Can Increased Attention to Inflation Make Inflation More Persistent?
Pfäuti (2026) examines whether people pay more attention to inflation once it becomes sufficiently high and whether this increased attention amplifies inflation. Using U.S. inflation data and household inflation expectations from the University of Michigan Survey of Consumers from 1978–2024, he estimates that attention nearly doubles when annualized inflation crosses a threshold of about 4%, rising from 0.18 to 0.35. He finds that supply shocks increase inflation by about 40 basis points when attention is high, compared with 20 basis points when attention is low. Increased attention can therefore magnify inflation surges and make inflation more persistent.
Why This Article Was Selected for The Policy Scientist
Inflation affects purchasing power, household planning, business investment, wages, interest rates, and the broader stability of the economy, making its persistence especially consequential. This study is particularly timely after the recent inflation surge demonstrated how quickly expectations can change. The long U.S. data series is a major strength, and the findings may generalize to other advanced economies with similar inflation dynamics. The article also advances the extensive literature on inflation expectations by identifying a specific threshold at which attention changes. Its threshold estimation and causal analysis of externally identified oil-supply shocks provide stronger evidence than conventional multivariate regression, although additional quasi-experimental and experimental evidence would further strengthen causal interpretation.
Full Citation and Link to Article
Pfäuti, O. (in press). The inflation attention threshold and inflation surges. American Economic Review. American Economic Association.
Central Research Question
The study asks when people begin paying substantially greater attention to inflation and whether this change in attention alters the behavior of inflation itself. The central proposition is that attention does not necessarily increase smoothly as inflation rises. Instead, people may remain relatively inattentive while inflation is low and stable, then become markedly more attentive after inflation crosses a threshold. The analysis therefore examines both the existence and location of such an “inflation attention threshold” and the macroeconomic consequences of crossing it.
A related question concerns inflation persistence. Greater attention causes households and firms to update their inflation expectations more strongly when actual inflation changes. Those expectations may then influence price and wage decisions, potentially magnifying the effects of inflationary shocks. Conversely, once inflation falls below the threshold, attention declines and expectations adjust downward more slowly. This mechanism could help explain both sudden inflation surges and the difficult “last mile” in returning inflation to a central bank’s target.
Previous Literature
The study connects several major strands of macroeconomic research. One begins with models showing that expectations depend on how people process new information. Evans and Honkapohja (2001), Eusepi and Preston (2011), and Marcet and Nicolini (2003) develop approaches in which expectations adjust as people learn from new information. Sims (2003, 2010) provides a foundation for rational inattention: because acquiring and processing information is costly, people rationally devote limited attention to economic conditions. Reis (2006a, 2006b), Mackowiak and Wiederholt (2009), and Paciello and Wiederholt (2014) further develop imperfect-information and rational-inattention approaches.
More recent research examines whether attention itself changes with economic conditions. Korenok, Munro, and Chen (2023), using Google searches across numerous countries, find that attention to inflation increases particularly after inflation passes a threshold; their U.S. estimate is approximately 3.55 percent. Bracha and Tang (2025) similarly find that attention rises with inflation. Weber et al. (2025), drawing on randomized experiments conducted across countries and over several years, provide evidence that households and firms pay greater attention during periods of elevated inflation.
The study also contributes to research showing that economic shocks can have different effects under different macroeconomic conditions. Auerbach and Gorodnichenko (2012a, 2012b), Ramey and Zubairy (2018), and Tenreyro and Thwaites (2016) provide important examples of such state-dependent effects. The distinctive contribution here is to identify attention as a mechanism through which the same inflationary shock can produce different outcomes depending on the inflation environment.
Data
The principal data come from the University of Michigan Survey of Consumers. The baseline sample contains average and median household inflation expectations from January 1978 through May 2024. Respondents report expected price growth over the following year. These responses are transformed into shorter-horizon expectations so that they correspond to the quarterly structure of the macroeconomic model.
Actual inflation is measured using the Consumer Price Index from the Federal Reserve Economic Data database. The primary measure is quarter-over-quarter CPI inflation expressed at an annualized rate. The long time span is important because the United States experienced substantially different inflation environments between 1978 and 2024, allowing comparisons between periods of relatively low and high inflation.
Several additional sources are used for robustness tests. Individual household expectations come from the Federal Reserve Bank of New York’s Survey of Consumer Expectations for 2013–2024. Firm expectations come from the Cleveland Federal Reserve’s Survey of Firms’ Inflation Expectations. Google Trends searches for “inflation” provide an alternative measure of public attention. Regional CPI and survey data permit comparisons across the four major U.S. census regions. To identify supply disturbances, the study uses Känzig’s (2021) oil supply news shocks, constructed from oil-futures movements surrounding OPEC announcements and an instrumental-variable approach to identifying structural oil supply shocks.
Methods
The empirical analysis first estimates a threshold regression in which the relationship between inflation forecast errors and subsequent expectations can differ depending on whether inflation is above or below an unknown threshold. The threshold and the parameters governing attention in each regime are estimated jointly. Importantly, the procedure does not assume beforehand that attention must be higher when inflation is high; that relationship emerges from the data.
Attention is interpreted as the fraction of an inflation forecast error incorporated into revised expectations. An attention estimate of 0.35, for example, implies that a one-percentage-point forecast error produces approximately a 0.35-percentage-point revision in expectations.
The second empirical component examines whether supply shocks have different inflationary consequences in high- and low-attention environments. Local projections estimate inflation responses following externally identified oil supply news shocks. The models incorporate lags of inflation, unemployment, inflation expectations, and the shocks themselves. Regional specifications and alternative measures of attention provide additional tests intended to separate attention from other characteristics of high-inflation periods.
Finally, the empirical findings are incorporated into a New Keynesian model in which attention changes endogenously. This allows the analysis to trace the interaction among inflation, attention, expectations, economic shocks, and monetary policy over time.
Findings/Size Effects
The central empirical result is a pronounced attention threshold at approximately 4 percent annualized inflation. The baseline estimate places the threshold at 3.91 percent. Below this level, the estimated attention parameter is 0.18. Above it, attention rises to 0.35—almost exactly doubling. Thus, after a one-percentage-point inflation forecast error, households revise expectations by roughly 0.18 percentage points in the low-attention regime but approximately 0.35 percentage points in the high-attention regime.
The threshold is economically consequential. Between 1978 and 2024, U.S. inflation was above the estimated threshold approximately 32 percent of the time. The most recent transition into the high-attention regime occurred in early 2021, coinciding with a pronounced increase in public interest in inflation.
Supply shocks also have substantially different effects across attention regimes. A one-standard-deviation adverse oil supply shock increases inflation by approximately 20 basis points when attention is low but approximately 40 basis points when attention is high. The inflationary response is therefore roughly twice as large in the high-attention regime. Moreover, these effects can remain evident for one to two years rather than disappearing immediately.
The 2021–2022 inflation surge illustrates the magnitude of the mechanism. Oil supply news shocks account for approximately half of the increase in inflation between the economy’s transition into the high-attention regime in early 2021 and the inflation peak in mid-2022. The analysis indicates that, without the accompanying increase in attention, the inflation attributable to those shocks would have been only about half as large.
Robustness tests generally support the threshold interpretation. Results remain similar using median rather than mean expectations, quarterly rather than monthly observations, alternative surveys, regional data, different measures of attention, and alternative definitions of the threshold variable. Model-selection tests favor one discrete attention threshold over multiple thresholds or a purely smooth relationship between inflation and attention.
Conclusion
The study concludes that public attention is not merely a passive response to high inflation. Once inflation reaches approximately 4 percent, attention increases sharply, expectations become considerably more responsive to incoming inflation information, and inflationary shocks consequently have larger and more persistent effects.
This mechanism also creates an important asymmetry. As inflation rises above the threshold, heightened attention causes expectations to respond rapidly, potentially reinforcing the initial inflationary impulse. When inflation subsequently falls below the threshold, attention declines. People then revise their still-elevated expectations downward more slowly, contributing to a prolonged return toward the inflation target. The same mechanism can therefore help explain both rapid inflation surges and slow final stages of disinflation.
More broadly, the results imply that the economic consequences of a supply or monetary shock depend partly on the attention environment in which it occurs. Inflation near the attention threshold may behave differently from inflation well below it because a sufficiently large shock can trigger a discrete change in how households and firms process inflation information. The study therefore adds state-dependent attention to the set of mechanisms that can explain why seemingly transitory shocks sometimes develop into persistent inflation episodes.
Comments