimates that a 1 point rise in VIX is worth -0.6% to IG ETFs in aggregate, and -1.3% to the universe of HY funds.

Such sharp moves have yet to be observed: thus far, an index of IG ETF’s has fallen 1.2%, whereas the VIX shock should have pushed it 3.3% lower, and an index of HY funds is down just 2% versus a VIX implied drop of 7.2%.

In other words, the adverse shock to the credit space should be roughly 3 times greater to regress to its fair value levels.

This in turn lays out the main concern for credit investors, which according to Deutsche Bank is that "if credit spreads do indeed reprice to higher volatility, the drawdown in credit ETFs could trigger meaningful liquidation, resulting in further pressure on spreads."

To attempt to put some numbers around it, empirically a 1% m/m selloff in IG ETFs is consistent with a 1.9% decline in IG ETF AUM (currently about $130bn), meaning a 0.9% liquidation; for HY ETFS, a 1% sell-off equates to a 2.8% drop in ETF AUM (currently about $45bn), and therefore a liquidation worth 1.8% of AUM.

It gets worse: assuming the full repricing implied by the move higher in volatility occurs, it would imply $3.7bn redemption in IG funds, and about $6bn in HY ETFs (and the longer that VIX remains elevated, the more this risk grows). The acute risk for the credit market should such a flow materialize should be for HY, conditional on how concentrated it is – average daily volume in IG was about $17bn last year, whereas for HY is was $7.7bn.

Konstam's conclusion:

Should credit sell-off to where the current level of volatility implies, liquidity would likely deteriorate anyway, and the added pressure from fund outflows would likely further exacerbate the spread widening.

And while credit spreads have so far failed to blow out to levels suggested by the recent surge in the VIX, troubling signs that investors are already rushing for the door have emerged. On one hand, US junk bonds funds (both mutual funds and ETFs) saw outflows of $6.3b for the week ending February 14th, adding to the outflows HY funds started to see since mid-October. The outflow from riskier corporate debt funds occurred against the backdrop of rising UST yields. The past five weeks’ outflows now total to over $15bn or 7.2% of AUM. These combined outflows from both ETFs and MFs have now reversed much of the $21bn of combined inflows in 2017.

Worse, according to JPMorgan, higher frequency data via ETFs shows highest ever weekly outflow of $3.2bn, or 6.4% of AUM, from high-yield bond ETFs this week, bringing cumulative outflows for the current year to $6.8bn, or 13.6% of AUM. On a monthly basis, outflows from HY ETFs have continued uninterrupted since October 2017 and reached $11bn, or 22% of AUM. This reversed almost all of the cumulative inflows into HY bond ETFs since the US election.

Meanwhile, as JPMorgan further cautions, short interest ratios in the two largest US high yield ETFs, HYG and JNK, have been spiking up since mid-Jan (Figure 11).

Both HYG and JNK short interest are at their highs for the period we track data from, suggesting that institutional  investor participation via shorting ETFs has contributed to the sell-off in recent weeks. This is similar to the rise in the short interest ratio on the largest investment grade corporate bond ETF, LQD, which has moved higher during the same period this year, albeit from very low levels (Figure 12).

All of this points to the risk of a sharp blow out in credit spreads in the coming days and weeks should risk off sentiment return and/or should VIX fail to recover its recent "complacent" levels.

While in itself a move wider in spreads would not be catastrophic, it could still lead to a broad liquidation panic at the synthetic credit level, at which point the main risk becomes the underlying threat latent within all ETF products, first voiced by Howard Marks in March 2015: "what would happen, for example, if a large number of holders decided to sell a high yield bond ETF all at once?" This is how Marks answered his own question:

"in theory, the ETF can always be sold. Buyers may be scarce, but there should be some price at which one will materialize. Of course, the price that buyer will pay might represent a discount from the NAV of the underlying bonds. In that case, a bank should be willing to buy the creation units at that discount from NAV and short the underlying bonds at the prices used to calculate the NAV, earning an arbitrage profit and causing the gap to close. But then we’re back to wondering about whether there will be a buyer for the bonds the bank wants to short, and at what price. Thus we can’t get away from depending on the liquidity of the underlying high yield bonds. The ETF can’t be more liquid than the underlying, and we know the underlying can become highly illiquid."

Should the VIX fail to revert to its recent historic lows and remain at current "elevated" levels, which just happen to be in line with the "fear index" long-term average, we may soon find out if Marks' "worst case" scenario plays out as envisioned, and what exactly happens when everyone tries to sell a synthetic product that is far more liquid than its underlying constituents, especially during a market panic.