Not-So-Great Expectations: Why Treasuries Don’t Tell The Whole Inflation Story

|
Posted: Apr 12, 2021 10:26 AM
Not-So-Great Expectations: Why Treasuries Don’t Tell The Whole Inflation Story

Source: AP Photo/Pablo Martinez Monsivais

Ramesh Ponnuru recently took up the case against inflation... well, more like the case against predictions of inflation. If I’m reading him correctly, he is making the case for policies in pursuit of higher inflation. To his credit, he does not shy away from a deep dive into intellectually challenging analysis.

Ramesh argues as follows

  1. Anxiety about inflation is high.
  2. The case that inflation risks are rising is based partly on a particular metric, which is calculated by subtracting the yield of an inflation-protected five-year Treasury note from that of a regular five-year Treasury note. This metric has been rising and, as of Ramesh’s writing, was above the Fed’s target rate for CPI.
  3. That metric is not reliable, because the Fed buys Treasury securities, which distorts the rate. So that factor can mislead users of that metric as to how much inflation fear there is among real investors (as opposed to central bankers, who buy these bonds as a policy instrument, not an investment).
  4. The Fed knows about this problem, so it recalculates the spread to reflect this reality. Ponnuru includes a helpful explainer from the Fed.
  5. For example, at the end of 2019, the Fed’s adjusted inflation expectation was 1.83 percent, above the 1.68 percent unadjusted expectation.
  6. As of March 3 of this year, the day before Ponnuru’s article was published, the adjusted inflation risk had fallen to 1.67 percent — lower than before the COVID scare.
  7. Failing to take this adjustment into account makes inflation risks look higher than they really are.
  8. If markets know more than experts, then we have no need for anxiety.

Alas, I remain unconvinced, but still convincible. I’m not in a camp. My job is to make forecasts. I don’t get demerits for failing to hew to a certain party line; I get demerits for being wrong.

So let’s answer point by point

  1. Yes, inflation anxiety is high, but the question is whether it is too high or not high enough.
  2. Yes, TIPS spreads are part of the evidence cited for elevated inflation, but they are hardly the only piece of evidence, nor the dominant one. I don’t know a single analyst who is making the hawkish case based on TIPS spreads alone.
  3. Yes, the Fed distorts interest rates, but it distorts both inflation-protected and regular interest rates. There are other distortions to these rates as well. An accurate projection accounts for all distortions, not just the ones that help the ”Relax, nothing to worry about” case.
  4. Yes, the Fed recognizes the insufficiency of observing yield spreads to predict inflation, and attempts to adjust for it, but it also fully acknowledges the limitations of its method. Fed analysts are humble about their ability to “decompose” real-world observed interest rates into some theoretically reconstituted rate to derive a theoretical inflation risk.
  5. This is where I find myself a bit confused about what Ramesh is doing. Regarding the liquidity-distortion effect, he says, “That makes the market’s projection of inflation look larger than it is. The Fed also publishes a data series that attempts to correct for this distortion (which is available at the bottom of the page in the previous link).” But that seems to leave quite a lot of relevant material out. We’ll put an analysis at the end of this paper for the math-phobic, but suffice it to say that these distortions can pull in either direction, either overestimating or underestimating inflation expectations, but tend in general to underestimate inflation risks. More at the bottom of this article.
  6. I have no reason to doubt the Fed’s or Cornerstone Macro’s expert adjustments to risk spreads, but I’m confused by Ponnuru’s position on expert knowledge versus market knowledge. These adjustments are performed by economists; in other words, experts. No one is simply observing inflation expectations in market data: They are building complex models to try to uncover some real yield or some real inflation risk premium. If Ponnuru’s supposition is that markets know more than experts, “even highly informed ones,” then what precisely is the point of presenting expert calculations in order to suggest the data are exaggerating inflation risk?
  7. See section below.
  8. Returning to the markets-vs.-experts issue, I wonder why we are depending only on one market in order to build the case against inflation, while ignoring other markets. I like the TIPS spread, and, for the record, I think the adjusted version is a real improvement, but there are other market signals to look at. Gold is a market. So is silver. So are currency markets. When many in the supply-side movement were pushing the Fed to hike higher and higher in 2006, they pointed to gold markets but ignored bond markets. I argued at the time that all market indicators needed to be listened to, not just the ones that fit our outlook. Data are a gift (literally in Latin, “things given”); let’s open all the gifts to get our insights.

More detail than you want about TIPS spread adjustments

Yes, the Fed paper Ramesh links to acknowledges that Fed purchases can distort yields, but the main point of the paper is to describe ways in which there is a “liquidity premium” that tends to raise rather than lower TIPS yields, which means that spreads usually underestimate rather than overestimate inflation risk. Bluntly, this factor generally (but not always) favors the hawks. Higher yields for TIPS means smaller spreads with regular Treasuries. The authors say that explicitly (sorry, algebra ahead . . .):

In the model, TIPS yields typically exceed “true” real yields (or real yields that are consistent with nominal yields) due to the TIPS liquidity premium:

TIPS yield = real yield + TIPS liquidity premium.

In other words, usually TIPS yields are elevated by the liquidity premium. This, of course, makes the spread smaller, meaning that in general, an adjustment for this effect will reveal greater inflation risk.

Let’s look at Ponnuru’s example. He says that the adjusted inflation expectations rate at the end of 2019 was 1.83 percent. He’s right, but the unadjusted rate was 1.68 percent. In other words, the adjustment raised the inflation outlook! The adjustment doesn’t generally make inflation risks look lower — it generally makes them look higher. 

That’s why the Fed subtracts the liquidity premium to get to an estimate of inflation compensation:

As a result, TIPS inflation compensation (IC)—defined as the nominal yield minus the TIPS yield—can be decomposed into three components,

TIPS IC = expected inflation + inflation risk premium – TIPS liquidity premium.

Why does the liquidity premium tend to work this way? Because TIPS are a bit exotic compared to regular Treasuries. They’re a good deal more complex. In my own experience, I remember during a consulting engagement recommending to a portfolio analyst (one with a CFA, no less) that he consider adding TIPS to the portfolio. He objected to the notion based on a lack of familiarity with the instrument and confusion about the nature of the calculation. The Fed recognizes that a significant part of TIPS price distortion results from uncertainty towards adoption:

By contrast, TIPS liquidity premium is not connected to inflation risk but rather reflects factors that drive a wedge between TIPS yields and true real yields. This premium was high when TIPS was first launched, as it took some time for TIPS to gain popularity among investors, and again surged during the 2008-2009 Financial Crisis, as investors fled from less liquid or more risky instruments and sought the safety and liquidity of nominal Treasury securities. Apart from lower liquidity of TIPS relative to nominal Treasuries, this premium may also reflect supply-demand imbalance of TIPS versus nominal securities and a greater concentration of buy-and-hold investors in TIPS market compared with the nominal Treasury market.

The bottom line is that the difference between regular Treasuries and TIPS is partly based on inflation risk, but partly based on other factors, one of which is that TIPS investors require a higher yield in return for the complexity and relative illiquidity of their choice security. On the other hand, there are other complicated flows involved: whether the Treasury Department is growing the supply of TIPS relative to regular Treasuries and whether the Fed is buying relatively more TIPS than regulars. In general, though, these distortions underestimate inflation, not overestimate it.

Assuming that Fed analysts have gotten things right, on average, these supply and demand effects have amounted to an average of 0.67 percent over the time they have been calculating this metric. Though as of latest data available, February 25, the premium is a -0.63 percent, amounting to a slight understatement of inflation.

This article originally appeared on National Review.