Running the economy hotter for longer could steepen Phillips curve

Running the economy hotter for longer could steepen Phillips curve

The Phillips curve describes the relationship between slack and inflation. In the short run, running the economy hot, or with output growth above potential, comes with the cost of additional inflation.

But policymakers cannot exploit this relationship forever because the public will come to expect a higher level of inflation for any given level of output. That is, the statistical relationship between slack and inflation shifts due to changes in policy behavior-an example of the Lucas critique.

This point is well understood and reflected in models of inflation such as the New Keynesian Phillips curve where agents' expectations of future inflation are key determinants for current inflation. It also motivates Fed policymakers' attention to measures of long-run inflation expectations and a determination to keep them anchored at levels consistent with the 2 percent target.

Analysts and forecasters often express the view that as long as long-run expectations remain anchored-and assuming a Phillips curve that was as flat as estimated over the decades before the pandemic-then even a persistent positive output gap would lead to inflation only slightly above 2 percent.

However, an underappreciated aspect of the New Keynesian Phillips curve is that even if the output gap is expected to close eventually and long-run expectations are anchored, the speed with which the output gap closes also matters for current inflation. If the public perceives the Federal Open Market Committee-seeking to lessen the chances of a recession-is aiming for a much slower cooling of the economy than was typical prepandemic, the Phillips curve could steepen. The result would be a meaningfully higher inflation rate than would be implied by those old inflation dynamics.

A simple model to make a simple point

This relationship is illustrated with a version of the model from a 2019 Brookings Institution paper by Richard K. Crump, Stefano Eusepi, Marc Giannoni and Ayşegül Şahin. (The authors make a similar point on pages 174-75).

Suppose that the current output gap is 1 percent, meaning economic activity is somewhat above a sustainable level, and the Phillips curve is relatively flat, consistent with data from 1998 to 2019 (Chart 1). Historically, this would be consistent with an unemployment rate around half a percentage point below its neutral level, or 3.8 percent using the Congressional Budget Office estimates of unemployment's long-run trend.

In the model, if the output gap is taken to be exogenous and expected to close by 5.0 percent each quarter, the implied current inflation rate would be 2.2 percent. The orange line in Chart 1 illustrates this relationship.

This is qualitatively consistent with forecasters' expectations of slightly below trend GDP growth, unemployment further ticking up and inflation falling to near 2 percent in 2025.

Does a flat Phillips curve mean that the unemployment rate can be kept below 4 percent for an extended period, and inflation will run only slightly above target? Not if that strategy is anticipated, according to the New Keynesian Phillips curve.

That's because the current inflation level varies with the current output gap level and the expected future output gap path. Higher expected output gap persistence raises short-run expectations. In other words, running the economy hotter for longer causes short-run inflation expectations to rise and raises current inflation, even if inflation is still expected to converge to 2 percent in the long run.

If expectations about the output gap path change and economic agents start to expect the output gap to close at a very slow pace of 0.5 percent per quarter, then the current inflation rate rises to 3.0 percent with the current output gap level unchanged (Chart 1, blue line).

Slack persistence as a policy decision

In this model, economic slack is taken as exogenous and assumed to converge to a sustainable level at a fixed pace. But the persistence of slack can be thought of as a monetary policy decision. Implicitly, monetary policy sets interest rates in a systematic manner that determines the behavior of real economic activity or slack. In turn, the level and persistence of slack determines the current level of inflation relative to target. In this case, running the economy hotter for longer is essentially a policy decision to allow economic activity to normalize more slowly.

The channel through which slack persistence matters is by way of expected inflation. The public learns about policymaker expectations over time and forms its own expectations. If the public never learns about a policy change to run the economy hot for longer, then this decision will not lead to a rise in current inflation for a given current output level under the theory outlined here.

Building on these scenarios, if policymakers decide that the economy will run hot for longer, but the public maintains expectations that output gap normalization will occur at about 5 percent per quarter, then inflation will remain at about 2.2 percent. Conversely, if the public learns about such a policy choice, inflation expectations will adjust, and current inflation will rise. In the case of normalization at 0.5 percent per quarter-hotter for longer-inflation would rise to 3.0 from 2.2 percent, while the current output gap level would remain unchanged at 1 percent.

'Hotter for longer' comes with a price

This forward-looking Phillips curve logic shows that the policy decision to run the economy hotter for longer can cause higher current inflation because expectations about future inflation rise as well. This would imply that short-run inflation expectations do not just move with food, gas or similar-type prices but can vary with changes in the public's understanding of the monetary policy strategy.

In practice, short-run inflation expectations as a predictor of future inflation or as capturing the public's perception of monetary strategy have had mixed results, and some backward-looking element in the Phillips curve is needed to fit the data.

So, while this is a stylized example, the point is an old and broad one that should not be forgotten: The statistical relationship between slack and inflation should not be expected to follow prepandemic dynamics if the behavior of policy has changed.