U.S. Consumer Confidence: Conference Board vs. Michigan Sentiment

News reports often call both releases “consumer confidence,” but that label describes a broad subject, not one shared statistic. The Conference Board's Consumer Confidence Survey and the University of Michigan Surveys of Consumers are independently designed surveys with their own questions and composite indexes. The Conference Board publishes the Consumer Confidence Index, Present Situation Index, and Expectations Index. Michigan publishes the Index of Consumer Sentiment, Index of Current Economic Conditions, and Index of Consumer Expectations. Each index summarizes answers from its own survey; neither is an official government count of purchases or household income.

The different names also point toward different emphasis. The Conference Board index includes assessments of current business and employment conditions and expectations six months ahead for business conditions, employment, and family income. Michigan's five sentiment questions cover household finances, near- and longer-run views of national business conditions, and whether it is a good time to buy major household goods. The two organizations explain these components in their [survey technical note]({source:conferenceBoardConsumerConfidenceTechnicalNote}) and [index-calculation documentation]({source:umichConsumerSentimentIndexCalculations}).

As a result, one month's releases can point in different directions without either being wrong. Respondents might become more positive about jobs today while worrying more about household finances over the next year. Each index gives those answers a different role and horizon. Before interpreting a headline, identify which survey, component, reference month, and data vintage the number describes.

The Conference Board Consumer Confidence Index is built from five questions. Two ask about current business conditions and employment conditions. Three ask about expected business conditions, employment conditions, and total family income six months ahead. The Present Situation Index averages the two current-condition question indexes; the Expectations Index averages the three forward-looking question indexes. The overall CCI averages all five. The [Conference Board technical note]({source:conferenceBoardConsumerConfidenceTechnicalNote}) documents the questions and construction.

Each question offers favorable, unfavorable, and neutral responses. The technical note describes seasonally adjusted response proportions and a relative value based on favorable responses divided by favorable plus unfavorable responses. Each question's value is then indexed to its 1985 average, and the component indexes are averaged into the headline measures. A neutral answer therefore does not enter that relative-value denominator in the same way as a favorable or unfavorable answer.

The index is a summary of survey responses, not the share of households who feel confident. A CCI reading of 100 does not mean that 100% of respondents are optimistic, and a move of 5 index points is not a five-percentage-point change in households' attitudes. The [current series page]({source:conferenceBoardConsumerConfidenceSeries}) identifies the U.S. index and its 1985=100 reference scale. The base provides a comparison point for that series; it does not turn the index into a probability or a spending total.

The University of Michigan's Index of Consumer Sentiment (ICS) comes from five different questions. They ask whether a household's finances are better or worse than a year earlier; whether its finances will be better, worse, or about the same a year from now; what the national economy will be like over the next year; whether the country will have sustained good times or periods of widespread unemployment over roughly five years; and whether now is a good or bad time to buy major household durables. Michigan also publishes separate Current Economic Conditions and Consumer Expectations indexes from subsets of those questions. Its [survey description]({source:umichSurveysConsumerDescription}) explains the broader questionnaire.

For each index question, Michigan calculates a relative score as the percentage of favorable answers minus the percentage of unfavorable answers, plus 100. It rounds the five scores to whole numbers, adds them, divides by 6.7558, and adds 2.0. The constants preserve the series' historical scale and sample-design adjustment. The official [index calculation sheet]({source:umichConsumerSentimentIndexCalculations}) shows the formula and the distinct calculations for the two subindexes.

The main ICS is therefore not just a renamed CCI. It includes a household's own financial history and expected finances, long-run national conditions, and current durable-goods buying conditions. Michigan's [survey description]({source:umichSurveysConsumerDescription}) also makes clear that the sentiment index is only one summary measure from a broader set of monthly questions. The headline does not capture every answer in that questionnaire.

The Conference Board series uses 1985 as its reference year, while Michigan's sentiment index is scaled to its 1966 base period. More importantly, the indexes start from different answers and transform them differently. The Board's question-level relative value uses favorable responses relative to favorable plus unfavorable responses before indexing to its base; Michigan uses a favorable-minus-unfavorable score, adds a constant, and applies its own scaling factor. Equal-looking numbers would not imply equal underlying attitudes.

For example, suppose a chart shows a hypothetical Conference Board reading of 92 and a hypothetical Michigan reading of 58 for the same month. It is not valid to say the Board's value represents more confidence because 92 is greater than 58, or that confidence differs by 34 points. The numbers belong to different scales. A value below 100 means below the relevant index's own base-period level under its construction; it does not mean that fewer than 100% of households are optimistic.

Within-series changes are more informative than cross-series level comparisons, but even a point change is not a percentage change in confidence. A rise from 90 to 93 is a three-index-point increase in the Conference Board measure, not a 3% rise in the number of confident households. To compare the two surveys, describe their direction and components separately, then check whether their question wording, horizon, and release timing explain the difference.

The surveys ask about different time horizons. The Conference Board's expectations components focus on six months ahead. Michigan asks about the household's finances a year ahead, national business conditions over the next year, and business conditions over a longer horizon of about five years. A respondent could be worried about the distant outlook while expecting near-term hiring to hold up, or the reverse. Such a response can affect the two composites differently.

Their current-condition questions differ too. The Conference Board asks about present business and employment conditions. Michigan's current-conditions subindex combines the household's comparison of its own finances with a question about buying major household durables. Changes in perceived job availability, current personal finances, and willingness to make a large purchase need not happen together. Inflation, borrowing costs, income uncertainty, and the timing of planned purchases can influence the answers through different channels.

The index formulas also weight response patterns differently. The Conference Board averages five question indexes, with two in its Present Situation measure and three in Expectations. Michigan's headline calculation combines five rounded relative scores, while its two subindexes use their own question groupings and scaling constants. Because the answers, groupings, sample designs, weighting, and scale conversions are distinct, the two monthly changes should not be averaged or subtracted into a new “true” confidence number.

Sampling variation and release conventions are additional reasons to avoid reading too much into a single month. Both measures estimate attitudes from survey responses, so small changes may partly reflect which households answered. Release timing, preliminary versus final values, seasonal adjustment, and later revisions should be checked in the original release. Michigan's [FAQ]({source:umichSurveysConsumerFaq}) directs readers to its technical documents for the index questions and construction; The Conference Board's technical note describes its own response adjustment and publication process. A comparison should keep each series' vintage attached.

<!-- learn:illustration -->

Imagine, purely as a hypothetical, that the Conference Board's Expectations Index falls by four points while Michigan's ICS rises by two. Do not collapse those changes into an average increase or declare one survey mistaken. The first question is which component moved: the Board's index may reflect a six-month outlook for jobs, business, or family income, while Michigan's headline can also respond to current household finances, durable purchases, and longer-run national expectations.

Next, inspect the component tables and underlying response shares rather than relying only on the headline. Did current conditions improve while expectations weakened? Did households become less negative, or did favorable answers actually increase? A composite can move because different components offset one another. The Michigan formula's favorable-minus-unfavorable score and the Conference Board's relative-value transformation are not interchangeable response percentages.

Then compare each survey with data that measure the outcome you care about. If the question is how much households spent, examine actual expenditure series such as the BEA's [Personal Consumption Expenditures]({source:beaPersonalConsumptionExpenditures}) rather than treating confidence as spending. If the question is hiring or job availability, compare survey expectations with labor-market releases and their definitions. A sentiment index may add context about perceptions or intentions, but it does not identify the cause of a change in realized sales, employment, or income.

Consumer attitudes can help describe how households view their finances and the economy, and the Michigan survey explicitly studies how attitudes relate to decisions such as saving, borrowing, or discretionary purchases. But a survey answer is not a purchase record. A household can report concern and continue spending because groceries, utilities, or rent are necessary; another may be optimistic yet postpone a car or appliance purchase because credit is expensive. Prices, income, debt, wealth, credit availability, and household circumstances all affect spending.

The same caution applies to recession claims. A weakening expectations index can be a warning worth investigating, but one threshold or monthly fall does not establish that a recession has begun or will occur. A composite cannot by itself reveal what caused respondents' views to change, whether a change is temporary, or how firms and households will act. Recession dating and economic assessment require a broader set of realized data and a defined analytical method.

Use survey indexes as one input alongside employment, income, production, and spending data. Match the question to the measure: a broad sentiment headline can summarize attitudes, while a component can focus attention on current conditions or expectations. Keep the survey's reference month and release vintage visible, and avoid using a survey index as an investment trigger or a forecast with a guaranteed hit rate.

Use the Conference Board Consumer Confidence Index when you want a composite built from current business and employment assessments and a six-month outlook for business, jobs, and family income. Its Present Situation and Expectations indexes help separate those two horizons. Use Michigan's Index of Consumer Sentiment when you want a five-question composite that also incorporates household financial comparisons, durable-goods buying conditions, and national outlooks over multiple horizons. Its Current Economic Conditions and Consumer Expectations subindexes reveal a different split.

For a direct comparison, write down the index name, component, reference month, publication date, data vintage, and scale before comparing changes over time. Compare each series with its own history rather than comparing the raw level of one index with the other. Read the question definitions before translating a change into a claim about shoppers, workers, or the economy.

If the practical question is actual consumer spending, follow survey context with BEA PCE or Census retail-sales measures. The guide to retail sales versus PCE explains why those realized-spending series have different coverage. For price changes that can shape household perceptions, see CPI, PCE, and the GDP deflator. For the labor-market backdrop, see CES payroll jobs versus CPS household employment. These comparisons keep attitudes, purchases, prices, and employment in their proper statistical lanes.