Why consumer expectations can move in different directions

August 2026’s New York Fed Survey of Consumer Expectations reported rising average chances of higher US unemployment in one year, at 44.4%; falling personal job-loss chances within 12 months, at 13.8%; and falling job-finding chances if the current job were lost, at 45.4%. The supplied release specifies no replacement-search window.1

The first concerns unemployment across the country; the second, losing one’s own job. National deterioration does not require every worker to lose employment. The third asks about finding work only if that personal loss occurs. A lower chance of entering that situation says nothing by itself about prospects once in it: exposure and replacement prospects answer separate questions.1

These percentages average respondents’ assigned chances, bounded between zero and 100. The mean is the arithmetic average. They are neither realized unemployment rates nor percentages answering yes. Assigning a chance allows degrees of confidence instead of a yes-or-no prediction.3

Fieldwork ran August 3–31; the release followed September 8. The nationally representative internet panel includes approximately 1,300 household heads. Participants stay for up to 12 months, with roughly equal numbers entering and leaving each month.12

That design allows researchers to follow the same individuals over time. But the reported averages do not identify which people combined national pessimism with confidence in keeping a job, or whether particular respondents changed those views together. Establishing that pattern would require examining linked responses. Why beliefs changed remains unknown from these aggregates. Logical compatibility is not evidence that particular people held the combination.21

July’s separate SCE Labor Market Survey makes the importance of the question clearer. The average expected likelihood of receiving at least one job offer over the next four months fell to 18.0%. This asks about an offer arriving during a specified period, without posing hypothetical job loss. An offer is not itself replacement employment. The different events, conditions and survey months prevent treating July’s and August’s figures as interchangeable observations or a continuous trend.41

The labor module runs every four months, versus monthly for the core survey. First-time core respondents do not receive that month’s supplements. These New York Fed pages describe one survey program, rather than independent confirmations.432

Inflation extends the same interpretive problem: a summary across people does not describe all the possibilities within each person’s outlook.23

August’s one-, three- and five-year median inflation expectations were 3.6%, 3.2% and 3.0%. Only the three-year measure declined; the others were unchanged.1

The latter two horizons cover one-year changes beginning two and four years ahead, not cumulative inflation. Respondents assign chances to possible inflation outcomes. A fitted distribution—a representation of those outcomes and chances—yields each person’s mean forecast. The reported median is the middle summary across those individual means. A point forecast instead comes from directly asking for a single prediction.3

An outlook can spread chances across a broader range of possible outcomes while keeping a central forecast similar to other people’s. That broader spread expresses greater uncertainty about what will happen; similarity between central forecasts concerns agreement across people. A central forecast alone cannot convey both.32

Disagreement measures the span covering the middle half of respondents’ inflation forecasts. At one year, it fell while median uncertainty rose. These summaries can move oppositely without establishing matched changes for particular respondents.1

Observed prices introduce a related distinction. The Bureau of Labor Statistics defines the Consumer Price Index as average price change over time for a representative basket of consumer goods and services. Its published example shows why the basket matters: someone spending a larger-than-average budget share on medical care may experience higher personal inflation if medical prices rise faster than other basket items. Both conditions matter—the spending share and the relative price movement. A household’s experience can differ from the index without either measurement being wrong. CPI records observed changes; expectations concern future periods. A current price observation cannot settle a forecast before its horizon occurs.51

Report an error · Corrections policy