When Oracles Lie

If I hear the term “false optic” from smart people a few more times, I’m going to start feeling smart myself.

Historically, that’s perilous, by the way. Hubris is famous for creating asymmetrical downside risks vis-à-vis life outcomes, but in my case, it doesn’t just bode poorly — it usually presages outright disaster. As a superstitious precaution, I temper my penchant for arrogance with self-deprecating humor. And whenever I start feeling smart, I re-read some classic novel or famous philosophical treatise, to remind myself that history’s smartest men is not a group to which I belong.

All of that said, there have been innumerable instances since Q4 2018 when I’ve suggested the bond market was a “Fata Morgana” (and I should thank Deutsche’s Aleksandar Kocic, because I think he substituted that for “false optic” in a note once, and it sounds so much better) — a mirage that tends to trick market participants into harboring macro delusions which then become self-fulfilling prophecies. In at least four cases, I’ve been a semblance of right.

This isn’t unique to rates, but it’s especially pronounced in bonds. Why? Well, because traders and investors are conditioned to believe bonds are “smarter” than other asset classes, so they assign special weight to what the bond market is purportedly “saying” about the economy.

Of course, the Fata Morgana characterization can be taken literally as well as figuratively — there is no “bond market” outside of the people and computers who trade it. It’s a literal mirage. As I put it Monday, investors habitually fail to recognize their own reflection in the mirror. Like a dog barking at itself in a reflective glass window, investors point to, say, a flattener, and then promptly sacrifice a goat to the market gods, hoping to appease the notoriously overzealous deities before Pythia’s prophecy plays out in a recession.

The whole thing becomes absurdly self-referential, especially once it starts to affect monetary policy outcomes in true “hall of mirrors” fashion. Throw in the (wholly lamentable) fact that a not insignificant number of bond market investors don’t really understand the role of systematic players and hedging dynamics, and you’ve got a recipe for… well, for false optics.

The above is just a (characteristically) laborious way of saying this: I never trust the bond market. If you want to conceptualize it as a thing that exists “out there,” so to speak, then that thing is notoriously deceitful. We like to say bonds have something like a “perfect” track record when it comes to predicting recessions. But how large is that sample size exactly? Can you really talk about track records with an “n” that small? I “taught” (don’t ask about the scare quotes, it’s a long story) probability and statistics briefly, so I can answer that: No. And besides that, how many times has the curve almost (but not quite) flagged a recession and been wrong? And so on.

Well, as it turns out, GPIF slashed its weighting of US Treasurys to 35% of foreign debt holdings in the fiscal year ending March, down from 47% previously. That’s a large reduction, to put it mildly.  “Some strategists suggested the pension giant may have sought to trim Treasurys because of an extended period of underperformance [while] others said this could have been incidental as it moved to reduce risk by aligning weightings with global indexes,” Bloomberg wrote, in the linked piece, adding that “GPIF’s Treasury holdings had shot up in the year earlier… as it was mulling the new investment plan,” which may suggest that the “extra funds had only been parked temporarily in US bills and notes as a substitute for cash before the fund settled on more permanent allocations.”

Whatever the rationale, this adds to the “false optic” case — it suggests that once again, the Oracle of Delphi may have been lying to us.

“The UST yield surge experienced in the back-half of the Japanese FY20… was perceived as largely attributable to fiscal stimulus [and the] renormalization impulse, but we now know [it] coincided with GPIF selling over the Japanese FYQ4,” Nomura’s Charlie McElligott said Tuesday. “So it’s possible that a large part of the UST weakness was not just the economic recovery and overshoot, but was also a flow as well, which in hindsight risked a potentially false optic,” he added.

McElligott went on to say there’s now at least some doubt around the notion that rising yields in Q1 were “entirely due to growth and inflation optimism alone, particularly as UST yields top-ticked exact[ly] at end-March, but have ground lower ever since.”

(I should note that McElligott doesn’t necessarily espouse the view that the entire Q1 Treasury selloff was “false.” Rather, he just mentioned the possibility that some of the impulse may not have been attributable to US macro factors.)

So here we are again, pondering a lying oracle. Can we trust her now? That is, if Q1’s Treasury selloff was at least part false optic, then what are we to say about 10-year US yields at ~1.16%? Is this just a lie in the other direction?

Morgan Stanley clearly thinks so. I touched on their take Monday, in “Why One Bank Isn’t Buying ‘Forced’ Bond Narrative,” but it’s probably worth citing a couple of short additional passages from a separate note.

“We think positioning has magnified the decline in yields and has been one of the key reasons why Treasury yields broke their range to the downside,” the bank’s Matthew Hornbach said, adding that despite the dour tone purportedly implicit in falling long-end yields, stocks are resilient, as is credit. In addition, Hornbach noted that “the market continues to keep the timing of the first hike stable while adjusting the pace of hikes,” the opposite of what you’d expect if the balance of macro risks were truly skewed to the downside.

But the most amusing passages from Hornbach came a few paragraphs later (in the same note). “While the Treasury market is forward-looking, the past suggests it’s usually wrong,” he said, flatly, referencing the figures (below).

Over the past decade, the policy rate midpoint averaged 0.64%. Morgan’s term structure model shows 10-year Treasury yields imply an average of 1.05% over the next 10 years. “Has the 10-year Treasury yield had much success in predicting this forward-looking average?,” Hornbach went on to ask, before answering himself: “Not really.”

The chart on the right (above) contrasts market-implied 10-year rate expectations with, as Morgan put it, “the ex-post reality.” “The question investors must answer today is, will the target fed funds rate average more or less than 1.05% through July 2031?,” Hornbach said.

As it turns out, predicting the future is hard — even for a purportedly infallible oracle.