Deconstructing long-term interests may seem like a purely academic exercise. This is certainly how Economists treat it, coming at them using their statistical models. The goal is always to properly interpret these most basic of economic, financial, and monetary fundamentals so as to understand where everything that matters stands.
Getting this wrong is the difference between night and day; between driving a car with perfect vision or attempting to do so with a blindfold on.
In the latter case, you might be able to “sense” where you’re headed for a few minutes (seconds) before being rudely shown how much of a mistake it would’ve been to ever have tried.
Ever since Irving Fisher wrote The Rate of Interest in 1907, it’s been widely accepted how changes in nominal bond yields like those on US Treasuries come from two sources: changes in real rates plus changes in expected inflation. This kind of Fisherian deconstruction or decomposition has been examined thoroughly by Economists over the century in between.
The problem, from their point of view, is how do we know which is what and then quantify all for each? It’s not as if bond investors get together and describe these discrete breakdowns to the world for publication in the next day’s Wall Street Journal (for those who remember having to look up securities prices in a physical newspaper). In lieu of that or mind reading, econometrics created a regression analysis purporting to accomplish something close to this objective.
Right from the start, you can see where this just goes wrong. The Federal Reserve Bank of New York publishes daily data derived from one widely recognized model of yield decomposition. The branch’s “ACM Model” (which stands for its authors: Tobias Adrian, Richard K. Crump, Benjamin Mills, and Emanuel Moench) combines the real interest rate with inflation expectations via an affine methodology:
The term premium estimates that we present are obtained from a five-factor, no-arbitrage term structure model…Our model belongs to the affine class of term structure models which characterize yields as linear functions of a set of pricing factors.
As such, it only runs into big trouble. To figure out what the market must be thinking in terms of inflation and underlying fundamentals, linear functions of a set of pricing factors is a purposefully vague term for winging it but making it look scientific because…math.
What that has meant, by and large, is that Economists following this approach have fooled themselves especially in the post-crisis era at every step and stage along the way.
Let me show you exactly what I mean:
I like to use a conceptual model where the expected path of short-term rates isn’t lumped together with expected future inflation (as I’ll get to below) because there are certainly times when those things might (and have) diverge(d). What this econometric modeling does is combine these two factors as their view of inflation plus real rates and then merely compare those calculations to the market yield of long-term rates (of whichever maturity; I’m going to use the 10-year here).
In other words, this thing about “term premiums” is that they are nothing other than a remainder after all the estimating is done elsewhere. That’s it. Therefore, this puts total emphasis on the ability of Economists to properly calculate the blue stuff. Yeah.
Because they believe QE works (even though their own studies can’t really describe how or why), and that it eventually leads a recessionary economy into recovery, what is supposed to happen is that long-term bond yields rise due to the market believing those same things: inflation and real rates will absolutely go higher as they always would whenever legitimate business opportunities pick up creating full employment which then leads to tighter price conditions (labor and consumer prices).
Following especially the introductions of QE3 and then QE4 in late 2012, by mid-2013 with the unemployment rate dropping rapidly the modeled calculations all changed in this way. In FRBNY’s numbers, shown below, you can see how the models shift upward when QE was tapered in December 2013 and then kept following that rising path even though mere weeks later long-term bond yields reversed course.
The dark blue line is the nominal rate (market) while the light blue dotted line is the ACM fitted guess for what the math says the market must be thinking about inflation and such. Unsurprisingly, even as long-term yields dropped more and more during 2014 the model still came out with higher guesses for ST/inflation in the future.
No matter how far market yields would continue declining, the modeled decomposition kept going up because that was consistent with Bernanke and then Yellen’s view (not the market’s) of the situation (supported if only by the unemployment rate).
Since “term premiums” are nothing more than a made-up remainder, with falling market yields and rising modeled inflation expectations term premiums could only have declined (red dashed line). By the end of Euro$ #3 in the middle of 2016, the disparity grew so substantial that this model (and others) indicated somehow term premiums must have become outright negative.
And this all was what Ben Bernanke used in 2015 to try to justify his stance why falling yields – a clear warning sign – had worked out to something else entirely; even as rates dropped, according to the former Fed Chairman and QE’s American father, if it was because of “term premiums” then that was OK, even a positive signal!
This isn’t science and it is barely mathematics; all of it, pure rationalizing in an ongoing attempt to disguise unwelcome, outright contrary developments under the cloak of scientism.
But, as discussed yesterday, if your model for determining the bulk of rate decomposition is off – or just plain farcical – then the remainder for the rest of the equation is perfect junk. That’s all these term premiums have ever been.
Not only have you fooled yourself, as a central banker, Economist, or the multitudes of the financial media who only follow either of those you don’t actually understand what’s truly going on leading to not just widespread confusion but in economy and money terms far less than desirable results.
For one: whatever must be causing the nominal rates to fall goes undiagnosed, unappreciated and is therefore left unfixed. And that’s true even when low rates otherwise tell you exactly what it is that is wrong. This term premium approach is little more than a way to delude oneself about this fact.
And that’s why all these people, Bernanke, Yellen, Powell, and the media, all made the same mistakes again in 2018. They repeated the rationalizing and the term premium remainders that only got more ridiculous as the market kept rejecting the subjective scenario put into the affine models in the first place (garbage in, garbage out).
Look at this unqualified mess:
What the ACM model suggested (as well as the others like it) was that long-term yields were so low they left zero room for their remainder; and that’s not how real recovery is supposed to happen in bond yields. What should’ve happened, if the dotted light blue line was anything close to a useful approximation, is that LT nominals (dark blue) would’ve gone up a lot more than they did (below).
The fact they didn’t, this was a very clear sign something was very wrong (as I wrote repeatedly at that time: sour not soar). And then, post-2018 landmine, it just got plain ridiculous.
As LT rates really started falling in 2019 – with the models behind every leg lower – Economists by that summer just came out and claimed LT yields must’ve been wrong. Models over markets every time for them.
Even as the term premium remainder worked out to more deeply negative, the obviousness of the nonsense becoming only more obvious, it should have been a wakeup how their assumptions and interpretations must have been way, way off. Not just about 2019, but everything up to that point, too.
In truth, there is no reason to try to decompose yields in this econometric fashion. The ACM fellas at FRBNY, however, only point to one alternative to their affine affinity:
An alternative approach to estimating the ten-year term premium is to subtract the expected average level of short-term interest rates over the next ten years obtained from surveys of professional forecasters from the current ten-year Treasury yield.
Surveys of professional forecasters? Of course they’d only suggest them because what are professional forecasters but Economists themselves. Groupthink echo chamber.
No, as I discussed at exhausting length here, we have market prices for both the expected path of ST rates (eurodollar futures which do require a little bit of interpreting and should never be taken literally) as well as inflation expectations (TIPS breakevens, also never taken literally). In other words, the bond market has available for us whatever we need to diagnose in relatively close terms what’s going on inside nominal long-term yields using the three-part or two-part structure.
During reflationary periods, the blue parts changed upward but never all that much. They kept being pressured downward by “other risk factors” never accounted for by closed-system Economists who distrust markets this much.
What’s missing from the econometric models is, well, pretty much everything that’s important. As discussed yesterday, the intellectual deficit only starts by omitting how the monetary system, therefore financial markets then the real economy by the end, aren’t parceled up into individual patchwork closed systems.
Global money, global economy.
If your linear functions of a set of pricing factors don’t factor that fact then the deconstruction models will be way off meaning the remainder, term premiums, are just the trivial leftovers from a garbage model.
It’s yet another area where the bond market and low yields have proved quite predictive well in advance of mainstream conjecture on inflation, growth, and all the big basics.
If inflation really was coming in 2021, then we’d see much higher rates and wouldn’t have to stake everything on so much objectively bad math. “They” have told you all this time, and hysterically claiming the same thing again, how low bond yields are wrong or not what they otherwise seem. Put another way, here’s another key instance where you can clearly see how false promises of that inflation and growth those were never more than dressed up illusions.