Deaths of Despair

Anne Case and Angus Deaton unleashed one of the most influential memes in recent social science: the idea that rising death rates among white, non-college Americans reflected a wave of "despair" caused by economic decline — a narrative that seemed to explain everything from the opioid crisis to the rise of Donald Trump. The problem is that the narrative is wrong in almost every particular, and its critics have been too polite about saying so.¹

The Statistical Mirage

The original 2015 paper observed rising death rates among middle-aged white Americans and attributed this to "deaths of despair" — drug overdoses, suicide, and liver disease. But the increase was overwhelmingly driven by drug overdoses. And the authors acknowledged, buried in their own text, three facts that should have been fatal to their thesis: European countries experiencing the same or worse economic conditions saw no similar mortality increase; the trend began before the economic problems they blamed; and Black and Hispanic Americans living through the same conditions didn't show the same pattern.¹

Andrew Gelman then demonstrated that the statistical trend itself was partly an artifact. Because older people have higher death rates, and the average age within Case and Deaton's 45-54 age bin rose during their study period, the increase looked less impressive once you used age-adjusted rates. Gelman was characteristically generous, calling this "in no way a debunking." But generosity may have been part of the problem — the narrative kept trundling forward.

The Switcheroo

Case and Deaton have since published updated versions that quietly abandon their original claims while maintaining the same narrative frame. The 2017 version, post-Trump, dropped the focus on economic conditions to emphasize the college/non-college divide — convenient, since non-college whites were the key swing voters. The 2023 version dropped the race angle entirely and shifted to a pure education divide. The once-central "despair" theme? Now a minor component of a broader mortality divergence driven by cancer, heart disease, and COVID.¹

The deeper problem, as Caroline Hoxby noted in her formal comments on the latest paper, is that the non-college population is a moving target. As more people get degrees, the remaining non-college group becomes smaller and more negatively selected — increasingly composed of people who, for various reasons, didn't or couldn't complete education. The widening mortality gap between college and non-college may largely reflect this compositional change, not a worsening of conditions for the working class.

When Novosad, Rafkin, and Asher actually accounted for these compositional changes, they found a very different story: instead of an educated elite pulling ahead of a suffering majority, the bottom ten percent of the distribution was falling behind while everyone else stayed roughly flat. It's a concentrated problem, not a systemic wave of despair.

What's Actually Happening

If it's not "despair," what explains the specific, real mortality patterns?

The opioid crisis is about drug policy, not economic despair. The US has uniquely permissive pharmaceutical marketing laws, which enabled the Purdue Pharma playbook of promoting supposedly safe time-release opioids. Europe, buffeted by the same economic trends and experiencing the same right-populist backlash, doesn't have the same opioid crisis — because it has different drug regulation. The sequence of supply shocks — first OxyContin, then fentanyl — tracks the death rate better than any economic indicator.¹

The geographic divergence in health outcomes tracks state policy, not economic despair. Places that adopt paternalistic public health measures — higher tobacco taxes, stricter seat belt laws, more aggressive health intervention — get better outcomes. Places that don't, don't. A Washington Post investigation found stark mortality differences across the Lake Erie borders of Ohio, Pennsylvania, and New York — similar populations subjected to different state policy regimes.¹

Even the COVID mortality divergence maps onto vaccine uptake, which maps onto political identity, not economic conditions. Republicans and Democrats have different demographic profiles, and a naive analyst could characterize the differential death rates as "a wave of deaths hitting white, rural, non-college people for mysterious quasi-spiritual reasons." But we all watched the vaccine debates in real time. There's no mystery.

The Cost of a Good Story

The "deaths of despair" narrative is a case study in how a compelling frame can survive its own refutation. Case and Deaton offered a story that felt true — that connected American mortality to American decline, that gave emotional weight to the hollowing out of the working class, that seemed to explain Trump. It was embraced by both Paul Krugman and Ross Douthat, a rare bipartisan consensus. And it sent public health discourse on what Yglesias calls "a years-long wild goose chase away from well-known ideas like 'smoking is unhealthy' or 'it would be good to find a way to get fewer people to use heroin.'"

The real lesson is about the toxoplasma of ideas. A narrative that connects discrete policy challenges — opioid regulation, tobacco taxation, vaccine uptake — into a single grand theory of "despair" is more viral than the boring truth that we face a set of discrete public health problems requiring discrete policy solutions. The discrete solutions aren't glamorous. Raise alcohol taxes. Enforce pharmaceutical marketing regulations. Fund smoking cessation programs. None of these makes for a TED talk about the soul of America. But they're what actually moves the numbers.