The Michigan gloom gauge has stopped predicting the consumer

America has a new favourite doomsday gauge: the University of Michigan Index of Consumer Sentiment, whose preliminary September reading printed at 47.8, down 7.5% in a month and 13% below a year earlier. On any iPhone that price feed is labelled simply "S". It is hard not to startle. The index is deep in the territory that used to accompany recessions, and its record low of 44.8 in May beat the gloom of 2008, the 1980 inflation panic and the covid lockdowns. The natural investor reaction is to ask whether the American consumer — two-thirds of the economy — is cracking, and to trim accordingly.

The trouble is that the gauge has stopped measuring what it purports to predict. Sentiment is a perception of personal finances and the news cycle, heavily priced by petrol. Pump prices are almost 50% above late February after the conflict with Iran and the disruption in the Strait of Hormuz. Tariffs are on the worried mind of a third of respondents. None of that surprises; it is the standard mechanism by which a visible price shock depresses mood.

Two further distortions are doing the heavy lifting, and both are measurement, not economy. The first is partisanship. The Michigan survey has become a mirror of White House control: the gap between Republican and Democratic respondents now runs to roughly 55 index points. It widened from about 21 points under Mr Bush to 45 under Mr Biden. When the party in power changes, sentiment re-orders accordingly — a property no real indicator of demand should have. The second is method. In mid-2024 Michigan moved from telephone to online-only sampling, and researchers reckon that shift alone shaved eight or nine points, more than a tenth, off the index. Hold the roster of respondents steady and the "historic" lows largely vanish.

The reality test is blunt. In the very weeks print after print of record despair, retail sales ran 4.9% above a year earlier, unemployment sat at 4.3%, and the economy had grown at a 2-3% annual clip for three straight years; the S&P 500 has more than doubled while sentiment has hugged territory below past-recession thresholds. The Federal Reserve's own research finds sentiment a poor predictor of spending. Its chairman has called the link between the two "not a strong link at all". A Kansas City Fed paper from February this year found that feeding survey feelings into forecasting models improved nothing over three decades. The Conference Board's rival confidence gauge, which weights the labour market more heavily, has stayed near its long-run average. Mood records and spending records diverge because mood tracks petrol and partisanship, while spending tracks jobs and real incomes.

The reading has not lost all value. Buried inside it is the one number that genuinely moves markets and that the Fed watches: inflation expectations. The year-ahead measure jumped to 4.6% in September, from 4.0% in August, its highest since June and well above the 3.4% of February before the Iran conflict. The more consequential print is the long-run one, which ticked up to 3.4% after three months at 3.3% — above the 2.8-3.2% band of 2024. It is the long run that matters, because a durable rise there would signal that households no longer believe prices will return to target, testing the anchor on which central-bank credibility rests.

Here is the inversion worth carrying. The mood headline is the least reliable part of the release; the expectations subcomponent is the most consequential. A retail investor is well served to ignore the vibe and watch whether long-run expectations keep ratcheting away from target. If they settle back toward the low twos, the noise is just noise — people unhappy about petrol and politics. If they keep drifting up, the story stops being about consumers' feelings and becomes one about the price of money, and about every asset whose value rests on a low discount rate. That is the real signal in the survey, and it is not the one the flashing "S" invites you to fear.