America's gloom index is a poor guide to the market

Few economic statistics announce themselves with such drama as the University of Michigan's Survey of Consumers. The preliminary reading for September 2026, released on September 11th, put its index of consumer sentiment at 47.8, down 7.5% from August and 13% below a year earlier—close to the record low it touched this spring. The sub-index of expectations fell even harder, by 11% in a single month. Any reader glancing at the headline would assume Americans are about to stop spending and that stocks, hovering near their highs, are in trouble.

That is the natural reading, and it may be the wrong one. The Michigan index is a well-built instrument; it has been published, in this format, since 1978 and its predecessor since 1952. But it measures something narrower than the economy's health: how optimistic people feel, which is not the same thing as how resilient they are. The evidence for the distinction has rarely been starker than now, when gloom and growth have co-existed for years.

Start with what actually drives the number down. Economists at the Federal Reserve and elsewhere have long noted that sentiment responds as much to headlines and politics as to pay packets. In the current slump, the University of Michigan itself attributes the September drop to a resurgence in fuel prices and trade tensions; both Democrats and Republicans reported sizeable declines, while independents were little changed. Fuel and tariffs plausibly matter. Yet the partisan skew of the survey is so large that it has become a political instrument in disguise. In 2008 the gap between how Republicans and Democrats felt was under 20 index points; today it is around 55 points. No other division in the population comes close—the gap between the top and bottom income terciles is a mere 13 points.

There is also a quieter methodological story. In the summer of 2024, the survey switched from phone to internet polling to save money. That quietly changed the mix of respondents, reducing the share of Republicans. holding the party mix at its early-2024 level would raise the current index by about seven points—enough all but to erase the "record low" that made so many headlines. The gloom, in other words, is partly an artefact of the ruler used to measure it.

None of this would matter if gloomy consumers went on to spend less. The trouble is they mostly do not. The Kansas City Fed found in February 2026 that adding sentiment to a forecasting model did not materially improve its predictions of real household spending across the past three decades; the Boston Fed reached a similar conclusion back in 2014, calling the predictive role "marginal at best". Jerome Powell, the Fed's chairman, put it bluntly last year: "the link between sentiment data and consumer spending has been weak". Retail sales have kept rising even as consumers insisted they felt terrible, and the same pattern has repeated across market cycles—in 2022 a record-low reading was followed by a bull market.

The reason is that behaviour and feeling are governed by different forces. Sentiment is a lagging, backward-looking measure: households vote with their political identities, react to the price on the fuel pump, and answer questions about an abstract "economy" that they experience mostly through the media they consume. Spending, by contrast, is anchored by jobs, incomes and savings—the things that determine whether a household can actually honour its obligations. As the University's own numbers show, people can hold a dire view of the national economy while maintaining a tolerable view of their own finances. Markets, being forward-looking, have already priced in most of the bad news that sentiment is only now registering as emotion.

The index is not, however, worthless. Two of its components point toward the future rather than reflecting the past. The first is the expectations sub-index, which in September tumbled far more than current conditions—the classic signature of a shock the public thinks is coming. The second, more consequential, is inflation expectations. The year-ahead reading jumped to 4.6% in September, from 4.0% a month earlier, its highest since June and well above the 3.4% recorded before the Iran conflict began in February. Expectations of this kind are the one channel through which a survey of feelings can genuinely alter the economy, because they bear on the Federal Reserve's interest-rate decisions and, through them, on the valuation of every rate-sensitive asset.

The lesson for the investor is one of calibration rather than alarm. A record-low confidence reading is a poor reason to abandon a position, because the number is part politics, part methodology, and only weakly tethered to what people actually buy. The signal worth watching is narrower: whether the gloom begins to show up in realised behaviour—a jump in the saving rate, a slump in retail sales, and the first warnings from retailers and consumer cyclicals about demand. Until then, a thermometer that says "freezing" is not proof that the patient is ill. It is a reminder that a large part of the population is unhappy, which is a political fact with economic consequences that arrive slowly, if they arrive at all.