Deaths of Despair
and the Future of Capitalism
by anne case and angus deaton
*Princeton University Press 2020.
All rights reserved.
Deaths of Despair and the Future of Capitalism,* the superb new book by Anne Case and her spouse, economic Nobelist Angus Deaton, is at once a survey of the dimensions of the social disaster overwhelming America’s white working class and a mystery story. For while the fatal epidemic of drug abuse, alcoholism and suicide – dubbed “deaths of despair” by Case and Deaton – is widely acknowledged, the causes are much disputed. Some point to the decline in community and family, some to the availability of spectacularly lethal opioids, some to disillusionment with elites, some to the winner-takeall nature of contemporary capitalism. Case and Deaton, who both teach economics at Princeton, have other ideas. Here, we excerpt the chapter that picks apart the dominant view that, one way or another, economics lies at the heart of the problem. Start here for a taste of their meticulous logic. Then read the rest of this accessible, jargon-free tome for their surprising and deeply troubling conclusions.
— Peter Passell
Published July 30, 2020
Deaths of despair are concentrated among those with less education, and the epidemic is widening the gap in years lived between those with and without a bachelor’s degree. But we have said little about money, or about its absence, and just how income or poverty fits into the story. Even for those who are not poor, people with higher incomes live longer – and there is evidence that education matters, too, even among people with the same incomes. In America, money buys access to better health care and, beyond that, life is easier when you do not have to worry about how you are going to pay for a car repair, childcare or an unexpectedly large heating bill after an especially cold winter month.
Financial worry can suck the joy out of life and bring on stress, often a trigger for pain and ill health. It would be surprising if money did not have its own beneficial effect on health even if much of the link between wealth and health is explained in other ways – through the impact of poor health on earnings, through education’s effects on both health and wealth, or through childhood circumstance setting the stage for adult health and wealth.
The United States has a much less comprehensive safety net than rich countries in Europe and elsewhere. The absence of benefits gives people sharp incentives to work and earn, which is good for those who can, but can be disastrous for those who, for one reason or another, cannot. The United States is also different from other rich countries in having several million extremely poor people, who arguably live in conditions as bad as poor people in Africa and Asia. Poverty is an obvious place to look when trying to explain an epidemic of death that is unique to the U.S.
Income inequality often features in popular discussion of deaths of despair and of American ill-health more generally. Inequalities in income and wealth are higher in the U.S. than in other rich countries, so that inequality is a popular candidate to explain other outcomes where the U.S. is exceptional. Poverty and inequality are seen as twin curses that are routinely, if usually not very precisely, blamed for all manner of evils – not just for contributing to poor and declining health, but also for undermining democratic governance, for slowing economic growth, for inducing economic instability, for eroding trust and happiness, and even for spurring the rise in obesity.
Poverty may be harder to bear in more unequal societies: poor people not only have to suffer their own poverty but can also see that there are others who have vastly more than they need. We have much to say about inequality in this book. We will argue that deaths of despair and income inequality are indeed closely linked, but not, as is often argued, with a simple causal arrow running from inequality to death. Instead, it is the deeper forces of power, politics and social change that are causing both the epidemic and the extreme inequality. Inequality and death are joint consequences of the forces that are destroying the white working class.
We resist the notion that income inequality is like pollution in the air or deadly radiation, so that living in a more unequal society is something that sickens everyone, rich and poor alike. For one thing, the huge expansion of income inequality in the U.S. came after 1970, precisely during the period when mortality was falling rapidly and life expectancy rising rapidly. Beyond that, although some states in America are much less equal than others, the epidemic of deaths of despair is no worse in less equal states. New Hampshire and Utah, two states with the lowest levels of income inequality, have been much harder hit than New York and California, two states with the highest.
The Great Recession began in 2008 with the collapse of Lehman Brothers and quickly led to large-scale unemployment and distress, not only in the United States but in other rich countries. The U.S. unemployment rate, which had been less than 5 percent in February 2008, was nearly 10 percent by the end of 2009, and it did not regain the 5 percent level until September 2016. The recovery is still incomplete in some aspects, especially for the less educated. Throughout the period from January 2010 to January 2019, the number of college graduates aged 25 and over in employment increased in total by 13 million (about a quarter). Employment for those without a degree rose by 2.7 million – but by only 55,000 for those with a high school degree or less!
Job growth for those with a college degree was barely affected by the Great Recession.
The recovery – although it has seen some growth in wages for the least-skilled – has not provided them with jobs. From 2008 to 2016, when deaths of despair were rising rapidly and progress against heart disease mortality was reversing, income and employment were much lower for less-educated Americans than they would (or should) have been had the bubble not burst.
The policy response to the financial crisis in the U.S. was less than it should have been, but was relatively successful compared with Europe. Different countries in Europe experienced the recession in different ways. Some remained untouched while others, either because of their own choices or because they had no choice given their debt position and their membership in the eurozone, experienced more or less severe austerity with cuts in state spending and benefits. The crosscountry variation in Europe gives us a laboratory in which we can compare health outcomes with different degrees of economic distress.
We do not believe that poverty or the Great Recession is central to our story of a rapid increase in deaths of despair. We do not deny the depth of poverty, nor the misery and ill-health that comes with it, and we acknowledge and deplore the disgraceful living conditions and low life expectancy in parts of America. That these are worse than in western Europe is a direct testament to the inadequacies of the American safety net and its health care system. But it is not possible to explain deaths of despair in terms of America’s exceptional poverty or the Great Recession.
The year 2017 marked the third consecutive year of decline in the poverty rate. This looks nothing like the pattern of deaths of despair, which rose uninterruptedly and increasingly rapidly from the early 1990s.
We pick up the story of what did happen in the next chapter. But the detour is important because when people are asked what might be causing deaths of despair, the common responses are poverty, inequality, the financial crisis or all three. All are important, but none is the main cause of deaths of despair. Yet the contrary view is so widespread that we need to explain why it is wrong while, at the same time, fitting poverty, inequality and the crisis into our story.
We know a lot about the people who have died from the information on their death certificates, including their educational attainment. But there is a great deal more that we do not know, but would like to, including their occupations, income, wealth and whether or not they were poor. Without that, we cannot immediately see whether deaths of despair are tied to poverty. We have to work indirectly.
There was no increase in the national poverty rate to match the timing of the epidemic. Official poverty counts, which tally the number of people living in households with incomes below the poverty line, were falling throughout the 1990s, when the epidemic was getting under way, down to a low of 11 percent of the population in 2000. There was then a slow rise to 13 percent on the eve of the Great Recession, followed by a rapid rise through the recession and slow decline thereafter. The year 2017 marked the third consecutive year of decline in the poverty rate. This looks nothing like the pattern of deaths of despair, which rose uninterruptedly and increasingly rapidly from the early 1990s.
The official poverty counts have a number of serious flaws. In particular, they take no account of taxes or benefits, such as the Earned Income Tax Credit, or food stamps.
While it is important to adjust for these, especially when assessing how the welfare system helped people during the Great Recession, none of the adjustments would make the poverty counts better match the steady rise in deaths of despair. There is simply no increase over time in poverty that can explain the surging epidemic.
The racial patterns of the epidemic are also difficult to reconcile with a poverty story. Among adults without a bachelor’s degree, the fraction of white non-Hispanics living in poverty was less than half that among blacks over the period from 1990 to 2017. Yet, at least up to 2013, African-Americans were almost exempt from the epidemic. From the early 1980s through the start of the Great Recession, midlife white poverty rates were roughly constant at 7 percent (9 percent for those with less than a bachelor’s degree), while deaths of despair among whites increased year by year.
More generally, a wide range of measures of living standards are worse for blacks than for whites, but it was almost exclusively whites who were dying from deaths of despair from the 1990s through to 2013. Whatever was differentially affecting white non-Hispanics, it was not that they were poorer than other groups.
Long-established deep poverty in the United States exists, especially among African- Americans. Indeed, it is the long and disgraceful history of race in America that has done much to prevent poverty relief in the South – where governments have long been white and the actual or potential recipients of relief are black. Deep financial poverty maintained over many years brings health poverty, too, worsened by racism and the low levels of health care, education and sometimes even sanitation.
Yet poverty is not the source of the surge in deaths of despair. The timing is wrong, and the deaths are too white. The geography is wrong too. The figure below shows ageadjusted mortality rates for accidental (or intent undetermined) drug overdoses for whites ages 25 to 64, by state, mapped against state poverty rates for whites, in 2017.
Certainly, Appalachia – especially West Virginia and Kentucky – are centers of drug overdoses and have high rates of poverty. But poverty does a poor job of matching deaths across the country. Drug overdoses are also prevalent in less economically deprived states along the Eastern seaboard, from Florida up the coast to Maryland, Delaware, New Jersey, Connecticut, Rhode Island, through Massachusetts to New Hampshire and Maine. There are also states with high levels of poverty, like Arkansas and Mississippi, that are much less affected by overdose deaths.
At the same time, suicide is much more prevalent in the Rocky Mountain states, where poverty is not particularly high. Suiciderate increases from 1999 to 2017 were also larger in the Mountain states, where they were already taking the highest toll. There is no part of the country that has been left untouched by suicide – two-thirds of all states saw midlife suicide rates for whites increase by at least 50 percent between 2000 and 2017. There is a positive correlation between alcohol- related liver mortality and state poverty. But the states with the highest poverty rates (West Virginia, Kentucky, Arkansas) do not have the highest death rates from alcoholism, in part because a large fraction of their residents are non-drinkers. Alcohol-related death rates are highest in Nevada, New Mexico and Florida, and are rising most rapidly in Western states (Wyoming, New Mexico, Oregon, Washington) and in the South.
Whatever is the nature of the despair driving the epidemic, it is widespread, and is not captured by poverty at the state level.
Deaths of despair are prevalent among those who have been left behind, whose lives have not worked out as they expected. Income is a part of this story, though we shall argue in a later chapter that declines in income work alongside negative social and political factors. But the fact that working-class people have been left behind economically is important, especially when there has been real economic growth in the economy. That growth has accrued to a more educated elite, leaving others behind. Widening income gaps are a consequence of this process, as are deaths of despair.
There is another line of thought that blames inequality itself for social disruptions, including mortality. In this account, inequality hurts people by undermining the social solidarity and relationships that are essential to a good life. The British epidemiologist Richard Wilkinson has argued that healthy societies have relationships that “are structured by lowstress affiliative strategies which foster social solidarity,” while unhealthy societies are characterized by “much more stressful strategies of dominance, conflict and submission.” Which kind of society we live in is “mainly determined by how equal and unequal a society is.”
This is different from arguing that poverty is the root of ill-health. In that story, the poor are unhealthy because of their poverty. By contrast, if inequality makes a society unhealthy, everyone’s health suffers, rich as well as poor.
Wilkinson’s theory has a lot to recommend it, especially its focus on social as opposed to individual circumstance. But here we are interested in whether it can help account for mortality in America today, and whether income inequality is linked to the epidemic of deaths of despair.
We agree that the increase in American inequality since 1970 is indeed linked to increases in deaths of despair. Not directly – as in “inequality makes us all sick” – but because in the U.S. the rich got richer at the expense of the rest of America – “reverse Robin Hood” in action. For 20 to 30 years, while income inequality rose, mortality declined. But eventually (after 1990), we began to see an increase in deaths of despair among the less-educated. We will argue that this is due not to the increasing prosperity of those in the top 1 percent, but to what has been happening to the white working class itself. Of course, the increasing prosperity at the top may have a good deal to do with the distress at the bottom, and will be central when we come to think about what should be done. But those in despair are in despair because of what is happening to their own lives and to the communities in which they live, not because the top 1 percent got richer.
The increase in American inequality since 1970 is indeed linked to increases in deaths of despair ... because in the U.S. the rich got richer at the expense of the rest of America – “reverse Robin Hood” in action.
Income inequality is different in different cities and states across the country. At times in the past, mortality rates were higher and life expectancy lower in states with higher income inequality. But the relationship is much weaker today.
Historically, the South and Border States where mortality is high – West Virginia, Alabama, Kentucky, Mississippi, Arkansas, Oklahoma, Louisiana and Tennessee – had higher levels of income inequality than the majority of states. In most cases this was because they had large African-American populations, who were relatively poor, which drove up overall income inequality, and who had relatively high mortality, which drove up overall mortality.
States in the Central Plains and in most of the West had more homogeneous populations and lower mortality rates. Today, however, New York and California are two of the most unequal states; they have a high degree of heterogeneity, with large Hispanic and Asian populations, but have among the lowest mortality rates. To search for a simple and direct relationship from income inequality to mortality is to follow another false trail.
Many people feel that income inequality is less of a problem if it is easy for people to move from poverty to riches, or at least for children to do better than their parents. To check this, we need a measure of intergenerational mobility – for example, the fraction of children whose parents were in the bottomfifth of the income distribution and who managed to move to the top-fifth. When intergenerational mobility is high, we might suppose that everyone has a chance of succeeding (or failing!), and when it is low people are trapped by the accident of their birth.
The economist Raj Chetty and coauthors have calculated measures of intergenerational mobility for children born between 1980 and 1991 for different places across the United States. Children in the Southeast U.S. have had the least chance of moving up, at least for this particular cohort. While there is a good deal of overlap between low mobility and deaths of despair, the relationship is no closer than for inequality itself. Indeed, inequality is itself strongly related to low mobility.
Incomes and the Great Recession
The stock market crash of October 1929 was followed by a decade of misery and economic failure that remains the worst-ever crisis of Western capitalism. Millions lost everything – their jobs, their savings, their homes or their farms. More than 20 percent of the population was unemployed, depriving not only themselves but also their families of economic support. Personal income per head fell by a quarter from 1929 to 1933, and did not recover until 1937. Suicide rates reached peaks that they have never subsequently attained; this is true in both the U.S. and Britain. In Europe, the Depression and its aftermath saw the rise of fascism.
Nothing as bad as this has happened since, but the events after 2008 are the next worst. During the Great Recession, unemployment doubled, from 5 percent to 10 percent, not as bad as 20 percent, but no better for the millions affected. (For those with a college degree, unemployment peaked at only 5.3 percent.) Because the crisis was rooted in a housing bubble, with bankers making huge sums from mortgages that should never have been issued, millions of people lost their homes. People who had worked hard to hold on to a middle-class life were suddenly without a job, without a place to live and without the means to continue their own or their children’s educations. Banks stopped lending, and millions of businesses went bankrupt.
There have been many studies of how mortality varies over the business cycle, whether more people die in bad times – as we might at first expect – or during economic booms, when times are good. Perhaps the first study to look at the question was published as long ago as 1922, by the sociologist and statistician William Ogburn and sociologist and demographer Dorothy Thomas. Thomas was the first woman to be a professor at the Wharton School of the University of Pennsylvania.
Ogburn and Thomas discovered, to their surprise, that the good economic times were bad times for mortality. Their conclusion has been replicated many times since, including in work for the United States for national and state-level business cycles before the Great Recession. Much the same happens in other rich countries, with slumps better for mortality than booms, though not every study confirms the pattern. While it is true that suicides are higher in bad times, as was true in the Great Depression – remember the famous images of bankrupt ex-millionaires jumping out of skyscrapers in 1929 – there are other mechanisms at work. During slumps, people have less money to spend on hurting themselves by driving fast cars or drinking too much. Working less can reduce stress and heart attacks and, when wages are low and labor available, it can be cheaper to find good people to take care of the elderly.
Yet each boom and each slump is different and has its own history. In the Great Recession, quite apart from the economic disaster, there was also the epidemic of deaths that is the topic of this book. So, what happened this time?
An essential starting point is to look at the trajectory of deaths of despair after 1990. The upward trajectory rises inexorably, and there is no sign of an effect of the 2008 crash nor of its long aftermath. Suicides were certainly high after the crash, but they had been rising for many years before. Whatever the other consequences of the crash, there is no evidence of a jump in deaths of despair after Lehman failed, nor as unemployment doubled from the fall of 2008 through 2009. The idea that the crashed economy caused deaths of despair is another false trail.
Even so, the Great Recession may be implicated in other types of deaths, or in deaths for some groups but not others. For example, whites ages 45 to 54 saw their median household incomes per member rise through the 1990s, and then fall after 2000. Their all-cause mortality rate – not just from deaths of despair, but including them – fell from 1990 to 1999 and then rose until 2016, matching income trends in reverse, as one might expect. But if we look more closely, the match is coincidental.
The down and up in all-cause mortality happened because deaths of despair, although rising in the 1990s, were small at first, and were more than offset by improvements in death rates from heart disease.
As the decline in deaths from heart disease slowed and as deaths of despair became large enough, the all-cause mortality decline turned into a mortality increase. Yet the underlying components, deaths of despair and heart disease mortality, are each unrelated to the pattern of incomes, and that their sum mirrors incomes is simply coincidental.
Older whites saw their mortality rates fall throughout the period as their median incomes rose. The incomes of older Americans were better protected than those of people in middle-age – Social Security benefits for retirees have done better than median wages for people who are working. But all we have here is one trend up – incomes – and one trend down – mortality – and it would be a real leap to attribute the falling mortality rates of the elderly to their rising incomes.
Just as poverty fails to explain the spatial pattern of deaths, so does the pattern of incomes after 1990 fail to explain the pattern of deaths over the same period.
When we look more widely, it becomes even clearer that patterns of incomes from 1990 through 2017 do not match patterns of deaths. For example, while whites ages 45 to 54 with a bachelor’s degree earn more than those with some college who, in turn, earn more than those with no more than a high school degree, median household incomes per member for the three groups tracked one another – rising until 2000, and falling thereafter. But the mortality rates for the three groups diverged, rising for the least educated, almost flat for the middle group and falling for the most educated.
The link also fails when we compare blacks and whites, both of whom experienced the rise and fall of median income, but with sharply divergent mortality experiences – blacks good, whites bad. Just as poverty fails to explain the spatial pattern of deaths, so does the pattern of incomes after 1990 fail to explain the pattern of deaths over the same period.
We are not arguing here that incomes (or wages) do not matter. In the rest of the book, we put much of the blame for deaths of despair on the long-term deterioration in opportunities for less-educated Americans. The operative expression here is “long-term.” Our skepticism about the effects of the Great Recession has to do with the lack of a relationship between movements in mortality and movements in income in the past two decades, a much shorter period.
Recession, Austerity and Mortality in Europe
The Great Recession was different in the U.S. and Europe and, for many European countries, it was much worse, both in terms of increases in unemployment rates and declines in income. For countries that implemented austerity policies, voluntarily or involuntarily, there were cuts in unemployment benefits in spite of rising levels of unemployment, as well as cuts in health spending, especially on preventive services such as vaccinations and breast cancer screening, and on pharmaceuticals. Government spending for the elderly, such as pensions and long-term care, was largely protected.
Greece, which suffered the most, cut public expenditure on health by 30 percent, though even there spending on long-term care for the elderly was protected.
In spite of the austerity in some countries, or lack of it in others, there was (and is) no epidemic of deaths of despair in Europe. The slowdown and reversals in mortality rates in the United States have no general counterpart in Europe. Indeed, between 2007 and 2013, while unemployment rates in Greece and Spain more than tripled to the point where more than a quarter of the population was unemployed, life expectancy was rising more rapidly than in most European countries.
In Europe over this period there was some convergence in life expectancy. Countries that started out at the low end, such as Estonia, Poland and the Czech Republic, had faster growth in life expectancy than those, such as Norway, France and Switzerland, that already had relatively high life expectancy. But Greece and Spain are not explained this way; both had high life expectancy to start, and both had substantial mortality improvement over the austerity years. Looking to Europe does nothing to help us develop any general story of unemployment, declining incomes and mortality.
Is there another wealthy, industrialized country at risk of following the U.S. down the path to poorer long-run outcomes for working class families and deaths of despair? Storm clouds appear to be gathering in the UK. Low-earning working households have seen little rise in household earnings since the mid-1990s. Life expectancy, which rose at more than a fifth of a year per year from 1980 to 2011, has flat-lined since then. Like the U.S., progress on midlife heart disease mortality has faltered, and deaths of despair in England and (especially) Scotland have risen. (The numbers here are small relative to what we currently see in the U.S., but the U.S. also started from a small base in the early 1990s.)
Storm clouds appear to be gathering in the UK. Low-earning working households have seen little rise in household earnings since the mid-1990s. Deaths of despair in England have risen.
Britain is currently experiencing an extended period of austerity and of rising (geographically centered) inequalities – London is flourishing while most of the rest of the UK is not. Like the U.S., the country is politically divided, with half having voted for Brexit, and half for Remain. While the long-run effects on mortality remain unclear as of mid-2019, it is possible that the long-term decline in working class lives witnessed among U.S. whites from 1970 may be on its way to the UK, with deaths of despair beginning to stir. But there is, as yet, no clear and accepted understanding of recent changes in mortality in Britain.
Deaths and Deindustrialization
We are not quite done with income and unemployment. Some of the writing about the epidemic, such as Sam Quinones’s outstanding book, Dreamland, highlights opioid abuse and deaths in once-prosperous towns or cities where jobs have disappeared, where factories have been lost to automation or have moved abroad, and where at least some of the people who remain are abusing opioids.
Correlating deaths of despair with employment rates at a fine level confirms what Quinones saw. Places with a low fraction of the prime-age population in work are also places with high rates of deaths of despair; this holds for suicides, drug overdoses and alcoholic liver disease taken separately. Several studies have looked at a more specific episode, the accession of China to the World Trade Organization in 2000, and the sharp local increases in unemployment that resulted from the sudden competition from much cheaper Chinese goods. Again, those increases in unemployment are associated with increases in mortality.
Our main argument in this book is that the deaths of despair reflect a long-term and slowly unfolding loss of a way of life for the white, less-educated, working class. Unemployment is part of that story, but only a part. People who have given up on finding a job and are not looking are not counted as unemployed, but they still reduce the fraction of people in work. Unemployment rates rise and fall, certainly for the country as a whole, but also in specific places as one kind of job is replaced by another – often a worse job replacing a better job.
In some places where manufacturing has disappeared and people have lost jobs with a high-paying employer, they find other jobs, in services, or order fulfilment, or call centers, or driving for Uber. These jobs may pay less and working conditions may be more stressful, but they keep people in the labor force. The journalist Amy Goldstein tells such a story about Janesville, Wisconsin, Paul Ryan’s hometown, after General Motors – known as “generous motors” because of its high hourly wages – closed its plant in 2008 after making Chevrolets there for 85 years. By the end of the story, the unemployment rate is only 4 percent, yet that hardly means that all is well.
Lower unemployment rates are associated with lower rates of deaths of despair. Yet there may be others who lost their jobs and remain out of the labor force and out of the unemployment statistics. They, too, contribute to distress and to despair, so that deaths are high even when unemployment is low, but when there are many people doing nothing and with nothing to do.
In summary, unemployment is not always good at identifying the places where social and economic structures have been destroyed. Worse jobs are still counted as jobs, and when people give up altogether and stop looking for work, they are no longer counted as unemployed. But these changes, prolonged for long enough undermine social life and the structure of society, and it is that destruction that brings deaths of despair. The link between mortality and unemployment is part of this process, as is the “China shock.” It is important for our story that these results point in the same direction, but they are simply the latest installment in a larger and longer running movie.
The Great Recession Redux
We have emphasized that the Great Recession did not bring deaths of despair in the way that the Great Depression brought epidemics of suicide in the U.S. and Britain, but that does not mean that it did not matter. We suspect that the upsurge of populism on the right and of rage against inequality on the left have much to do with the financial crisis.
Until the crash, it was possible to believe that the elites knew what they were doing, that the salaries of the CEOs and the bankers were being earned in the public interest, and that economic growth and prosperity would make up for the ugliness of the system. After the crash, when so many ordinary people lost so much, including their jobs and their homes, the bankers continued to be rewarded and went unpunished, and politicians continued to protect them. Capitalism began to look more like a racket for redistributing upward than an engine of general prosperity.
related topics: Books