When wording changes what we find: The impact of inflation expectations on spending 5.5.2026 Tiziana Assenza; Stefanie Huber; Anna Mogilevskaja; Tobias Schmidt | Deutsche Bundesbank How do inflation expectations influence household spending? The answer depends on how consumption questions are asked. This study reveals that subtle differences in survey wording can substantially change the estimated relationship between inflation expectations and consumption. The results highlight the importance of careful survey design and interpretation, as measurement choices can lead to misleading empirical conclusions and potentially inaccurate policy advice. Macroeconomic theory offers unambiguous predictions about the role of inflation expectations in shaping household consumption decisions through intertemporal incentives, with important implications for aggregate demand and inflation dynamics. In contrast, empirical studies that measure how households actually respond to changes in inflation expectations reach strikingly different conclusions. Depending on the study, inflation expectations appear to increase consumption, have no effect, or even reduce spending. We investigate in this paper whether the conflicting findings in the literature can be explained by differences in how surveys elicit consumption behavior. To address this question, we embedded a randomization experiment in the Bundesbank Online Panel-Households (BOP-HH). The experiment systematically varies the wording and framing of consumption questions to uncover how the specific design of questions affects the estimated relationship between inflation expectations and spending. To be more precise, the experiment varies the framing of consumption questions along three dimensions: (1) the reference unit (individual vs. household spending), (2) the time horizon (past 1, 3, or 12 months), and (3) the question type (attitudinal, planned spending, qualitative recall-based, and quantitative recall-based). Respondents are randomly assigned to different question formats, allowing us to isolate the causal impact of survey design on the observed relationship between inflation expectations and consumption. Framework: Why question design should matter The paper builds on the premise that survey responses are shaped by how questions are framed, rather than simply revealing fixed underlying behaviors. We hypothesize that framing effects systematically influence the estimated relationship between inflation expectations and spending. One key reason is that consumption questions capture different constructs: attitudinal questions, such as whether it is a good time to buy, capture willingness to spend; recall-based questions measure realized consumption, which is influenced by constraints like income and necessity. We argue that these different constructs – willingness versus realized behavior – are key to understanding the mixed findings in the literature. Key findings: Question design does matter The study reveals four important insights: Question type matters: Attitudinal and planned spending questions yield a negative relationship between inflation expectations and consumption, suggesting that higher expected inflation reduces willingness to spend. In contrast, recall-based questions (both qualitative and quantitative) show a modestly positive or insignificant relationship, reflecting the fact that realized spending is often constrained by factors like income or necessity. This divergence highlights that attitudinal measures capture forward-looking sentiment while recall-based measures reflect past behavior, pointing to distinct economic constructs. Time horizon influences results: The relationship between inflation expectations and consumption is stronger for shorter recall periods (e.g., past 1 or 3 months) than for longer ones (e.g., past 12 months). Shorter horizons are likely to reduce recall bias and make recent purchases more salient, leading to more accurate responses. Reference unit has no detectable impact: Whether respondents report individual or household-level spending does not significantly affect the relationship between inflation expectations and consumption. This finding suggests that durable goods, which are often shared within households, are reported similarly regardless of the reference unit. Conflicting results in the literature are partly due to different questions: Replicating prior studies using the unified dataset from the experiment reproduces the divergent findings in the literature. This demonstrates that differences in survey design alone can explain much of the variation in results. Conclusion Our study demonstrates that survey-based measures of consumption are not interchangeable and that survey design plays a central role in shaping empirical conclusions about the effects of inflation expectations on household behavior. Different question formats capture different economic constructs, and failing to account for this can lead to incorrect inferences. For researchers, our results highlight the importance of aligning survey questions with the specific behavior or concept of interest. For policymakers, the findings raise concerns about the interpretation of survey-based indicators used to assess the strength of the expectations channel of monetary policy. Misinterpreting survey responses may lead to distorted estimates of household responsiveness to inflation expectations.We recommend that future surveys include multiple question formats to capture the full spectrum of consumer behavior. Our results also call for greater transparency in reporting survey methodologies to facilitate comparisons across studies. Assenza T., S. Huber, A. Mogilevskaja, T. Schmidt (2026), When wording changes what we find: The impact of inflation expectations on spending, Bundesbank Discussion Paper, No 13/2026. https://www.bundesbank.de/en/publications/research/discussion-papers/when-wording-changes-what-we-find-the-impact-of-inflation-expectations-on-spending-961174