Average Inflation Targeting: How the Fed Replaced Strict 2 Percent in 2020

For most of the decade after the 2008 financial crisis, the Federal Reserve had a problem it could not solve through normal policy. Inflation kept running below its 2 percent target. The Fed’s preferred measure, the core PCE price index, averaged about 1.6 percent from 2009 through 2019, leaving the price level roughly four percentage points lower than a steady 2 percent path would have delivered. Under the old framework, this miss was treated as water under the bridge: each year began with a fresh attempt to hit 2 percent, regardless of what had happened before. By the late 2010s, FOMC officials began describing this asymmetry as a structural problem. The Fed had been undershooting for a decade, expectations of future inflation were drifting lower, and policy was running out of room because nominal rates kept brushing against zero. In August 2020, the Federal Reserve announced a new framework. Average inflation targeting, or AIT, formally commits the central bank to letting inflation run moderately above 2 percent following periods of undershoot, with the goal of keeping the average at 2 percent over time rather than aiming for 2 percent each year.

AIT was adopted to address the asymmetric risk of persistent undershoots when the policy rate approaches zero. The framework’s make‑up strategy commits the Fed to tolerating inflation above target following periods of below‑target outcomes. The first stress test arrived within a year of adoption, when the 2021‑2023 inflation overshoot pushed the framework to its limits. The 2025 review softened the make‑up commitment and restored more symmetric language on employment, leaving the framework intact but more conditional than the original 2020 version.

The Problem AIT Was Designed to Solve

The case for AIT begins with what economists call the asymmetry of inflation targeting at the lower bound. Under standard inflation targeting, the central bank treats each period as independent. If inflation comes in at 1 percent this year, the bank does not try to deliver 3 percent next year to make up the difference. It simply tries to deliver 2 percent next year. The shortfall is forgotten. This works fine when policy has plenty of room to move in both directions, but it creates a one-sided bias when nominal rates approach zero. The central bank can always raise rates to bring high inflation down. It cannot easily cut rates below zero to bring low inflation up. Persistent undershoots, therefore, have an asymmetric quality: they accumulate over time without any built-in mechanism to reverse them.

The consequence is that long-run inflation expectations drift downward. Households and firms observe that the central bank misses its target on the downside more often than on the upside, infer that 2 percent is more of a ceiling than a midpoint, and adjust their wage and price-setting behavior accordingly. Once expectations anchor below the official target, the central bank loses some of the credibility that the target was designed to provide in the first place. Japan’s experience after the late 1990s provided a vivid demonstration: years of mild deflation became self-reinforcing as expectations adjusted, and the Bank of Japan struggled to convince the public that its inflation target meant what it said.

The conventional inflation-targeting framework, used by most advanced-economy central banks since the early 1990s, was developed in a higher-rate environment where the lower bound was a theoretical concern rather than a practical one. The neutral rate of interest, r-star, had fallen far enough by the late 2010s that even normal recessions could push policy rates to zero. The Fed staff and external academics began publishing analyses through the mid-2010s, arguing that the framework needed to adapt to a permanently lower-rate world. The 2020 framework review formalized the response.

The August 2020 Statement

On August 27, 2020, Chair Jerome Powell delivered a speech at the Jackson Hole conference announcing the conclusions of the Fed’s first formal monetary policy framework review. The accompanying revised Statement on Longer-Run Goals and Monetary Policy Strategy contained two material changes from the 2012 version it replaced.

The first change was a single phrase. Where the 2012 statement described 2 percent as a symmetric inflation goal, the 2020 statement specified that the Fed “seeks to achieve inflation that averages 2 percent over time” and that, “following periods when inflation has been running persistently below 2 percent, appropriate monetary policy will likely aim to achieve inflation moderately above 2 percent for some time.” This is the average inflation targeting clause. It is the formal commitment to make-up strategy: a willingness to tolerate inflation above target for some period following an undershoot, so that the longer-run average can hit 2 percent.

The second change concerned the employment mandate. The 2020 statement said policy would respond to “shortfalls” from maximum employment rather than to “deviations” in either direction. In practice, this meant the Fed would no longer preemptively raise rates simply because the labor market was strong, as long as inflation remained controlled. The two changes were paired: AIT relaxed the inflation side, the shortfall language relaxed the employment side, and the combined effect was a framework explicitly designed to run the economy “hot” for longer than the pre-2020 framework would have allowed.

AIT versus Conventional Inflation Targeting

The differences between AIT and standard inflation targeting matter less in normal times than in extreme periods. In a steady-state economy with inflation at target and stable expectations, the two frameworks prescribe nearly identical policy. The differences open up in three specific situations: after a sustained undershoot, when the lower bound becomes binding, and when an unexpected inflation shock arrives. The table below sets out the contrast across the dimensions that matter for policy design.

Table 1. Average Inflation Targeting vs Conventional Inflation Targeting Across Key Policy Dimensions

Dimension	Standard Inflation Targeting	Average Inflation Targeting (AIT)
Treatment of past inflation	Bygones are bygones. Each year reset.	Past undershoots are recognized and partially made up.
Tolerance for overshoot after undershoot	None. Inflation above target triggers tightening.	Moderate. Inflation above 2 percent tolerated “for some time.”
Treatment of strong labor markets	Strength can trigger preemptive tightening.	Shortfalls from maximum employment are the policy concern.
Behavior at the lower bound	Same as away from the bound.	Designed to keep expectations anchored at 2 percent.
Quantitative make-up rule	Not applicable.	Not specified. Discretionary judgement.
Reference period for average	Not applicable.	Not specified. “Over time.”
Net policy bias	Symmetric in principle, downward-biased at the lower bound.	Tilted toward allowing modest overshoots after misses.
Source: Federal Reserve Statement on Longer-Run Goals and Monetary Policy Strategy, 2012 and 2020 versions.

Two features of the table are worth highlighting. First, AIT under the 2020 framework is not a mechanical rule. It does not specify how much make-up inflation is appropriate, over what window the average should be computed, or how aggressive the overshoot tolerance should be. These are left as judgments for the FOMC to make in real time. This is sometimes called “flexible average inflation targeting”, or FAIT, to emphasize the discretionary character. Second, the framework was designed for the undershoot problem. It says little about the symmetric problem of how to handle an unexpected overshoot, which is exactly the situation the Fed faced from mid-2021 onward.

The PCE Path Before and After 2020

The chart below shows the Fed’s preferred inflation measure, the core PCE price index, alongside the 2 percent target band that the FOMC effectively communicates. The pre-2020 period is dominated by persistent undershoot. The post-2020 period contains the largest inflation overshoot in forty years.

Core PCE Inflation vs the Fed’s 2 Percent Target, 2010-2025

Annual core PCE inflation rates approximated to whole-decimal precision based on Bureau of Economic Analysis monthly data. Target band of 1.5%-2.5% reflects the FOMC’s communicated comfort zone around the 2 percent point target. Not a substitute for source data from BEA.

Three patterns are visible in the chart. First, core PCE inflation spent almost the entire 2010-2019 decade in the lower half of the target band or below it, with only one annual reading at the 2 percent point itself. Second, the overshoot that began in 2021 was much larger than anything the framework had been designed to handle: a peak of 5.5 percent in 2022 against a target of 2 percent. Third, the post-peak path shows a gradual but incomplete return to target, with the 2025 reading still above the band. The asymmetry that motivated AIT did not prevent the symmetric problem of how to respond to an unexpected upside shock from emerging.

The 2021-2023 Stress Test

AIT was adopted in August 2020 against the backdrop of a decade of inflation undershoots. Within a year, the framework faced the opposite problem. Reopening after the pandemic, fiscal stimulus, supply chain disruption, and energy shocks combined to push inflation to its highest level since the early 1980s. The FOMC’s initial response was framed explicitly in AIT terms: the overshoot was characterized as “transitory” and as partly desirable given the prior decade of undershoot. The Fed kept policy rates at zero through most of 2021 even as inflation rose past 5 percent.

By the end of 2021, the transitory framing was no longer credible. Inflation expectations were beginning to drift upward, the labor market was tightening sharply, and the FOMC pivoted in December 2021 toward an aggressive tightening path. The federal funds rate moved from 0-0.25 percent in March 2022 to over 5 percent by mid-2023, the fastest tightening cycle in decades. Several FOMC participants subsequently acknowledged that the AIT framework had contributed to the slow initial response. The framework’s bias toward tolerating overshoots after undershoots, combined with the language of “shortfalls” from maximum employment, made it harder for the committee to recognize that the overshoot had become non-transitory.

Critics, including former Treasury Secretary Lawrence Summers and former IMF Chief Economist Olivier Blanchard, argued that the framework had not just failed to prevent the overshoot but had actively delayed the response. Defenders, including Vice Chair Lael Brainard and several Fed staff, argued that the overshoot was driven by supply shocks that no monetary framework could have prevented, and that the AIT structure helped preserve the credibility of the 2 percent anchor when expectations might otherwise have unmoored. Both views have analytical merit, and the verdict on the framework’s first stress test remains contested as of 2025.

The 2025 Framework Review

The Federal Reserve committed in 2020 to reviewing its framework every five years. The first such review, completed in 2025, addressed the lessons of the 2021-2023 episode directly. The conclusions of the review, published by the Fed in August 2025, retained the core commitment to averaging inflation at 2 percent over time but adjusted the framework’s tilt in several ways.

First, the 2025 statement de-emphasized the “shortfalls” framing for employment. The new language treats deviations from maximum employment as a concern again, restoring some of the symmetry that the 2020 statement had removed. Second, the make-up strategy language was softened: the commitment to aim for “moderately above 2 percent” inflation following undershoots was retained but with explicit acknowledgement that the appropriate response depends on the size and persistence of any miss, and on whether expectations remain anchored. Third, the review acknowledged that the AIT framework, as originally specified, was designed for a low-inflation regime, and that its application during the post-2021 inflation surge had required more discretionary judgment than the framework could provide.

The 2025 framework is therefore best described as a softened version of AIT rather than a return to strict inflation targeting. The Fed continues to express its goal as an average of 2 percent over time, but the makeup commitment is more conditional, the employment side is more symmetric, and the FOMC’s discretion to depart from the framework in response to large shocks is more explicitly recognized. Other central banks watching the US experience have generally chosen not to adopt formal AIT in their own frameworks, with the European Central Bank revising its target to a symmetric 2 percent in 2021 without the make-up language, and the Bank of England retaining its standard 2 percent point target.

The Choice of Inflation Measure

The technical details of how inflation is measured become unusually important under AIT. The Fed targets the core PCE price index rather than the headline PCE or the more commonly cited CPI. Core CPI and core PCE differ in coverage, weighting, and methodology. The PCE index uses a chain-weighted methodology that adjusts more quickly to changes in consumer spending patterns, includes a broader range of goods and services, and historically runs about 0.3 to 0.4 percentage points below CPI on average.

This matters because the AIT make-up calculation depends on which measure is used. If the Fed measures the prior undershoot using core PCE and finds inflation averaged 1.6 percent for a decade, the implied make-up overshoot is one figure. If it were measured against headline CPI, it would be a different figure, possibly substantially different. Other central banks that have considered AIT in their own frameworks have had to confront the same question: which inflation measure is the operational anchor, and how should the average be computed across measures that move differently?

The Fed’s choice of core PCE as the primary measure was made for reasons that predate AIT. The 2020 framework simply inherited the existing operational target. But under AIT, the choice has framework-design implications it did not have under standard inflation targeting, because the average requires a longer-run measure that is consistent across decades and that resists short-term volatility from food, energy, and other transitory components.

The Verdict on AIT, So Far

The honest assessment of AIT as of 2025 is mixed and contingent. The framework was designed to address a real problem (the asymmetry of inflation targeting at the lower bound) and was based on serious analytical work by Fed staff and external academics. Its adoption was a legitimate response to a decade of undershoots. The framework also faced an unusually severe stress test within one year of adoption, and the test exposed limitations that no one had fully anticipated.

The framework’s defenders point out that long-run inflation expectations remained anchored at around 2 percent through the entire 2021-2023 episode, which is what the AIT structure was designed to deliver. Five-year, five-year forward inflation expectations from TIPS markets stayed in the 2.0 to 2.4 percent range throughout the inflation surge, well below where they might have moved without a credible framework. By this metric, AIT performed its core function.

The framework’s critics point out that the cumulative inflation overshoot from 2021 to 2024 went far beyond what any reasonable make-up calculation would have prescribed. The 2010-2019 undershoot averaged about 0.4 percentage points below target. The 2021-2024 overshoot averaged more than 1.5 percentage points above target. The asymmetry that AIT was designed to remove now exists in reverse: the price level is meaningfully above the 2 percent path, and the framework provides no mechanism for bringing it back down. Whether this matters depends on whether the price level itself is the relevant anchor or whether the inflation rate is.

What seems clearer from the experience is that AIT in its 2020 form was a framework for an undershoot world that had to operate in an overshoot world. The 2025 revisions reflect the lessons of that mismatch. Whether the framework continues to evolve, whether other central banks adopt similar approaches in the next round of reviews, and whether the next low-inflation episode reveals that AIT’s diagnosis was right after all, are open questions for the rest of the decade.

Explains

Conclusion

The average inflation targeting framework adopted by the Federal Reserve in August 2020 was the most significant change to the monetary policy framework of a major central bank in more than two decades. It was designed to address a structural problem: the asymmetric risk of persistent inflation undershoots when the policy rate approaches zero. By committing to make up past undershoots with periods of inflation moderately above target, the Fed sought to keep long-run inflation expectations anchored at 2 percent rather than letting them drift down.

The framework’s first stress test came earlier than expected and in the opposite direction. The 2021-2023 inflation overshoot was larger than the prior decade’s undershoot and exposed the framework’s discretionary character. The 2025 review softened the make-up commitment, restored more symmetric language on employment, and acknowledged that the framework, as originally specified, had not anticipated a large supply-driven overshoot so soon after adoption. The current US framework is best described as AIT with more conditional make-up language and a less aggressive employment bias than the 2020 version. Whether AIT remains the right framework for the next decade depends on whether the structural conditions that motivated it (low r-star, periodic encounters with the lower bound, persistent undershoot risk) reassert themselves once the current inflation cycle fully unwinds.

Frequently Asked Questions

Thanks for reading! If you found this helpful, share it with friends and spread the knowledge. Happy learning with MASEconomics