The income-wellbeing gap is broader than you think

America is not well. It’s hard to come to any other conclusion, especially when looking at the data we released through the State of the Nation Project. We see worsening self-reported wellbeing on almost every available measure: life satisfaction, youth depression, adult depression, anxiety, emotional support, trust in other people, and trust in institutions.

It’s tempting to think of these measures as unreliable because they are subjective. But that’s precisely why they are important. With humans, perception is reality. As Nicholas Kristof noted in his New York Times column about our project, low perceived wellbeing could become a “self-fulfilling prophecy.” Decisions about education, marriage, childbearing, and other relationships might follow the same downward spiral. Extensive research by one of us on how behaviors and long-term outcomes are influenced by individuals’ wellbeing levels provides empirical support for this proposition.

What is different about our work, and what we think deserves more attention from economists and policymakers, is demonstrating just how broad-based and uniform the decline in wellbeing has become. For years, researchers have puzzled over fairly narrow disconnects: consumer confidence diverging from GDP, or life satisfaction failing to rise alongside average incomes—a modern version of the Easterlin Paradox (named for Richard Easterlin, the first modern economist to study happiness), which provides (much debated) evidence that higher incomes within countries are associated with rising life satisfaction. These patterns are real but limited in their focus. What our data now show is something more troubling.

Our most recent report, the State of the States, finds that of 225 possible opportunities for states to show improvement on six self-reported wellbeing measures, only 12 show genuine progress—a 0.050 batting average. No state has improved on life satisfaction and almost all the others listed above are worsening, in every state, regardless of income level or political leadership. And this builds on our national-level findings showing simultaneous declines in those measures, even as average income continues to rise.

This is a systematic, nationwide divergence between what the economic numbers say and what Americans report about their own lives. It demands a serious explanation. Three broad types of explanations deserve attention—and all are probably at least partially right.

Explanation one: The economic measures are off

One possibility is that the economic statistics themselves are not accurately capturing what is happening to the material lives of typical Americans.

The first key point is that we are comparing changes in average income and average wellbeing. This is important given another finding in the State of the Nation Project: that income inequality is rising. This distinction is important because most economics textbooks, as well as a vast literature on life satisfaction, show that each additional dollar matters less as income levels increase, i.e., diminishing marginal utility of income. Further, with income gains going mainly to the top earners, we should not really expect personal wellbeing to rise.

One reason we suspect this is not the main explanation is simply that real income per capita isn’t shrinking. It’s just increasing more slowly for lower-income groups. So, flaws in the economic statistics cannot explain why the income-wellbeing relationship has reversed.

Beyond distribution, there are serious questions about whether our inflation measures accurately capture the actual cost of living for typical households. CPI does a poor job of capturing certain costs that have risen sharply for middle-income households: housing prices, homeowners insurance, college costs and debt, and health costs. While the cost of college has risen less than the popular imagination, the price of college makes up only a tiny share (1.3%) of the CPI market basket since so few people are in college at a point in time; for those who do go to college, however, college is almost the entire market basket and loan payments are a large share of income even for those who find a job.

Similarly, childcare is less than 1% of the CPI market basket because, again, most people don’t have children. But try telling that to a working parent, who generally spends at least 10% of income on this essential service. CPI can only do so much to account for the diverse circumstances of the entire population, some of whom are experiencing sharply rising costs.

Standard income measures also entirely miss the dimension of economic uncertainty—the financial vulnerability that comes from having little buffer against a large medical bill (something most other countries address through universal health insurance). For households at the median and below, that vulnerability is real and consequential for wellbeing even in periods when measured income is rising. Also, a rising share of people live alone, so that housing is immediately at risk when jobs are lost. Married couples, meanwhile, might require two incomes to maintain their expected lifestyle, which means losing one income could force them out of their homes.

Taken individually, none of these measurement problems seems large enough to account for the full income-wellbeing gap. But taken together—with rising inequality concentrating gains at the top, with real purchasing power eroded by costs that the CPI understates and with economic insecurity excluded from the ledger entirely—they likely explain a meaningful portion of what we observe. But probably not all of it.

Explanation two: The wellbeing measures are off

A second possibility is that the self-reported wellbeing data are unreliable—that people are echoing negative messages rather than genuinely reporting their own experience.

The evidence cuts against that, too. If rising depression were merely a survey artifact, we would not expect to see parallel increases in suicide rates over the same period, which are measured through administrative records rather than self-report. We also would not expect the pattern to be so geographically systematic: self-reported wellbeing is declining broadly across Western countries while improving across much of Asia and Eastern Europe. Survey bias does not produce that kind of structured cross-national pattern.

The more plausible version of this concern is not bias in the classical sense but something subtler: that people’s genuine self-assessments are being shaped by the media environments they inhabit. We turn to that possibility next.

Explanation three: The gap is real, but driven by something else

Our instinct is that the income-wellbeing relationship has genuinely weakened, but that the main drivers are not economic in the conventional sense. Two candidates stand out.

The first is digital media. The smartphone and social media have restructured how Americans spend their time, what they see and pay attention to, and how they compare themselves to others. The flood of disturbing content available on demand—violence, crime, political conflict, pornography, social grievance—creates a sense of ambient threat that shapes self-assessment even when individual circumstances are unchanged. Crime is a good example. Our data show that the murder rate is down 40% since 1990, and have been steady since the early 2000s, aside from a small spike during COVID. Nevertheless, people report feeling more threatened, a direct reflection of the gap between the social media environment and material reality.

Simultaneously, the “highlight reel” problem means that peer group comparisons, which the satisfaction literature identifies as a key driver of subjective wellbeing, are increasingly made against curated best-selves rather than actual lives. This is “keeping up with the Joneses” operating at a scale and intensity that has no historical precedent. There is also now robust evidence that device use has contributed to rising loneliness by crowding out face-to-face interaction, and that these effects are particularly acute among younger people—consistent with the especially sharp declines we observe in youth mental health.

Another driver, and the one we find most compelling, is political polarization. The United States is experiencing some of the fastest growth in polarization, and the highest polarization level in our analysis of the most recent data from 92 countries. Support for violence to achieve political aims is also on the rise. When partisans perceive political opponents as enemies rather than fellow citizens, that animosity bleeds into assessments of everything else, including economic conditions and life satisfaction. Pew Research has documented sharp discontinuities in the public’s assessments of the country and the economy at the moments of presidential transitions, with the optimism of one party’s supporters collapsing and the other’s surging almost overnight. In a country this divided, something close to half the population is, at any given moment, primed to report that things are going badly—regardless of what the underlying data show. And critically, this is not merely a survey artifact. Polarization and the media environment that feeds it appear to be generating genuine psychological distress, not just negative survey responses. A useful reference point is the recent finding that iPhone use has contributed meaningfully to declining fertility rates. If these technologies can move something as fundamental as the decision to have children, they can certainly move self-reported wellbeing.

What follows

Each of these explanations carries different implications. If the problem were primarily bad survey data, there would be little to act on. If it were primarily bad economic measurement or genuine material deterioration—rising inequality, understated inflation, or increased uncertainty—then the response lies in economic (and health) policy.

But if some of the most powerful drivers are digital media and political polarization, the appropriate responses look quite different. They might include rethinking how media platforms are regulated, finding ways to strengthen institutions that rebuild cross-partisan trust, and addressing the structural features of American democracy—electoral rules, primary systems, campaign finance—that have made polarization so durable and so extreme. We are not in the business of prescribing specific political reforms, and reasonable people disagree sharply about which interventions would help. But the data make clear that part of the story—especially for the young—is these “economic” issues, involving domains that go well beyond the usual economic boundaries.

What the State of the States 2026 report ultimately shows is that pointing to strong GDP or low unemployment as evidence that things are going well has lost its persuasive force. The American public has been delivering a consistent, cross-partisan verdict about their own lives, and it diverges sharply from what the standard scorecard records. Taking that verdict seriously, and explaining it rigorously, is where the work needs to go.