Earlier this month, following a Fed statement that perhaps didn’t provide as much support for the hawkish narrative as some folks were anticipating, Bloomberg ran a story called “A Rare Treasury Yield-Curve Trade Emerges After Fed Decision“.
That “rare” mode of the curve: bull steepening. Here’s Brian Chappatta:
The Fed sparked a move in the U.S. yield curve that’s been virtually absent of late.
The spread between 5- and 30-year yields widened after the announcement to 33.6 basis points, the highest since April 27. Curve steepening is a rare enough occurrence — it’s near the flattest levels in more than a decade amid bets on continued gradual Fed rate hikes. But it’s the manner of the steepening that’s striking.
Specifically, the 5s30s bounced off what, at the time, was a multi-year low below 32bp the day before, and steepened by more than 2bp in the 10 minutes following the Fed statement.
As I was quick to point out, that dovetailed with recent commentary from Deutsche Bank’s Aleksandar Kocic, who just days before, suggested that “tail risk is shifting from bear steepeners to bull steepeners” as the Fed attempts to pull volatility back to the front end.
The week before the May Fed statement, Kocic outlined the extent to which – and I’m quoting his latest note here in order to describe his previous piece – “the Fed is daisy chaining the two ends of the curve with the equities market, effectively buying back convexity from equities, recycling it through the front end and sending it as convexity supply to the back end of the curve.”
You can read more about this in the linked post above (aptly entitled “mind the convexity flow”), but here’s the schematic:
The destabilization in equities catalyzed by the persistence of the Fed, helps underpin the long end as volatility in equities engenders a safe haven bid for Treasurys while Fed hikes underpin the dollar and help cap inflation expectations, with the latter effort helping to ensure the tail risk of an unwind in the bond trade isn’t realized (and as a reminder, you don’t want that tail risk to be realized).
“Restriking of the Fed put is re-syphoning of convexity,” Kocic wrote, adding that “withdrawal of convexity from equities means higher volatility and their underperformance, which fosters preference for bonds and reinforces their stability.”
In a regime where everything is prone to selling off, volatility is effectively all that matters when it comes to preferences. It becomes a story about what’s most vulnerable and in that context, elevated stock volatility (relative to other assets) means duration is preferable to equities and that’s self-feeding – vol. gets pushed away from rates.
Ok, so getting back to the “anomalous” bull steepening mentioned here at the outset, Kocic’s latest note finds him reminding you that “in normal market conditions, shocks arrive to the front end of the curve.” To wit, from a note dated Friday:
[Normally] the curve evolves through bull steepening or bear flattening modes. During the QE period, directionality of the slope was reversed: As the front end remained anchored at zero, shocks arrived through the back end with the curve repricing through bear steepening and bull flattening modes.
He illustrates this with the following chart, which shows the regime shift at the beginning of the QE era, when suddenly, the typically negative correlation between the 2Y rate and the 2s10s flipped:
So here I’m going to reverse the order in which Kocic presents things in the note, because I think that will make it easier to follow for folks who might not be steeped (Heisenberg double entendre alert) in this stuff.
Implicit in that chart is the general idea that in the post-QE era, whatever 2s are doing, 10s are doing, only more of it. Hence the bear steepeners and bull flatteners. Here’s Kocic:
2s and 10s track each other (most of the time), but whatever 2s do, 10s do more of the same: If both go up, 10s goes up more, and if they both go down, 10s go down more than 2s.
Contrast that with the pre-QE mode:
2s and 10s track each other (most of the time), but whatever 10s do, 2s do more of the same. If both go up, 2s underperform, i.e. goes up more, and if they both go down, 2s outperforms, i.e. goes down more than 10s.
Ok, so what does that entail for vol.? Well, it entails two distinct regimes with regime shift coming at exactly the same point as illustrated above and with that shift being characterized by 3M2Y/3M10Y ratio being higher than the 2s10s correlation in the pre-QE era and lower thereafter.
“In this way, the question of normalization becomes synonymous with distribution of vol along the curve and, therefore, with the shape of the volatility surface,” Kocic goes on to write.
Getting back to what I said above and characterizing it in terms of what’s “normal” and what’s indicative of the post-QE regime, normalization means the Fed will attempt to pull volatility back to the front end. Here’s what that presages, according to Kocic:
Currently, 3M2Y is at 41bp and 3M10Y at 59bp. Their ratio is at 0.7 while correlation between 2s and 10s, the threshold of normalization, is 86%. For correlations to flip sign (for rates to normalize), 3M2Y gamma would have to rise above 51bp (all else equal), 1Y2Y above 56bp.
Now, at this point you’re either tuned out or you’re asking the following question: why is it “normal” (so to speak) for volatility to be concentrated at the front end?
That gets back to the discussion we had a month ago in “Inner And Outer Limits: Resolving An Inconsistency And What It Means For Risk“, in which we brought in multiple quotes from analysts we’d spoken to off the record and combined that insight with commentary from both Kocic and his colleague Steven Zeng to explain the whole story behind a JPMorgan note that flagged “the first signs of curve inversion in the U.S.”
Recall the following assessment from one analyst we spoke to for that post, commenting on the JPMorgan piece:
So these are spreads to greens (2y fwds) and the 2y fwd is basically the terminal Fed, so the spread between reds and greens (1y vs 2y fwd) is how aggressively the Fed is going to hike. The spread between blues/greens (3y vs 2y) is basically how likely it is that greens are the terminal Fed.
That seems to suggest that there’s some ambiguity about the destination of the Fed’s quest. Happily, Kocic’s latest note touches on this in the context of how things used to be. To wit:
Prior to 2008 financial crisis, Red contracts on the Eurodollar curve used to be the most volatile rates sector. Typically, policy relevant horizon was set to be around two years. On the back of that, rate hikes become an optimal control problem. The Fed sets on its “journey” with the market making up its mind about where the terminal rates should be. This “anchors” Green contracts (2Y forwards) and stabilizes rates beyond the 2Y horizon. While the market remains confident about the success of the “journey”, it grants Fed an option to optimize rates path. As rate hikes occur at discrete time intervals, the very front end remains relatively inert, and with terminal points being already set, the most volatile sector falls somewhere in the area of 1Y forwards. The arrival of new information does not affect materially either the immediate Fed action or its final destination, but mostly the path along the way to terminal rate.
Got it. And that means that if volatility resides primarily in reds, then things are as they “should” be, but if that volatility starts to move outward, it suggests instead that folks might be a bit unsure about how things will ultimately end up. Or, as Kocic puts it, “migration of volatility from the Red sector to the long end signals erosion of confidence.” Well witness that “migration” as the red line trends lower post-crisis:
This raises obvious questions about why, despite multiple Fed hikes, things still aren’t normalizing. Kocic attributes this in part to excessive transparency and the two-way communication loop between the Fed and markets, a concept that, frankly, informs a lot of what we write.
As usual, the withdrawal of that transparency could come in the form of a more data-dependent Fed and that, in turn, means that transitioning back to a more “normal” regime for rates depends on the evolution of the econ.
But there’s a problem with this setup, namely that you don’t need transparency if you’ve got trust, right? I mean, if you trust your significant other, you don’t need to be rifling through his/her text messages while he/she is in the other room, for instance. To a certain extent, the more trust you have the less transparency you need and vice versa.
Volatility residing in reds (see above) entails trust about the assumed success of the Fed’s quest, while the migration outward reflects the opposite; that is, it reflects transparency (suppressing volatility at the front end as the committee remains predictable) but no trust (less faith in the success of the journey).
“Pre-2008, trust was reflected through high confidence regarding the estimate of the policy relevant horizon — Reds used to be more volatile than Greens or long rates in general,” Kocic writes, before reminding you that “this is no longer the case.”
In the end, the question is whether the Fed can pull this off – i.e., whether they can succeed in withdrawing transparency while simultaneously restoring the trust that’s necessary for a less transparent regime to exist.
If they can’t stick the proverbial dismount on that, well then destabilization of one kind of another would appear to be in the cards.
It is, to borrow the title of Kocic’s note, “Fed in wonderland.”