For readers who are not familiar with the concept of breakeven inflation, it is the nominal yield on a conventional Treasury bond less the quoted yield (indexed, or "real yield") on a matched maturity inflation-linked Treasury (TIPS). I discuss the mechanics of breakeven inflation in my handbook, Breakeven Inflation Analysis. The key observation is that the breakeven inflation rate is (roughly) equal to the required rate of inflation for the TIPS and conventional bond to have the same rate of return when held to maturity -- that is, the TIPS is priced at a break even level versus the nominal benchmark.
Since inflation-linked bonds are based on overall ("headline") CPI, the inflation payout is greatly influenced in the near run by changes in energy prices. On longer horizons, it is difficult for energy prices to significantly diverge from the overall price index, so averages end up looking closer to core (ex-food and energy) inflation that economists tend to focus on. (Some believers in Peak Oil argue that energy prices will spiral out of control, but I believe that overstates what will happen to measured prices.)
The easiest way to deal with the effect of energy prices on expected inflation is to break the investment horizon into two intervals.
- The near run outlook, based on the current state of the business cycle.
- A longer run average, which is an estimate of where policymakers would like inflation to be.
As an example, I would break up the 10-year inflation breakeven into the first 5 years (the spot 5-year breakeven inflation rate), plus the 5-year breakeven starting 5 years forward. (The advantage of that separation is that the 10-year breakeven is just the average of those two levels.)
(The issue of risk premia in inflation-linked bonds is a thorny question. There are two plausible, countervailing forces. Inflation-linked bonds are less liquid, and they should be cheaper than conventional government bonds -- which are a pricing benchmark. Conversely, there is a mismatch between the supply and demand of inflation protection, with a far greater demand for protection than there is supply. This ought to result in inflation-linked bonds being expensive -- breakevens above inflation. Since we do not know exactly what market participants expectations for inflation are, and there is only a single spread between the observed breakeven and this forecast level, it is extremely hard to come up with a plausible model for the premium. That said, this is only a concern for academics or central bankers -- as an investor, you know what your own inflation forecast is, and you can set your own target risk premium.)
Although adaptive expectations are sneered at by modern academics, their use is justifiable as long as one understands the risks. The choice of 10 years in the moving average was not accidental -- 10 years largely matches the length of recent expansions. The cyclical average tells us where overall policy settings (i.e, fiscal, monetary, and regulatory policy) have let inflation settle. Unless those policy preferences change, the best guess is that future inflation will tend to settle near those levels. Meanwhile, if one looked at a long history of the average of inflation, we see that the average has settled into a relatively small range since the early 1990s.
I am extremely unconvinced by stories that there has been a secular change in policymakers' views towards inflation. As such, I see no reason to argue against where forwards are. As for the near-run inflation forecast, even if I wanted to be a forecaster (which I am not), my feeling is that the outlook is murky. The current situation is somewhat unprecedented in the modern era, with a mixture of unemployment alongside industries hitting capacity constraints. Meanwhile, it is hard to judge how large the post-vaccine bounce will be.
Finally, one popular belief is that "breakevens have nothing to do with inflation." That is the type of thinking that drove the forward breakeven inflation rate down to 1.5%, which generated a good trading opportunity for anyone who disagreed with that sentiment.
(c) Brian Romanchuk 2020