Analyzing Consumer Sentiment: A Leading Economic Indicator

Consumer sentiment surveys ask households how they feel about their finances, jobs and the economy, and markets watch them because attitudes can shift before official spending data does. The record is mixed: sentiment is published early and is sometimes informative, but a 2025 Kansas City Fed study found it added little to forecasts of household spending once past spending and income were known.

Key takeaways

The University of Michigan index reached a record low of 44.8 in May 2026, just under the previous low of 50.0 in June 2022; the preliminary October 2026 reading was 46.3.
In a June 2024 release the Conference Board called an Expectations Index below 80 "the threshold which usually signals a recession ahead"; its September 2026 reading was 63.6, while its Present Situation Index was 109.3. Readings below 80 in 2024 were not followed by an officially dated recession.
Households can report gloomy sentiment while still spending, as a 2025 Federal Reserve study of verified purchases found for 2024.
A Kansas City Fed analysis found that adding sentiment to its spending model moved the forecasts for the next two quarters of 2025 by less than half a percentage point.
The two surveys use different scales and methods, so their levels cannot be compared with each other.

What the two surveys measure

The University of Michigan Surveys of Consumers publishes an Index of Consumer Sentiment, built from two parts: Current Economic Conditions and the Index of Consumer Expectations. Until mid-2024 the survey interviewed people by cell phone. According to the university's April 2024 announcement, it moved to web interviews with address-based sampling in stages from April to July 2024, with a target of about 900 to 1,000 interviews a month, up from about 600 by phone. The university reported a 0.97 correlation between phone and web readings over seven years of parallel collection.

The Conference Board Consumer Confidence Index (1985 = 100) has a different design. Its May 2021 technical note says the survey went online that month, uses a panel run by Toluna, collects about 3,000 responses a month, and applies weights for age, gender, income and geography. It rests on five questions: views of current business and employment conditions, plus expectations six months ahead for business conditions, employment and income. For each question, the share of positive answers is divided by the sum of positive and negative answers and compared with the 1985 average, so a reading above 100 means views are more favorable than in 1985.

Both organizations split their headline into a present-conditions measure and a forward-looking one. The Conference Board labels these the Present Situation Index and the Expectations Index. Because the surveys differ in questions, scale and sampling, a Michigan number of 46 and a Conference Board number of 82 say nothing about each other; only changes within each series, and the direction of each, can be compared.

Where sentiment sits among leading indicators

Consumer expectations are formally part of one composite. The Conference Board's Leading Economic Index has ten components, and the tenth is average consumer expectations for business conditions. The other nine include weekly unemployment claims, building permits, the S&P 500, the Leading Credit Index and the spread between the 10-year Treasury yield and the federal funds rate. Sentiment is one input among several, not the whole indicator.

Recent readings and record lows

The Michigan monthly series, available as a CSV from the survey and on FRED (series UMCSENT), begins in 1952. June 2022 was a low of 50.0 at the peak of post-pandemic inflation. The survey's July 2026 special report says sentiment reached a new historic low in May 2026, just below that earlier trough.

Reading	Value	Note
Michigan, June 2022	50.0	Previous record low
Michigan, January 2025	71.7
Michigan, May 2026 (final)	44.8	Record low in a series that starts in 1952
Michigan, September 2026	48.1
Michigan, October 2026 (preliminary)	46.3	Current conditions 44.7; expectations 47.3
Conference Board, September 2026 (preliminary)	81.9	Present Situation 109.3; Expectations 63.6

Data as of October 17, 2026. The final Michigan October reading is due October 23 and the next Conference Board release October 27, so the latest rows will change.

Simple arithmetic on those figures shows the scale. May 2026 sits 5.2 points below the June 2022 low and 26.9 points below January 2025. The preliminary October 2026 reading is 7.3 points (13.6%) below the 53.6 recorded in October 2025, and 1.8 points below September. Michigan attributes the recent weakness mainly to cost-of-living concerns, and its October 9, 2026 report ties them to gasoline prices after a Middle East conflict began on February 28, 2026.

The survey's own check on the low matters. Michigan ran an experimental web survey alongside the official phone survey from 2017, and the July 2026 report puts the June 2022 web reading at 46.3, against 50.0 by phone. May 2026, at 44.8, is below both, so the record is not a product of the switch to web interviewing.

The expectations gap in the data

The "expectations gap" is the distance between how people rate conditions now and what they expect over the next six months. The Conference Board's September 2026 release gives a clear example: the Present Situation Index was 109.3 and the Expectations Index 63.6, a gap of 45.7 points. Both indexes fell that month (by 7.9 and 5.9 points), and the Expectations Index had declined for a third month in a row.

The Conference Board describes the 80 level in its June 2024 release as "the threshold which usually signals a recession ahead," and noted that the index had then been below 80 for five straight months. The wording is "usually," so it is a rule of thumb drawn from past cycles. It is not a statistical forecast, the Conference Board does not attach a probability or a timeline to it in that release, and its September 2026 release does not repeat it.

The rule applies to the Expectations Index, not the headline index, and it has misfired: as of October 2026, the NBER business-cycle chronology, the standard reference for US recession dates, lists no peak after February 2020, so the 2024 run below 80 was not followed by an officially dated recession.

The gap can also point the other way. In Michigan's preliminary October 2026 data, current conditions (44.7) were below expectations (47.3), reversed from September when current conditions were 50.9 and expectations 46.3. Current conditions fell 6.2 points in a month while expectations rose 1.0. A narrowing or reversal of the gap shows that the two halves respond to different things, such as prices people pay now versus hopes for later, and each should be read on its own.

Why mood and spending diverge

The older academic result is that sentiment adds only a modest amount beyond what is already known. Carroll, Fuhrer and Wilcox (1994, American Economic Review; working-paper version) found, in a Reserve Bank of Australia summary, that past Michigan readings "contributed marginally" to predicting household spending once past spending and labor income were included. The Kansas City Fed's May 2025 Economic Bulletin describes the same paper as finding that sentiment "adds little additional information."

The bulletin's own test compared a baseline model (past spending and labor income growth) with a version that included Michigan and Conference Board measures. For the next two quarters of 2025 the forecasts were within half a percentage point of each other, and the authors concluded that sentiment data "have not substantially improved forecasts of consumer spending growth." The baseline model was estimated on 1967 to 2019 data.

The inflation effect

A 2025 Federal Reserve Board note using verified retail purchases of about 9,200 panelists describes the gap in its summary: "consumers say they feel worse, but through the end of 2024, they are buying more." In the survey, 43% said they were doing worse than in 2019 and 25% better. The note reports that, since mid-2022, sentiment has tracked price concerns more closely than income concerns, and that respondents were more likely to overestimate than to underestimate the inflation they had actually experienced. The note's data suggest the gloomy feeling is real while its link to purchasing is loose.

A St. Louis Fed FRED blog post from September 2026 shows a similar split. The Michigan index averaged 95.9 in 2019 and stood at 71.7 in January 2025, while the share of adults telling the Fed's household survey they were at least "doing okay financially" was 75% in 2019 and 73% in 2025.

Intentions can still show stress

Divergence does not mean spending is unaffected by price shocks. In the October 9, 2026 report, based on 3,492 interviews from June 23 to September 21, 2026, Michigan found only 31% of consumers expecting to spend as usual on items with large price increases, versus 37% in July to October 2022 and 26% in April to July 2025; 54% expected to cut back and 16% to stop. The answers differed by income: 20% of the lower third expected to spend as usual, against 42% of the upper third. Those are stated intentions, not recorded purchases, and the report itself notes consumers may cut some items without reducing total spending.

How investors use the readings

Sentiment reports are usually treated as one input among many. They arrive earlier than retail sales or GDP, and the Conference Board's technical note calls the index one of the earliest monthly indicators. But the Kansas City Fed result above shows that adding sentiment has not substantially changed spending forecasts. Other releases, such as unemployment data, inflation numbers, interest-rate decisions and earnings reports, cover different parts of the picture.

Three reading habits follow from the evidence in this guide:

Change versus level. Month-to-month changes include ordinary sampling variation, as Michigan's 2024 announcement notes, and Michigan itself describes the 1.8-point October dip as "little changed."
Components over headline. The headline can hide opposite moves, as in October 2026, when current conditions fell while expectations edged up.
Behavior alongside attitudes. Retail sales and verified purchase data show what households did, which sentiment surveys cannot.

None of this makes sentiment a timing tool. The sources here describe how the indicator behaves and give no basis for predicting market direction.

Limits of sentiment data

Methodology changes affect long comparisons. Michigan's 2024 announcement advised users to be more conservative reading month-to-month changes between March and July 2024, because those months combined real change with method effects. The same document notes that web respondents tend to answer somewhat less favorably than phone respondents; Current Conditions, for example, ran consistently lower on the web while moving in parallel. The Conference Board applies adjustment weights to its online panel to offset the positive bias commonly seen in online panels, and says it will keep evaluating them.

Politics also colors answers. Michigan's May 2026 report says partisan gaps in sentiment are larger than a year earlier, but that national readings align closely with the views of independents. The July report gives May 2026 readings of 32.8 for Democrats, 40.6 for independents and 84.6 for Republicans, against a national 44.8, and adds that the alignment holds under both phone and web collection. Michigan's October report cites research linking partisanship to actual spending, so the gaps are not dismissed as noise.

Finally, different groups feel different things. Beyond the income split above, the October 9 report found that 18% of consumers who own no stock expected to keep spending as usual on items with big price increases, against 50% of those in the top third of stock holdings, so one average can hide a wide spread.

Frequently asked questions

How often are the two indexes released?

Michigan publishes a preliminary and a final reading each month; its twice-monthly schedule is unchanged by the 2024 move to web interviews. The Conference Board publishes monthly, with preliminary results closing about a week before release and late responses folded into the final figure published with the next month's data.

Does the Conference Board count recession worries in its index?

No. Its September 2026 release reports how many consumers think a US recession over the next 12 months is "somewhat likely" or "not likely" (the first share rose that month and the second fell), but says these measures are not included in calculating the Consumer Confidence Index. The index uses the five questions on business conditions, employment and income described in its technical note.

Is the 2026 Michigan low comparable with 2022?

Michigan says yes for this purpose. Its July 2026 special report says May 2026 would still be below the June 2022 trough using either the old cell-phone reading or the experimental web reading that matches today's method. The report does not claim the two readings are identical in every detail.

Why do Republican and Democratic respondents differ?

Michigan says consumers tend to hold more favorable economic views when their party holds the White House, a pattern it traces back to at least the Reagan administration, and that partisan gaps were larger in 2026 than a year earlier. The survey reports that national estimates still track independents closely, so the partisan split does not drive the headline.

How big are the sample sizes?

Michigan targets about 900 to 1,000 web interviews a month under its current design. The Conference Board's technical note says it collects about 3,000 responses a month. Both are weighted to represent the US population, but they sample differently (address-based web sampling at Michigan, an online panel at the Conference Board), so the raw sample sizes alone do not show which survey is more precise.

Checking a new release

A short routine keeps a sentiment release in proportion:

Note the release type (preliminary or final) and the survey dates, since the Conference Board's preliminary cutoff comes about a week before release.
Compare the change with the previous month within the same index only.
Check both halves: present conditions and expectations, and the size of the gap.
Look at what respondents cite (prices, jobs, borrowing costs) in the release commentary.
Set the reading beside hard spending data before drawing any conclusion about demand.

The site's purchasing-power calculator can show how a given inflation rate erodes what a dollar buys, which helps put the cost-of-living complaints in these surveys into numbers.