Market Research, Motorcycle Market research

The Sentiment Gap: Americans Say They Are Worried. They Are Still Buying Motorcycles.

by Francis Rozange | Sep 28, 2026 | 0 comments

By the measure that economists, bankers and cable-news anchors have watched for three generations, the American consumer is in a sour mood. The University of Michigan’s index of consumer sentiment fell to 51.7 in August 2026, a drop of 6.3 percent from July and 11.2 percent from a year earlier, and the final September reading slipped again, to 48.1. In May the index touched 44.8, which the survey’s own researchers describe as the lowest reading in its history, below even the trough of June 2022, when inflation was at its peak. The people who answer the survey say they are worried about prices, about business conditions and about their own finances.

And yet they are buying motorcycles. According to retail data compiled by the Motorcycle Industry Council and reported in the trade press on September 23, new motorcycle sales among leading brands rose 1.9 percent in the first half of 2026 compared with the same period of 2025. Shipments of replacement motorcycle tires, a quieter indicator of how much Americans are actually riding, rose 3.8 percent over the same six months. The council itself was careful not to overstate either signal. Its director of research and statistics said that consumers were signaling more caution about the months ahead, but that the real question for the industry was whether those concerns would begin to show up in actual purchasing behavior.

A Survey Built for Economists, Read by Dealers

That question, whether a stated feeling becomes a transaction, is older than the powersports industry and more stubborn than most dealers assume. The Michigan survey began as an annual study in the late 1940s, became quarterly in 1952 and monthly in 1978, and it remains, with one rival, the most closely tracked measure of American consumer confidence. It asks a national sample of adults how they judge their personal finances, business conditions over the next year and the next five years, and buying conditions for major household items. The composite index blends those answers into a single number that moves markets and headlines.

The September commentary from the survey’s director shows how much is packed into that single number. Views of current and expected personal finances each weakened by about 10 percent in the month. Year-ahead inflation expectations jumped from 4.0 percent to 4.6 percent, the highest since June, and the short-run outlook for business conditions fell sharply amid worries about elevated fuel prices and renewed trade disputes. The survey also noted that sentiment among self-identified Republicans was now 20 percent below its January level, and among Democrats 13 percent below. None of these answers is a question about motorcycles, and none of them is a question about what the respondent will actually do.

Two Thermometers, Two Different Readings

The first lesson for anyone who reads sentiment for a living is that there is no single thermometer. The Conference Board, a nonprofit business research group, publishes a rival consumer confidence index built on different questions, with a heavier emphasis on business and labor market conditions. Its August 2026 reading fell only modestly, by 0.8 points to 89.4. More striking, its Present Situation Index rose by 6.8 points to 121.2, as more respondents described jobs as plentiful, while its Expectations Index fell by 5.8 points to 68.2. In the same month, then, one major survey showed Americans feeling worse about current conditions and the other showed them feeling better.

This is not a statistical embarrassment; it is a clue. A 1998 study by two economists at the Federal Reserve Bank of New York ran what they called a horse race between the two indexes and found that the Conference Board measure generally had more power to forecast household spending. For motor vehicle spending specifically, lagged values of the Michigan index added about 5 percent to the variance explained by a baseline model, while using both indexes together raised the gain to 21 percent. The practical implication is simple. Different questions capture different parts of the household mind, and a business that relies on a single headline index is reading one instrument on a crowded dashboard.

Forty Years of Asking Whether Mood Predicts Spending

Academic economists have argued about the predictive value of sentiment for decades, and the argument has not been settled so much as refined. In a widely cited 1994 paper in the American Economic Review, Christopher Carroll, Jeffrey Fuhrer and David Wilcox found that lagged sentiment helped forecast household spending and suggested that it might act as an independent force on consumption. Nine years later, a study published by the Federal Reserve Bank of Richmond, covering data from 1959 to 2001, concluded that once current income, real interest rates and lagged wealth were properly controlled for, sentiment had no direct role in predicting spending, though it did help anticipate changes in income and rates.

Sydney Ludvigson’s 2004 survey of the literature in the Journal of Economic Perspectives offered what remains the most balanced verdict. Popular confidence measures, she wrote, contain some information about future spending growth, but much of that information is already present in other economic and financial indicators, and the independent contribution of confidence is relatively modest. In May 2025 two economists at the Federal Reserve Bank of Kansas City reached a similar conclusion after comparing forecasts over the past thirty years: adding sentiment to a model built on official spending and income figures did not substantially improve forecasts of consumer spending growth. Sentiment, in short, is informative mainly because it arrives earlier than the hard data, not because it knows something the hard data never will.

When Pessimism Does Not Reach the Wallet

A more recent strand of research explains part of the gap directly. In a study published in 2023 in the Review of Economics and Statistics, Atif Mian, Amir Sufi and Nasim Khoshkhou showed that Americans’ economic expectations swing sharply according to whether their preferred party holds the White House, and that this partisan effect has grown over twenty years. After the 2008 and 2016 elections, respondents reported large shifts in their outlook along party lines. Administrative spending data, however, showed no corresponding change in what those households actually spent. The mood moved; the wallet did not. For a market researcher the design matters as much as the finding: the authors set what people said against records of what they actually did, rather than against a second survey.

The Michigan survey’s own September note, with its sharp divergence between Republicans and Democrats, is a reminder that part of any monthly swing may be of this kind. The evidence is not one-sided. Christian Gillitzer and Nalini Prasad, working with Australian survey data and publishing in 2018, found that post-election shifts in sentiment did move stated spending intentions, and that geographic variation in automobile purchases suggested those intentions were reflected in actual buying. The fair reading of both studies is analytical rather than ideological: some sentiment shocks carry real economic content and some are expressive, and a business needs a method for telling them apart before it cuts orders.

The Paradox of Anxious Buying

One reason worried consumers keep buying durable goods is that worry itself can accelerate a purchase. The Michigan survey’s September release contained a sentence that deserves more attention in the powersports world than the headline index. Buying conditions for durables, the survey reported, improved slightly, in part because consumers perceived that completing such purchases now would help them avoid higher prices later. A household that expects prices to rise can be deeply pessimistic about the economy and still decide that this autumn, rather than next spring, is the moment to sign for a motorcycle. The same release put year-ahead inflation expectations at 4.6 percent, well above the 3.4 percent recorded in February, before the conflict involving Iran began.

Trade policy has made such timing decisions concrete. In late August the Government of Canada announced a 50 percent counter-tariff on motorcycles of American origin with engines above 800 cubic centimeters, effective September 8, 2026, and the Canadian industry association warned of consequences for a national network of roughly 900 dealerships. Our analysis is that dated price shocks of this sort tend to concentrate demand ahead of the deadline and leave a gap afterward, which means that a strong quarter can be a borrowed quarter. Pessimism about prices and optimism about a purchase are not contradictions. They are often the same calculation made by the same buyer.

What Cutting Back Actually Looks Like

A special report published by the Michigan survey in August 2025 offers an unusually clean test of stated intention. Asked between May and July of that year how they would respond to large price increases, only 24 percent of consumers said they would spend as usual on the affected items, compared with 36 percent in late 2022. Fifty-eight percent said they would cut back and 13 percent said they would stop buying altogether. The responses varied sharply by income: 31 percent of the top income third expected to spend as usual, against 18 percent of the bottom third. The same report noted that about 60 percent of consumers expected unemployment to worsen in the year ahead, a reading last seen during the Great Recession.

What happened in the motorcycle market that year suggests that “cutting back” is not the same as abstaining. The council’s full-year data for 2025, as reported in February, showed motorcycle and scooter sales down 7.6 percent and all-terrain vehicle sales down 3.5 percent. But the composition shifted more than the total. Sport bikes rose 13 percent, with all of the gain coming from models under 750 cubic centimeters; adventure models under 600 cubic centimeters rose 10 percent; touring motorcycles fell 13 percent and cruisers 6 percent. The council’s research director summarized the shift as a move toward smaller, more affordable and more versatile models. The pattern is consistent with what consumers said, although sales data alone cannot prove the link: restraint appears to have taken the form of trading down.

A Rebound, Then a Slower Pace

The first quarter of 2026 brought the reversal that the gloomy surveys did not foreshadow. According to council figures relayed in the trade press in May, new motorcycle and scooter sales rose 4.2 percent from the first quarter of 2025, and touring motorcycles, a category that had fallen 13 percent the year before, posted the largest first-quarter unit increase of any motorcycle category since 2022. Replacement tire shipments rose 5.3 percent in the quarter. All of this happened while the Michigan index was sliding from 56.6 in February toward its May record low. The council’s president said at the time that, despite profitability pressures across the industry, the increase in retail activity showed continued enthusiasm for riding heading into the peak season.

The first-half figures published in September are more subdued: a 1.9 percent gain in new motorcycle sales among leading brands and a 3.8 percent rise in tire shipments. If the first-quarter and first-half series cover comparable scopes, and the council’s wording suggests they may not be identical, the arithmetic implies that the second quarter grew more slowly than the first, or not at all. That is an inference, not a published number, and the council has not released a stand-alone second-quarter figure in the reports we reviewed. It is nonetheless the kind of deceleration a careful analyst would flag, because it is the first place where cautious attitudes could begin to leave a mark on actual purchases.

The Supply Side of the Gap

Retail sales are not a pure measure of consumer will; they are also a measure of what manufacturers and dealers choose to put in front of buyers and at what price. The largest American maker of heavyweight motorcycles told investors in its second-quarter filing with the Securities and Exchange Commission that its North American retail sales rose 3 percent from a year earlier, that its global dealer inventory of new motorcycles was 17 percent lower than a year before, and that its wholesale shipments had been kept below retail sales as a deliberate priority. The company also cited net pricing as a drag on revenue, and its gross profit reconciliation listed price and sales incentives among the negative factors.

These are company statements, not independent data, but they illustrate a mechanism that sentiment surveys cannot see. Leaner showroom inventory reduces the pressure on dealers to discount heavily, while targeted pricing support can keep a hesitant buyer in the market. Our reading is that at least part of the resilience in 2026 retail figures reflects supply discipline and pricing decisions after a difficult 2025, rather than a spontaneous burst of consumer confidence. For a dealer interpreting the gap between stated mood and actual sales, the distinction matters: resilience that is bought with incentives behaves differently in the next quarter from resilience that comes from demand.

The Price of Money

For many motorcycle buyers, the monthly payment is a more immediate constraint than any view of the national economy. On September 16, the Federal Reserve raised its target range for the federal funds rate by a quarter of a percentage point, to 3.75 to 4 percent, noting in its statement that domestic spending had been resilient but that inflation remained elevated. The decision came as the University of Michigan was recording a jump in year-ahead inflation expectations, and it pushes financing costs in the direction that discretionary purchases like least. Many consumers had expected as much: in the Conference Board’s August survey, 61.3 percent anticipated higher interest rates over the following twelve months.

The Federal Reserve does not publish an interest rate for powersports loans, so the closest official proxies are its figures for other consumer credit. Its September release put the average commercial bank rate on a 60-month new car loan at 7.14 percent in the second quarter of 2026, down from 8.16 percent for 2024 as a whole but far above the 4.82 percent of 2021. The average rate on a 24-month personal loan stood at 11.86 percent, compared with 9.38 percent in 2021. We found no published official series for motorcycle loan rates, and we do not substitute an estimate. The level, however, remains far above that of 2021, the September increase works against further relief, and the cost reaches the buyer at the finance desk rather than in the survey interview.

The Buyer Is Not the Median Respondent

A national sentiment index gives equal voice to every adult in its sample. The motorcycle purchase is concentrated in a much narrower group. The most recent owner demographics we found in a public council release, from its 2018 owner survey, put the median age of American motorcycle owners at 50, up from 45 in 2012, with 19 percent of owners female, 24 percent retired and a median household income of 62,500 dollars. Those figures are now several years old; the council’s current statistical annual, which includes rider demographics, is available only to members and paying buyers, and we did not review it.

Even the dated profile is enough to show why the headline index can mislead a powersports business. The Conference Board reported in August that confidence remained highest among consumers under 35 and generally higher among higher-income groups, with older generations trailing by a wider margin. Michigan’s 2025 report found that the top income third was far more likely to keep spending as usual than the bottom third. A market whose buyers skew older, whose purchase is often financed, and whose sales are shifting toward cheaper entry models is exposed to several different slices of the national mood at once. Averaging them into one number hides exactly the segmentation that matters.

A Second Reading From Canada

Canada offers a useful parallel, because its national industry association has just released its annual report for 2025. According to figures from the national industry association reported in the trade press on September 23, Canadians bought 61,623 new motorcycles and scooters in 2025, along with 48,994 all-terrain vehicles and 37,918 side-by-side vehicles. The association’s president described 2025 as a period of normalization, with new motorcycle retail sales down modestly from recent record levels and the market still significantly stronger than before the pandemic. Taken together, those three categories come to about 148,500 new vehicles, sold through a dealer network that the association elsewhere puts at roughly 900 outlets, many of them small and medium-sized businesses.

The more telling Canadian figure is not a sale at all. More than 915,000 on-road motorcycles were registered in Canada in 2025, according to the same report, the highest level in five years. The association’s president drew the distinction himself, observing that Canadians were continuing to ride and not only to buy. The American tire data make the same point from another direction. A rider who postpones replacing a motorcycle but keeps riding the old one is still buying tires, service, gear and insurance. For much of the industry, usage is the revenue base, and usage appears less sensitive to stated gloom than the decision to buy a new machine.

Why Buying Intentions Mislead

If headline sentiment is a blunt instrument, the obvious alternative is to ask people directly whether they intend to buy. That approach has its own long history of disappointment, and the reason was identified sixty years ago. In a 1966 study for the National Bureau of Economic Research, the economist F. Thomas Juster showed that surveys of buying intentions were inefficient predictors of purchase rates because nonintenders, the people who say they do not plan to buy, account for the bulk of actual purchases and for most of the variation in purchase rates over time. Juster had argued in earlier work that statements about buying intentions are essentially probability statements in disguise.

Juster’s remedy was to replace the yes or no intention question with an explicit probability scale. In an experiment run with the Census Bureau, the probability measure explained about twice as much of the cross-section variation in automobile purchase rates as the intentions question, largely because it separated nonintenders and undecided respondents into groups with systematically different buying rates. Once probabilities were in the model, the traditional intention answers added nothing. The lesson for motorcycle research is direct. The prospect who says “not this year” is not a zero, and a survey design that treats that answer as one will miss the market’s swing buyers.

Calibrating What People Say

Later work formalized the limits of stated intention. Charles Manski, in a paper published in 1990 in the Journal of the American Statistical Association, showed that even under the best possible assumption, that respondents answer as rational forecasters, intentions data bound but do not identify the probability that a person will act. Stated preference research in other fields points the same way. A 2005 meta-analysis of 28 valuation studies found that hypothetical willingness to pay exceeded actual payment by a median ratio of 1.35, a smaller gap than folklore suggests but a persistent one, and one that choice-based questions helped to reduce.

The conditions under which intentions work best are well documented. A 2007 study by Vicki Morwitz, Joel Steckel and Alok Gupta in the International Journal of Forecasting found that intentions correlate more closely with purchases for existing products than for new ones, for durables than for nondurables, over short horizons rather than long ones, and when respondents are asked about specific models rather than a whole category. A motorcycle is a durable good with a considered purchase cycle, which is favorable. The hazards are the long horizon of most annual surveys and the habit of asking about the category rather than about a specific machine at a specific price.

The Survey That Changes Its Respondents

There is a further complication that dealers who survey their own prospects rarely consider: asking the question can change the answer and the behavior. In a 1993 study in the Journal of Consumer Research, Morwitz, Eric Johnson and David Schmittlein found that merely asking people about their intention to buy raised their subsequent purchase rate, while repeated questioning of low-intent respondents lowered it. Both effects were weaker among consumers with prior experience of the product. In practical terms, a single well-timed question can nudge an undecided prospect toward a purchase, while a prospect questioned repeatedly without real intent may harden against it. Neither effect appears in the survey tabulation; both appear, if at all, in the sales ledger.

A 2005 study in the Journal of Marketing by Pierre Chandon, Morwitz and Werner Reinartz measured how much this inflates apparent accuracy. Across three large field studies, the correlation between underlying intentions and later purchase behavior was on average 58 percent greater among surveyed consumers than among comparable consumers who were not surveyed. For a powersports business, the consequence is uncomfortable. A follow-up survey of showroom visitors, or a post-event questionnaire at a demo ride, may appear to predict sales well precisely because the survey itself nudged the respondents. Without an unsurveyed comparison group, the validity of the instrument is partly self-generated.

Linking What People Say to What They Do

The remedy is not to stop asking but to design the asking around what people later do. That means probability scales rather than binary intentions, questions tied to specific models, price points and financing terms, short and dated horizons, and above all a link between each respondent’s stated answer and a behavioral record: a registration, a financing application, a service visit, a tire purchase. Panels that are re-contacted over time allow stated probabilities to be calibrated against realized purchases for the same people, and holdout samples that are never surveyed allow the measurement effect to be estimated rather than ignored.

This is the discipline that separates customer research from opinion polling, and it is central to the way CSM International approaches motorcycle research. The goal is not to produce a more optimistic or more pessimistic number than the national indexes, but to produce a number whose relationship with actual behavior is known and has been tested. In a year when the national mood and the showroom have moved in opposite directions, the practical value of a calibrated instrument is that it tells a manufacturer or a dealer group which part of the gloom belongs to its own buyers and which part belongs to everyone else.

What a Careful Dealer Should Watch This Autumn

The calendar for the coming weeks is dense. The Conference Board’s September confidence reading is scheduled for release on September 29, and the University of Michigan will publish preliminary October data on October 9. Neither will say much about motorcycles directly. More useful for the industry will be the council’s third-quarter retail figures, the path of financing costs after the Federal Reserve’s September increase, and whether the buy-ahead motive that the Michigan survey detected in durable goods fades once tariff deadlines pass. The Conference Board’s own August release contained a more relevant detail than its headline: on a six-month moving average basis, plans to buy an automobile remained strong.

The deeper point is that the gap between what Americans say and what they do is not a malfunction to be explained away each month. It is a permanent feature of measuring intention, documented from Juster’s experiments in the 1960s to the partisan swings of the last two decades. Sentiment surveys remain valuable as early and broad signals, and the council is right to watch them alongside sales rather than in isolation. But a business that sets production, inventory or staffing by the headline index is reading a national mood as if it were a local order book. The people who fill the order book are older than the national sample and more specific about what they are weighing, and they are best measured directly.