How memory cues shape inflation beliefs
Gennaioli et al., in this paper, present a model of inflation expectations based on selective memory and test its predictions using survey data. The authors develop a model where people form inflation expectations by selectively recalling past inflation experiences based on:
- The frequency of experienced inflation levels
- Temporal context (recency and primacy effects)
- Numerical similarity to current inflation and the inflation scenario being evaluated
They test the model’s predictions using microdata from two surveys: the University of Michigan’s Survey of Consumers (MSC) and the Federal Reserve Bank of New York’s Survey of Consumer Expectations (SCE).
Their key findings include:
- Selective memory mechanisms significantly influence inflation expectations: People who experienced higher average inflation report higher expectations. Both early-life experiences and recency effects are observed even though people overweight past experiences similar to current inflation.
- The model explains the sharp de-anchoring of inflation expectations in 2021. In particular, it accounts for why elderly people’s expectations rose more sharply than younger people’s.
- Out-of-sample nowcasting: The model successfully predicts the re-anchoring of expectations as inflation fell in 2022-2024.
- Quantitative importance: A one standard deviation increase in similarity based recall measures decreases inflation expectations by 0.18-0.29 percentage points.
Gennaioli et al. conclude that memory-based models, emphasizing similarity, frequency, and interference, offer a promising approach to studying belief formation in macroeconomics and finance.

How Inflation Expectations Dis-Anchor: the Role of Selective Memory Cues
Authors: Nicola Gennaioli, Marta Leva, Raphael Schoenle, Andrei Shleifer
From: Bocconi University, Brandeis University, Harvard University
Household shocks and inflation beliefs
Butera et al. in this paper examine how household-level events influence people’s beliefs about the economy, particularly inflation forecasts and backcasts. The authors link Danish Consumer Expectations Survey data (2012-2019) with administrative tax, health, and other registry data. Survey respondents were aged 25-60 but respondents with significant self-employment income were excluded. Finally, the authors constructed measures of recent and future household income changes using tax data and utilized emergency room (ER) visit data as a proxy for negative household events.
Key findings:
Inflation forecasts are strongly negatively associated with:
- Recent household income changes
- Expected future income changes (survey-based)
- Realized future income changes
- These associations violate rational expectations, as actual inflation shows no relationship with these variables
Role of Memory and Affect
- Inflation backcasts (perceptions of past inflation) show even stronger associations with income changes than forecasts
- ER visits in the survey month significantly increase both inflation forecasts and backcasts
- The effect of ER visits is larger for backcasts than forecasts
Mechanism
- Results suggest associative memory plays an important role in belief formation
- Positive (negative) household events cue recall of positive (negative) experiences, influencing inflation beliefs
- Even unrelated negative events (ER visits) impact inflation perceptions
The authors conclude that these findings challenge theories of limited-information rational expectations and suggest that “experience effects” may reflect a broader psychology where events in one domain influence beliefs in unrelated domains.
Beliefs About the Economy are Excessively Sensitive to Household-Level Shocks: Evidence from Linked Survey and Administrative Data
Authors: Luigi Butera, Chen Lian, Matteo Saccarola, Dmitry Taubinsky
From: Copenhagen Business School, University of California-Berkeley
Similar research on inflation expectations formation can also be found in two recent newsletters:
On experience-based inflation expectations (June 4, 2024)
Public perceptions of inflation (September 19, 2024)

For a quick exploration of this research, we invite you to listen to our accompanying podcast. This audio format offers a narrative that highlights the key insights and findings of all the studies included in the three newsletters.
14 minute podcast