The behavior of inflation was already a mystery before it started rising in 2021. For example, back in 2016, then-Chair of the Federal Reserve Janet Yellen gave a talk about important research topics in macroeconomics, and one of the topics was: “What determines inflation?” Other macroeconomists at about that time were also grappling with what I called the “mysteries of modern inflation.”

At that time, the main puzzle was that inflation seemed overly stable for the previous 20 years or so. Inflation had not noticeably taken off during, say, the boom years of the dot-com revolution in the late 1990s or the housing boom of the early 2000s. Inflation had not also not dropped in any sustained way during the Great Recession of 2007-9 or the earlier recession in 2001. The Federal Reserve had revamped monetary policy, pushing the policy interest rate (the “federal funds rate”) down to essentially 0%, as well as making large purchases of government and government-guaranteed debt in policies of “quantitative easing,” and inflation barely budged. So why was inflation so stuck? And why did it come unstuck?

Ricardo Reis discusses some possibilities in “The burst of high inflation in 2021–22: how and why did we get here?” (Bank of International Settlements, December 16, 2022, Working Papers #1060). He starts with a striking figure, using data from the Millenium dataset of the Bank of England, providing in this case 800 years of inlation data for the British economy. The data is grouped into 20-year periods. Average annual inflation in consumer prices is measured on the horizontal axis. The vertical axis measures the standard deviation of inflation during that 20-year period: loosely, you can think of this as how much variation in annual inflation rates during a certain period. Thus, if inflation spiked way down, or way up, the level of variation would be high.

The 20-year period of 1997-2016 is at the bottom middle: that is, the inflation rate was 2%, but the amount of variation in that inflation rate was the lowest for any of the 20-year time periods. Historically, whether inflation was lower or higher on average over a 20-year period, it involved a lot more movement–and bursts of deflation or inflation pose a challenge for economic performance. When inflation rates are moving, a lot of time and energy needs to go into reducing the risks of such changes; when inflation is stable, economic actors can focus more clearly on actual decisions about production, consumption, and investing.

Reis then moves to four arguments about why inflation came unstuck.

#1: Shocks and Misdiagnoses

When the pandemic hit in early 2020, there was a reasonable fear that the economy would suffer enormously, which led to a major burst of expansionary monetary and fiscal policy. This policy worked in the short run. For example, the pandemic recession was only two months long, and when the “Delta” variant of COVID hit in late 2020 with high numbers of deaths and widespread lockdowns, there was no follow-up recession. However, inflation did get underway. Reis writes: “The amount of fiscal and monetary stimulus in 2020 was perhaps excessive, although this judgement comes with the benefit of hindsight. A more pertinent criticism is that policy did not reverse course until at least the end of 2021, even as the signs that the fast recovery was leading to overheating became clearer.”

The pandemic shock was then followed by the supply chain shocks of 2021 and the Russian-invasion-of-Ukraine shocks in February 2022. In each case, Reis points out, these shocks were interpreted by central banks as having only a temporary effect on inflation–so the main policy response was to keep stimulating demand in the economy, to overcome these supposedly temporary issues. But an alternative interpretation was that these shocks were having real and at least moderately lasting effects in raising prices while reducing production. Reis writes:

Three times in a row in a short period of time, a set of shocks pushed inflation up. Three times in a row, monetary policy interpreted them using the lenses of the Phillips curve in the direction that concluded that monetary policy should be kept loose. Three times in a row, this diagnosis was plausibly right but disputable, and the risk was that inflation would rise too much and too persistently. After the fact, in all three cases this risk became reality. A policy framework should be robust to shocks, and it should correct misdiagnoses. So many successive errors in the same direction indicate more systematic problems.

#2: Short-term Expectations

One legacy of inflation being stuck in place from the 1990s up through 2020 was that people’s expectations of inflation–rationally enough–also seemed stuck in place. To put it another way, the memory that inflation could rise had faded for a generation of people and policymakers. Indeed, looking at survey data on inflation expectations in 2021 and into 2022, it looked as if inflation expectations still had not moved by much–which was one reason why many policymakers were not worrying about inflation. But as Reis points out, while the median expectation did not move move, the spread in opinions about inflation was rising. He writes: “The expectations anchor had left the seabed after a couple of decades during which it had barely moved.”

#3: Long-term Credibility

Behind the idea that inflation should be expected to stay low is a credible belief that even if inflation bumps up and down in the short run, policymakers will act as needed to keep inflation low in the long run. The anti-inflation credibility of policymakers was pretty high in 2020, after several decades of low inflation rates during a wide array of economic events. But again, Reis points out that if one looks at expectations of long-run inflation–as measured by survey data or as estimated by financial markets–the risk that inflation might still be exceeding 4-5% five years from now has risen. The anti-inflation credibility of policymakers is wavering.

#4: R-star and the Tolerance of Inflation

In the lingo, “R-star” refers to the level of interest rates that lead to an economy where there is no underlying pressure for the rate of inflation to rise or fall. But R-star can be set at a level where the resulting inflation is relatively low, with no tendency to rise or fall, or at a level where the resulting inflation is relatively high, with no tendency to rise or fall. As you might expect, estimating R-star is more art than science. But up to about 2020, it’s a common theme in the comments of central bankers like Jerome Powell at the Fed that perhaps inflation is too low, and thus when setting inflation rates the Fed should err on the side of tolerating a little more inflation.

Notice that all of these factors suggest that the Federal Reserve ultimately has power over whether inflation takes off or not. They also suggest that even if one can come up with arguments or justifications for Fed decisions to not raise interest rates in 2020 and 2021, and only to start doing so in March 2022, those earlier arguments and justifications in fact have turned out to be incorrect. Central banking is a results-oriented business. If you let inflation out of the cage, the quality of your earlier arguments and intentions doesn’t change the reality that you have failed in one of your main tasks.