Reference 12 June 2026 · 10 min

A plain-English guide to inflation expectations surveys.

The four surveys central banks watch — Michigan, NY Fed SCE, the Conference Board, and the SPF — what each one asks, why the Fed weights them differently, and the "anchored vs unmoored" framing that today’s Michigan long-run jump just put on the table.

By Igor Komolov

Four surveys produce the public, US-published inflation expectations that the Federal Reserve reads alongside market-implied measures like breakevens. Each survey asks a slightly different question, of a slightly different population, at a slightly different cadence. They do not all move together. Today's University of Michigan preliminary release moved its 5-year measure from 3.5 to 3.9 percent in a single month, which is the kind of jump that puts the "anchored expectations" framing we wrote about in today's analysis directly at risk. This is the explainer on what each survey is, why they diverge, and which of them the Committee actually pays attention to.

The four surveys, side by side

University of Michigan Survey of Consumers. Released twice monthly (preliminary on the second Friday of the month, final on the last Friday). Samples roughly 500 households, asks them about expected price changes over the next year and the next 5 to 10 years. Published since 1946; the 5-year measure has been published since 1980. The 1-year expectation is the noisiest of all four major surveys; the 5-year is much more stable and is the one the Fed has cited most often in speeches.
NY Fed Survey of Consumer Expectations (SCE). Monthly, since 2013. Samples roughly 1,300 households on a rolling 12-month basis. The methodology differs from Michigan in that SCE asks for a probability distribution ("what is the chance inflation is between 0 and 2 percent next year?") rather than a point estimate. The result is a smoother, less noise-sensitive series than Michigan. The SCE 1-year and 3-year measures are the median across respondents.
Conference Board Consumer Confidence. Monthly, since 1967. Samples roughly 3,000 households; the inflation question (a single-point estimate of 12-month expected inflation) is one of several supplementary questions. Conference Board is the most labour-market focused of the four; the inflation print from it tends to mirror Michigan's 1-year measure with a small lag and higher noise.
Survey of Professional Forecasters (SPF). Quarterly, since 1968 (originally ASA-NBER, since 1990 the Philadelphia Fed). Samples roughly 40 professional economists. Reports the median, mean, and a probability distribution for CPI and PCE inflation over the next 1, 5, and 10 years. SPF is the slowest-moving of all four surveys and the most stable; it sits very close to the Fed's 2 percent target on long-run measures in most periods.

The picture, four measures together

Four 5-year (or 3-year for NY Fed) inflation expectation measures, last 24 months. The Michigan series spikes at the right edge on today's print; the NY Fed series is much smoother; the SPF essentially does not move; the market 5y5y forward is the calmest of all. The dashed line at 2.0 is the Fed target. Illustrative monthly observations modelled on the published series. Chart by TradingFuse.

Why they diverge in any given month

Question framing. Michigan asks "what do you think the rate of inflation will be over the next 5 to 10 years?" with a point answer. NY Fed asks the household to assign probabilities to ranges. The two methodologies capture different things; Michigan over-weights respondents with strong views, NY Fed weights the whole distribution.
Sample composition. Michigan samples consumers; SPF samples economists. The same data print moves the two surveys in different magnitudes, because economists update against the Fed framework and consumers update against gas prices.
Question recency. Conference Board is sampled mid-month, Michigan preliminary is the second Friday, NY Fed releases for the prior month. A high CPI released between the surveys will show up in one before another.
Headline vs underlying. Michigan consumers respond strongly to gasoline prices, which are a one-off shock channel. Economists in the SPF strip out energy when forming long-run views. The two series can move 30 to 50 basis points in opposite directions on the same set of releases.

Which one the Fed actually weights

The Committee has cited all four surveys in speeches at different points. There is a soft hierarchy in how speeches treat them.

SPF and the market-implied 5y5y are the two anchors. When both stay near 2 percent, the Committee says expectations are "anchored". A divergence between SPF and 5y5y is the alarm bell on differential information.
NY Fed SCE sits in the middle. It is cited as a household-side cross-check on SPF; it is more stable than Michigan and lets the Committee distinguish true unmooring from a single-month spike.
Michigan and Conference Board are noisier and used more for narrative than for policy calibration. The Committee will cite Michigan when the consumer is front-of-mind in the public discussion (gasoline price cycles, food inflation, post-pandemic surveys), but it will not pivot policy on a single Michigan print.

"Anchored" vs "unmoored", operationally

The Committee uses "anchored" to mean: the public expects the central bank to bring inflation back to 2 percent eventually, and is willing to discount short-run inflation surprises in forming long-run expectations. "Unmoored" means the public no longer holds that expectation; near-term inflation surprises start to feed through into long-run expectations.

Three operational thresholds desk strategists use:

SPF long-run above 2.5 percent for two consecutive quarters. The cleanest single signal of unmooring. It has not happened in the post-Volcker sample.
Market-implied 5y5y above 2.5 percent for one month. Easier to cross than SPF; the 2022-2023 episode crossed it briefly and returned.
Michigan 5-year above 3.5 percent. Often noted in public commentary but treated by the Committee as a Michigan-specific signal. The post-pandemic peak was 3.3 percent. Today's 3.9 percent is the highest in the Michigan time series since 1995.

Today's Michigan print crosses the third threshold by a wide margin. It does not cross the first two. The Fed's response function will weight that combination carefully.

What surveys do not tell you

What the survey respondents will actually do. Inflation expectations matter for policy because they may feed into wage demands and price-setting behaviour. The surveys measure what people say, not what they do. The transmission from "expressed expectation" to "behavioural response" is slow and incomplete.
Why expectations have moved. A 40 basis-point Michigan jump can come from gasoline, from tariff coverage, from an FOMC press conference, or from news cycles. The survey reports the level, not the cause.
The full distribution. Median expectations hide tails. A 4 percent median with a 10 percent right tail (households expecting 10 percent inflation) is a different story from a 4 percent median with a tight distribution. NY Fed SCE publishes the distribution; the other surveys do not.
Cross-asset implications. Surveys are slow-moving; FX and bond markets price expectations continuously. The surveys are reads on the consumer; they are not the readout the trading desk uses for positioning. Breakevens are that read.