S6 Ep14: How should we measure expected inflation?

Episode Summary

Executive Summary: The episode explains why expected inflation is hard to measure and why no single measure—households, firms, professionals, or market-based forecasts—should be treated as definitive. Ricardo Reis argues that different measures capture different information, are noisy in different ways, and should be combined using economic theory. He also shows that households correctly anticipated the 2021 inflation surge better than professionals, especially in the euro area.

Main Topics: Why expected inflation is difficult to measure (Priority: 5/5): Reis explains that expectations are internal beliefs, observable actions, and market prices are all different and only imperfectly aligned with theory. Why no single measure should dominate (Priority: 5/5): He warns against choosing one 'best' measure based on forecast performance alone, arguing that measures should be combined rather than ranked in isolation. Households versus firms and professionals (Priority: 5/5): The conversation contrasts household, firm, banker, and professional forecaster expectations, showing each reflects different parts of the inflation process and economy. The 2021 inflation surprise and household accuracy (Priority: 4/5): Reis highlights that households in Europe and the US picked up the inflation upturn earlier than professionals, making them especially informative during regime shifts. Why past forecasting performance can mislead (Priority: 4/5): He argues that using only historical predictive success ignores changing inflation regimes and confuses statistical fit with the underlying concept of expectations. Policy implications for central banks (Priority: 5/5): Even noisy expectations matter if they influence behavior, reflect overreaction, or signal credibility problems, so central banks should pay attention and sometimes react strongly.

Key Arguments: Expected inflation is not directly observable; researchers must infer it from beliefs, actions, and prices, which often diverge. Household surveys are often dismissed, but Reis argues the standard criticisms are weak and can cause errors. Firm expectations are not automatically superior because firms respond to wages, interest rates, and other expectations embedded elsewhere in the economy. Professional forecasters can be useful in stable inflation regimes, but they often fail during large shifts in inflation, such as the 1970s, the post-2021 surge, and historical disinflations. Forecast accuracy alone is not the right criterion: what matters is whether expectations influence decisions and are relevant to the economic model. Central banks should combine multiple measures of expectations rather than select one winner, just as they do with labor-market indicators. If expectations are noisy but drive actions, central banks must still respond because those expectations help determine inflation outcomes. Households may have appeared especially accurate in 2021–2022 partly because they are sensitive to salient prices like food and energy, but their behavior still matters for macro outcomes.

Data Points: Paper count discussed: 2 CEPR discussion papers - Reis references two papers: one on mistakes in using expected inflation measures and one on euro-area expected inflation and policy responses. CEPR discussion paper number: 17850 - 'Four Mistakes in the Use of Measures of Expected Inflation'. CEPR discussion paper number: 17849 - 'Expected Inflation in the Euro Area: Measurement and Policy Responses'. Expected inflation horizon in stable regimes: Around 2% with some years at 1% or 3% - Reis describes professionals as useful when inflation is stable near target. Time span of some inflation-expectations datasets: 10 to 60 years - Different respondent groups have been surveyed for varying lengths of time. Inflation target/reference level: 2% - Used as the benchmark around which professional forecasters are said to be useful in stable regimes.

Pivotal Quotes: "we have prices, actions, and beliefs, if you want, what's in your mind" — Ricardo Reis: He summarizes the three distinct channels through which expectations are measured and why they rarely align perfectly. "you want to measure the, as I said, less rational agents, which are often the smaller ones" — Ricardo Reis: He explains why smaller, less perfectly rational agents can be more informative than elite experts in some inflation episodes. "The exact same thing applies to expected inflation" — Ricardo Reis: He compares inflation expectations to labor-market indicators and argues for combining multiple measures rather than picking one.

Implications: Policymakers should not rely on a single inflation-expectations gauge. Household surveys, firm data, professionals, and markets each reveal different information, especially during regime shifts, so central banks should combine them and respond when they influence behavior or signal credibility issues.