Inflation expectations are beliefs about the future rate of inflation over a specified horizon. For a quant, they help separate the outlook available when a position was taken from the price changes eventually observed. They also matter economically: expected purchasing-power losses can influence wage negotiations, firms’ pricing and borrowing decisions. Beliefs can affect outcomes without determining them. Cleveland Fed
An expectation is neither today’s measured inflation nor a promise about future inflation. A forecast error can be assessed only after the relevant outcome becomes available, using a compatible price measure and period.
One survey, several summaries
Suppose 100 fictional, equally weighted respondents each give a single forecast for the percentage increase in a stated consumer price index over the next 12 months:
| Forecast | Respondents | Share of respondents |
|---|---|---|
| 2% | 50 | 50% |
| 3% | 30 | 30% |
| 5% | 20 | 20% |
The cross-sectional mean forecast is
No respondent gave that exact forecast. It is a summary of their answers, not an additional respondent and not the inflation rate that will necessarily occur.
The sample median answers a different question. Order the 100 answers. Observation 50 is 2%; observation 51 is 3%. Using the usual even-sample midpoint convention, the median is . State the convention: survey publishers may use weights or interpolation, and a mean and median are not interchangeable.
Original fictional point forecasts for one common 12-month horizon. Bar lengths show shares of respondents, not probabilities assigned by one respondent. The sample median uses the midpoint of ordered observations 50 and 51.
For the weighted-average mathematics, Expectation, Variance and Covariance develops expectation and spread. The full primer topic is free with a signed-in account.
Disagreement is not personal uncertainty
The table describes disagreement across people. It does not tell us how sure any person is. Respondents could give the same point forecasts while assigning very different probabilities to alternative inflation outcomes. Conversely, everyone could report the same point forecast while each remains highly uncertain.
A personal probability distribution assigns probabilities to outcomes for one respondent. A cross-sectional distribution counts or weights different respondents’ forecasts. Replacing one with the other changes what is being measured.
The New York Fed’s SCE makes this distinction operational: it asks for probabilities over inflation ranges, fits each respondent’s distribution and reports a median across the resulting individual means. Its direct point-prediction measure is separate. The chart guide distinguishes the spread across individual expected rates from uncertainty within an individual’s fitted distribution. Our fictional table reproduces neither official estimator. SCE FAQ, chart guide, inflation and uncertainty definitions
Read the label before comparing the number
Record the population, question wording, price concept, horizon and aggregation rule. Consumers, firms and professional forecasters need not report the same beliefs. A question about inflation can elicit a different response from one about prices a household pays. The US SCE is monthly and surveys a rotating panel of household heads; its three-year-ahead question concerns the one-year change starting two years from now, rather than cumulative inflation over three years. SCE FAQ
Survey expectations and market inflation compensation are also different measurements. A nominal-government-bond yield minus a comparable inflation-linked-bond yield is commonly called breakeven inflation. It embeds inflation expectations alongside inflation-risk compensation and relative liquidity effects. A change in that spread need not be an equal change in expected inflation, and extracting a “pure” expectations component requires modelling assumptions. San Francisco Fed
For a historical trading study, save the survey’s fieldwork period, release timestamp and available vintage. Join a release to the information set available at the decision time; the survey month alone does not establish when a trader could use it.
What “anchored” adds
Longer-horizon expectations are described as anchored when they remain consistent with the monetary-policy anchor and do not react excessively to temporary shocks. One stable mean is incomplete evidence: disagreement, personal uncertainty, tails and the response to shocks can change underneath it. The ECB recommends assessing these dimensions across different groups. Anchoring is an assessment of beliefs, not a guarantee that realised inflation will stay on target. ECB, Economic Bulletin Issue 5/2026
Sources
- New York Fed — SCE Frequently Asked Questions: survey population, monthly collection, forecast horizons, point forecasts and density means.
- New York Fed — SCE Interactive Chart Guide, inflation charts 1–2: disagreement versus individual uncertainty.
- Cleveland Fed — Measures of Expected Inflation and Why They Matter: decisions, survey differences and model-based measures.
- San Francisco Fed — The Increase in Inflation Compensation: What’s Up?, “A yield curve model of nominal and real yields”: expectations, inflation-risk and liquidity components.
- ECB — Capturing inflation expectations (de-)anchoring: level, higher-moment and shock anchoring.