Consumer Sentiment and Expectations: Michigan, Conference Board and GfK

What you will learn

• How the University of Michigan index is calculated and why its inflation-expectation numbers move bond markets.
• How the Conference Board index differs and what the "labour differential" tells you.
• How the German consumer climate (GfK / NIM) is built and which month it describes.
• How to weigh a sentiment surprise against hard data.

Explanation

Consumer surveys are soft data: they measure what households say, not what they do. Markets watch them because they arrive before spending data and because two of them contain inflation expectations, which central banks read as a sign of whether inflation is "anchored".

University of Michigan Surveys of Consumers

The Michigan survey asks five core questions: personal finances versus a year ago, expected personal finances in a year, expected business conditions over 12 months and over five years, and whether now is a good time to buy major household items. For each question a relative score is built: percentage favourable minus percentage unfavourable, plus 100. The Index of Consumer Sentiment (ICS) adds the five relative scores, divides by 6.7558 (a constant so that 1966 equals 100) and adds 2.0. Sub-indices for current conditions and expectations are also published.

A preliminary reading (usually the second Friday) is followed by a final reading (usually the last Friday) with the whole month's interviews. Since mid-2024 the survey is conducted online rather than by telephone, with roughly 900 to 1,000 responses a month; the switch raised recorded inflation expectations somewhat, so comparisons across that break need care.

Markets care more about two side numbers than about the headline: the median one-year and five-to-ten-year expected price change. A jump in the long-run number is read as a risk that expectations are drifting, which makes a central bank less willing to cut rates.

Conference Board Consumer Confidence Index

The Conference Board survey has several thousand online responses, is released on the last Tuesday of the month and uses 1985 = 100. It has a Present Situation Index and an Expectations Index (business, jobs and income six months ahead); the Conference Board notes that an Expectations Index below about 80 has often preceded recessions, a historical observation rather than a rule. Because it asks whether jobs are "plentiful" or "hard to get", analysts compute the labour differential: percentage plentiful minus percentage hard to get. It moves with the unemployment rate and cross-checks the labour reports from lesson 4.

German consumer climate (GfK / NIM)

The German survey, long known as the GfK Consumer Climate and since 2026 published as the NIM Consumer Climate powered by GfK, uses about 2,000 interviews a month. It publishes four balances in points (positive share minus negative share, so values can be negative): economic expectations, income expectations, propensity to buy and propensity to save. The headline is a forecast of private consumption for the following month, so the late-October release describes November.

Rules of thumb: compare each reading with consensus and with its own one-year range; sentiment reacts to petrol prices and headlines and is a poor short-term spending forecast, so hard data such as retail sales (lesson 7) override it; inflation expectations matter most when they move with actual core inflation (lesson 2).

Worked example

Hypothetical example: a Michigan preliminary release reports relative scores of 95 (finances now), 110 (finances in a year), 120 (business 12 months), 80 (business 5 years) and 105 (buying conditions). Assumptions: constants as published; rounding at the end.

1. Sum: 95 + 110 + 120 + 80 + 105 = 510.
2. Index: 510 / 6.7558 = 75.49; plus 2.0 = 77.5.
3. If consensus was 80.0, the surprise is −2.5 points: a miss, but inside normal monthly noise of 2 to 4 points.
4. One-year inflation expectations rise from 3.0 % to 3.6 %; five-to-ten-year expectations stay at 3.0 %. Because the long-run number did not move, most analysts would treat the rise as a petrol-price effect, not a de-anchoring signal.

In the Conference Board release of the same month, 38 % say jobs are plentiful and 16 % hard to get. Labour differential = 38 − 16 = +22. If this was +30 six months earlier, the labour market is cooling even though the level is still positive.

Market reading: a weak headline with stable long-run expectations is mildly negative for the dollar and supportive for bonds; this tells you which way the surprise points, not what prices will do.

Common mistake

Practice activity

On paper: (a) A Michigan final release reports relative scores of 88, 102, 115, 76 and 99. Calculate the index. (b) German income expectations rose from −5 to +3 while propensity to save rose from +8 to +14. Write one sentence on what this implies for consumption. (c) In the Conference Board survey, 30 % say jobs are plentiful and 20 % hard to get. Calculate the labour differential and compare it with the worked example.

Key takeaways

• Michigan: five questions, preliminary and final readings, inflation expectations that bond traders watch.
• Conference Board: labour differential as a cross-check on the jobs data.
• German consumer climate: four balances in points; the headline refers to the following month.
• Soft data give early warnings; hard data decide.

Next step

In the next lesson you move from households to firms and analysts with the German and European leading indicators: ifo, ZEW, Bundesbank reports and the euro-area flash inflation estimate.