Hard-Hitting 'Soft' Data: Why Surveys Matter

"Soft" data can be bare-knuckled, with confidence and sentiment surveys often driving moves in stocks and bonds.

The question is, how much should it matter?

Survey data like monthly University of Michigan Consumer Sentiment Index and the Conference Board's Consumer Confidence Index^® are often called "soft" because the reports ask consumers how they feel, rather than measuring "hard" numbers like retail sales, home buying, corporate earnings, and initial jobless claims.

The same goes for regularly scheduled business surveys like the Federal Reserve's Beige Book and the ISM manufacturing and service PMIs^® released at the start of each month.

After many decades, the jury is still out on how accurately survey data reflects underlying economic trends or previews hard data to come. Even so, traders and investors ignore survey data at their own risk, because the reports often move markets.

For instance, a surprisingly weak consumer sentiment reading in early 2025 shocked Wall Street and sent stocks lower. Also, one element of the rally to record highs in mid-2026 was renewed strength in the ISM's manufacturing PMI after several years of contraction.

The ISM survey appeared to reflect the economic boom generated by the AI build-out, and hard data like earnings backed up the survey, including stronger sales at companies like Caterpillar (CAT) and memory chip makers like Micron (MU) that provided the muscle and tools to ramp up data center and chip production.

Even as hard data showed economic strength, however, consumer sentiment stayed historically weak, which may reflect a so-called "K-shaped" economy where the very top earners are doing well while the rest of the country struggled. Over time, if sentiment remains weak, it's possible consumer spending might drop, ultimately hurting the earnings picture and perhaps the overall economy. That's the idea behind watching survey data, anyway.

The University of Michigan Consumer Sentiment survey interviews 900 to 1,000 people around the country each month, asking how they're doing financially, whether they think they'll be better or worse off in a year, how they feel about business conditions, and gauging their inflation expectations—among other things. It began in 1946 and recently transitioned from randomly dialing cell phone numbers to interviewing via web surveys with address-based sampling.

The surveys "have proven to be an accurate indicator of the future course of the national economy," the University of Michigan Surveys of Consumers site noted.

That said, no survey is perfect.

"Earlier in the decade, all of this sentiment data was telling you that we were going into recession," said Kevin Gordon, head of macro research and strategy at the Schwab Center for Financial Research (SCFR). "That was the soft data. The hard data was telling you that the economy was largely fine."

Despite the soft data dip at that time, the economy continued growing and a recession didn't occur. "That has kept us in what I call a vibepression, where the consumer vibes have remained depressed despite corporate profits, GDP, and the stock market all continuing to grow," Gordon said. That continues to be the case today, as well, with surveys indicating consumer dread and hard data suggesting resilience.

While soft data can sometimes be vague, hard data can also mislead. For instance, retail sales often fall during particularly cold winter months or when a major hurricane blows in. It's important to sift through data to see if the impact lasts or if weather and other transient events outside the market's control fade as more data filters in. Each data point is simply a snapshot of that month, though exceptions occur, like the weak July 2026 nonfarm payrolls report in which the government also downwardly revised the previous two months of jobs growth.

While retail sales are a closely followed hard data report, they don't adjust for inflation, either. This means the jump in gasoline prices during both 2022 and 2026 played into rising monthly numbers, making bad news arguably look good.

"If in one month you spent $50 to fill up your car at the gas station, and the next month you spent $80 to fill up your car at the gas station solely because the price of gas has gone up, that's not a sign of stronger consumer spending," noted Liz Ann Sonders, chief investment strategist at SCFR, in a podcast. "The nominal dollars were more, but in real terms, it was because you had that move up in inflation. Make sure when you're looking at any kind of economic data, you understand whether it's being reported in nominal terms or real terms."

However, the headline quarterly gross domestic product (GDP) growth rate is reported in real, inflation-adjusted terms, making it a good indicator of actual economic growth on a seasonally adjusted basis.

Consumer sentiment data from the University of Michigan, released twice monthly in preliminary and final forms, came under a microscope recently. One criticism of the survey is that it can reflect partisan political differences among respondents. Survey administrators acknowledge that more positive views tend to come from those whose favored party controls the White House.

During the Biden presidency, for instance, Republicans surveyed consistently sounded more pessimistic about the economy than Democrats. The reverse held true after President Trump took office again in early 2025.

In a 2025 report, however, survey administrators said that "the Surveys of Consumers continue to reach a nationally representative sample of Americans across the political spectrum."

Still, some debate how seriously to take the survey data. First, there's the political element, despite administrators' recent findings. Then there's the relatively small sample size, with around 1,000 people chosen out of hundreds of millions.

Also, any survey could reflect all kinds of things, from what the person ate for breakfast to whether they had an argument with their spouse that day. In other words, moods can affect numbers.

That's why Gordon emphasizes the importance of paying attention to what consumers say, but even more to what they do. For instance, by mid-2026 sentiment dropped to historic lows as crude oil and mortgage rates stayed lofty amid continued U.S. tariffs and the war with Iran. Spending, however, seemed relatively resilient. "Real" personal spending that adjusts for the impact of inflation rose 0.4% month over month in May 2026 and 0.4% in June, the government's data said.

The same goes for the Fed's Beige Book of business conditions across its districts that comes out eight times per year. Fed researchers talk to many businesses for that survey but can't necessarily reach them all. As the Fed itself notes, the information is "anecdotal."

For instance, just before ISM manufacturing began improving in 2026, the Fed's January Beige Book noted "manufacturing activity varied" with five districts reporting growth and six reporting contraction."

But hard data by mid-year indicated actual improvement, with new orders for nondefense capital goods excluding aircraft—a proxy for business spending—rising 0.9% month over month in June following a 1.9% increase in May. These were very robust numbers that anyone reading the January Beige Book might not have expected.

All this doesn't mean soft data can't be a helpful indicator—only that it's not necessarily the final word.