I ran into an amusing article by a certain mainstream economist, but I have used up my quota for dunking on him in Q2, so I will just make a more generic point related to an underlying issue that often comes up in discussions of the bond market by people who are not fixed income strategists. The argument is straightforward: the “common sense” belief that a jump in measured inflation will cause a rise in bond yields is incorrect. The reason is straightforward: the bond markets are forward-looking, while measured inflation is a backwards-looking measure (and is normally considered a lagging economic indicator).
(Note: I have a big consulting project is ongoing, hence this article is light. It is exceedingly likely that I have written a variation of this article many times. I am just repeating it since it is somewhat topical, and there is a chance that I picked up new readers when I launched my Substack.)
REMINDER: Feedspot email support is being discontinued in July. E-mail subscribers need to opt-in to the mailing list on my Substack to get my articles mailed to them. Link: bondeconomics.substack.com.
The “common sense” belief is extremely common, and I assume that it is the result of showing up in some decades-old mainstream Economics 101 textbooks. Over my years of hanging around fixed income, I have seen many variations of this from comments from the broad public as well as economists.
I would summarise the belief as follows: a rise in inflation causes bondholders to “lose money,” and so bond yields will mechanically rise by the increase in inflation to compensate (creating a further capital loss). Strangely enough, bonds are the only asset class (other than possibly cash) which “loses money” due to a rise in inflation. Even if equity and bond returns are the same, a rise in inflation causes losses for bondholders, yet equity holders are unaffected.
The Economics 101 part of the story is that bond investors “demand” a constant real yield. (The real yield is defined here as the nominal bond yield less spot inflation, which is not the same thing as the quoted yield on an inflation-linked bond — like the U.S. TIPS.) Therefore, since the real yield is fixed, nominal bond yields must rise by the amount of inflation.
If one wishes, one could scour online economic time series databases and try to torture the data to “prove” this belief. However, I will merely assert that we really cannot see such a relationship in the data. Maybe it worked one time, but it failed horribly the next.
Why does it fail? There are a number of pillars.
- Bond investors mainly work for institutions, and they are not normally benchmarked to inflation. (Admittedly, this is the case for some portfolios.) Their job is to make money. And if they are levered — and a lot of price determination is done by levered investors — prospective returns are supposed to much higher than quoted bond yields. If I were involved in structuring a trade in U.K. gilts, the change of price of Toad in the Hole in the local in Middlesmoor was something that I would have had zero interest in.
- Bond investors cannot “demand” any particular yield; they are stuck buying at whatever the market is trading at. If yields are too low, they can go short duration versus benchmark, but risk-taking capacity is finite. This means that they cannot force real yields to any particular value.
- Finally, bond prices reflect expectations about the future, while the inflation rate is the change in the price index from a recent past month versus the year before.
The final point is not entirely obvious, and has some entirely pointless controversies associated with it. The fair value of a bond is supposed to equal the “expected” return of a money market portfolio over the lifetime of the bond (with the same credit risk as the bond) plus an added “term premium.” Back in the day, the working assumption was that term premia were positive, but nowadays, anything goes.
The “expected” return on a money market portfolio will roughly equal the “expected average” of the policy rate, set by the central bank. That is, it is based on what bond investors think will happen. Published inflation rates are what happened.
Nevertheless, there is a way to make the “common sense” belief work: the central bank needs to react to past inflation. And if we look at the more incompetent central bankers in the past (cough ECB cough), that is how they behaved. However, say what you want about the New Keynesian whiz kids running central banks, they tend not to overreact to backwards-looking data. That said, not everyone in the mainstream (and elsewhere) got that message.
It is a mistake to argue that inflation does not matter under any circumstance for bonds, but you need to be extremely careful about what inferences you draw. Markets are hard to beat, and some market participants can forecast inflation data a few months ahead relatively accurately. This means that anything you read about in the papers about the latest CPI print is likely to have already been priced in.
Email subscription: Go to https://bondeconomics.substack.com/
(c) Brian Romanchuk 2021
Post a Comment
Note: Posts are manually moderated, with a varying delay. Some disappear.
The comment section here is largely dead. My Substack or Twitter are better places to have a conversation.
Given that this is largely a backup way to reach me, I am going to reject posts that annoy me. Please post lengthy essays elsewhere.