Michigan versus Conference Board: a field guide to the split
Every few months the University of Michigan sentiment print and the Conference Board confidence print refuse to tell the same story. Commentators pick a favorite. The useful response is to treat the gap as a feature of instrument design.
Michigan’s index is built from five core questions covering personal finances, business conditions, and buying conditions for durables. It has a long academic memory and a reputation for moving with inflation surprises. The Conference Board’s index leans harder on labor-market perceptions — jobs plentiful versus hard to get — which is why it can stay firm while households complain about prices.
When jobs are fine and groceries are not
A classic split appears when unemployment is low and food-and-energy inflation is loud. Michigan, more sensitive to price questions and to buying conditions, sags. Conference Board, more sensitive to job availability, holds up. Equity markets sometimes treat the softer print as a growth scare. Credit markets sometimes treat the firmer print as evidence the consumer is still employed. Both can be true in the same month.
The reverse split shows up near labor turning points. Conference Board can roll over while Michigan, cushioned by falling gasoline prices or a stock-market rally that lifts the “investable” household, looks resilient. If you only watch one series you will eventually be the person explaining why “the consumer” did the opposite of your chart.
Revisions, prelims, and the gossip cycle
Michigan’s preliminary and final prints create a two-step news cycle. The preliminary is tradable because it is first; the final is often closer to the survey’s completed sample. Treating the preliminary as gospel is a choice, not a requirement. CrowdMood timestamps which vintage we store. A signal that cannot say whether it used prelim or final is not a signal. It is a mood board.
Question wording also ages. A “buying conditions for houses” item does different work at 3% mortgages than at 7% mortgages. The series name stays the same; the economic object moves. That is another reason we refuse to mash global surveys into one unlabeled number without a recipe.
A practical checklist
When the two U.S. prints diverge, walk this list: (1) Is the gap in current conditions or expectations? (2) Did gasoline, food, or rents dominate the month’s CPI? (3) Did payrolls or claims change? (4) Did equity prices rally into the survey week? (5) Is there a political event that contaminates “government policy” style items? If you can answer those five, you do not need a take. You already have a diagnosis.
Our U.S. public tape is calibrated to sit between those official series as an educational nowcast, not as a replacement. Licensed files include the mapping, the lag, and the revision flag. That is the unglamorous work that makes a sentiment product usable in a real operating cadence — merchandising meetings, not just Twitter arguments.