The data presented in previous chapters suggest that social and economic factors have contributed to mortality trends in the United States. During the study period (1990–2016), for example, the increase in working-age (ages 25–64) mortality was greatest among adults with less education. As shown in Chapter 3, large increases in working-age mortality occurred in the industrial Midwest and central Appalachia, areas deeply affected by the collapse of manufacturing plants and coal mines, on which many communities depended for economic vitality and stable employment. Working-age mortality rates increased during a period in the United States in which middle-class and low-income people faced reduced access to well-paying jobs, rising housing and health care costs, and difficulties ensuring that their children could obtain a good education and climb the economic ladder. Research has shown that exposure to prolonged economic adversity may affect health outcomes via multiple mechanisms, including gaps in health care; chronic stress;1 anxiety; depression; and unhealthy coping behaviors, from smoking and overeating to drug and alcohol abuse, suicide, and violent crime.
The complex interrelationships among economic conditions, place, and time, together with the inability to conduct controlled experiments, make it difficult to prove causal associations or confidently isolate the health impacts of economic trends (Gonsalves, 2019). Nevertheless, a growing body of literature provides some insight. For example, ethnographic
1 Chronic stress is itself biologically harmful to health, with known effects on neuroendocrine function, the immune system, and epigenetic transmission.
studies reviewed by the committee document the withering influence of economic pressures on the health of communities and their vulnerability to chronic stress, anxiety, substance abuse, depression, and suicide (Chen, 2015; McLean, 2016; Silva, 2019; Thompson et al., 2020). Much of the other empirical research examines changes in overall mortality, but some focuses on specific causes of mortality, with growing interest in deaths due to suicide and substance abuse. This chapter examines evidence of the relationship between selected economic factors and mortality. The focus is on general economic conditions, economic fluctuations in employment, plant closures, trade pressure, and economic inequality. Related discussion of policies intended to improve economic well-being (e.g., minimum wage laws) is included in Chapter 11.
Assessing the relationship between economic conditions and mortality is difficult for a number of reasons. First, the impacts of intermediate- and short-term economic changes may differ from those of more sustained changes, and empirical analyses of long-term impacts are challenging because of many intervening factors. Second, confounding and complex interactions may mask effects. For example, economic deprivation may be correlated with education, making identification of the pure effects of economic well-being difficult to isolate.
Moreover, different demographic groups may experience economic conditions differently, and economic effects may interact with other time-varying contextual and environmental factors, such as the availability of opioids. For example, economic factors may give rise to physical and psychological pain, but the connection between that pain and mortality may reflect the introduction of opioids. Thus the combination of pain and opioids may have created conditions that led to significant mortality, while the absence of either might have greatly dampened the dramatic rise in mortality that occurred.
Finally, the effects of changes in economic conditions may vary based on the reasons behind the changes. The effects of plant closings, for example, may differ from those of cyclical recessions. Likewise, broad-based economic decline in communities may have a broader effect relative to individual economic setbacks.
A considerable body of research demonstrates that lower-income Americans have worse outcomes relative to their wealthier counterparts. Thus,
adults ages 50 and above living in poorer areas have higher mortality rates than those living in wealthier areas (Currie and Schwandt, 2016). The difference in life expectancy between the wealthiest 1 percent and the poorest 1 percent of individuals in the United States is approximately 15 years (Chetty et al., 2016). The increase in life expectancy from 2001 to 2014 was not uniform: life expectancy increased by 2.34 years and 2.91 years for men and women, respectively, in the top 5 percent of the income distribution but by only 0.32 year and 0.04 year for their counterparts in the bottom 5 percent. Within the bottom income quartile, life expectancy was approximately 5 years longer in geographic locations with the highest versus those with the lowest longevity. The lower life expectancy among the latter individuals was significantly correlated with health behaviors such as smoking, but not with physical environment factors, labor market conditions, income inequality, or access to medical care. Life expectancy for these individuals also was correlated with government expenditures, fraction of college graduates, and fraction of immigrants.
Seminal research by Case and Deaton (2015) documents trends in mortality associated with economic status. Specifically, using death records and national survey data, these authors found that among those ages 45–54, mortality among non-Hispanic Whites (Whites) rose by 34 per 100,000 population between 1999 and 2013—approximately 0.5 percent per year. This increase in working-age mortality was driven largely by increases in drug and alcohol poisoning and suicide, known as “deaths of despair” (see Chapters 7 and 8, respectively), among those with a high school education or less, along with a slowdown in progress against mortality from heart disease and cancer, the two leading killers in middle age (see Chapter 9). These trends were accompanied by increases in morbidity, including deterioration in self-reported physical and mental health, and rising reports of chronic pain. Case and Deaton (2017, 2020) later extended their analysis and proposed a hypothesis whereby poor labor market conditions and cumulative disadvantage experienced by successive cohorts (delineated by birth year) can help explain this increased mortality (Case and Deaton, 2017, 2020). These basic associations are supported by the committee’s analysis.
Although the evidence relating economic disadvantage to mortality is strong, the relationship is complex. Some studies have found the effects of economic despair on mortality for Whites but not for non-Hispanic Blacks (Blacks) or Hispanics (Hollingsworth, Ruhm, and Simon, 2017; Pierce and Schott, 2020). Case and Deaton (2017, 2020) found that although mortality is higher in an absolute sense among Blacks than Whites, mortality rates among Blacks still improved until about 2010. Thereafter, absolute year-to-year increases in working-age mortality among non-White populations during the study period (1990–2017) matched or exceeded those among Whites, suggesting that past differences by race may have reflected timing more than
fundamental protections from the forces that contributed to increased mortality among working-age adults. The committee’s analysis supports the notion that recent trends in increased mortality extend beyond Whites.
Moreover, while drug (opioid) overdoses accounted overwhelmingly for rising working-age mortality during the study period, alcohol-related diseases, suicides, unintentional injuries, and organ diseases contributed as well. The breadth of factors contributing to increases in working-age mortality in the past decade—from external causes such as substance abuse to diseases involving multiple body systems and pathophysiological processes—eludes attribution to a single cause, such as opioids or obesity. Rather, it suggests the possibility of one or more upstream systemic causes, such as economic disadvantage or increasing stress that can affect health across multiple pathways and disease processes (Woolf et al., 2018). Yet as noted above, the relationship between mortality and economic disadvantage is complex, and it may be exacerbated or ameliorated by mediating factors. Different groups may experience economic deprivation differently, or economic well-being relative to expectations may matter more than economic well-being per se. But in any case, differential effects by geography and race suggest a more complex story than the conclusion that economic deprivation leads to increases in mortality.
One way to isolate the impact of economic conditions on mortality is to examine the effects of specific economic shocks. Research typically focuses on two types of shocks: job loss, particularly when caused by plant closings, and trade/import competition.
One body of quasi-experimental analysis looks at specific economic shocks at the individual level and tends to find that such shocks lead to poor health and mortality outcomes. One such shock examined in the literature is job loss. The literature suggests that the immediate impact of losing a job does not explain suicide risk, but being part of a mass layoff event or being unemployed for an extended period of time appears to be associated with increased suicide risk (Classen and Dunn, 2012). Sullivan and von Wachter (2009) suggest that for men, late-career job loss is associated with mortality rates in the year after displacement that are 50–100 percent higher than would otherwise be expected. The effect on mortality declines sharply over time, but even 20 years after displacement, the authors estimate a 10–15 percent increase in annual death hazards. In their study, they compare mortality rates among displaced workers with those among their nondisplaced peers and also find that workers with larger losses in earnings see more substantial increases in mortality rates relative to those with smaller earnings losses. A related study from the Netherlands found that after controlling for
worker characteristics, workers who lost their jobs because of firm closure had a 34 percent or 0.60 percentage point increase in probability of death within 5 years (Bloemen, Hochguertel, and Zweerink, 2018).
In other research, data from Denmark suggest that job loss due to a plant closure, particularly in the few years following the loss, increases the risk of overall mortality and of death from circulatory disease; suicide and suicide attempts; and death and hospitalization due to traffic accidents, alcohol-related disease, and mental illness (Browning and Heinesen, 2012). According to these data, the risk of overall mortality is 79 percent higher in the year of displacement, 35 percent higher 1–4 years after displacement, 17 percent higher 1–10 years after displacement, and 11 percent higher 1–20 years after displacement. Importantly, Venkataramani and colleagues (2020) document that the increase in opioid mortality between 1999 and 2016 was significantly greater in manufacturing counties that had experienced the closure of an automobile plant (Venkataramani et al., 2020). They estimate that 5 years after a county experiences a plant closure, mortality due to opioid overdoses will increase in that county on average by 8.6 deaths per 100,000 population compared with unexposed counties, an 85 percent increase relative to the preclosure mortality rate.
Overall, job loss appears to have negative impacts on certain health-related behaviors. However, employability (and the ease with which those laid off can find new employment) does appear to mitigate the effects of job loss, so these effects may be concentrated among an at-risk subset of workers (Deb et al., 2011).
Another line of quasi-experimental research examines the impact on health of adverse economic conditions attributable to international trade. Specifically, while international trade may be beneficial for overall economic growth, increased trade can prove disruptive to the domestic manufacturing industry, resulting in layoffs, plant closings, and reduced earnings for exposed workers (Autor et al., 2013). The particular role of import competition is important. Studies of the relationship between imports and mortality have found that areas more versus those less affected by imports experience worse health outcomes. For example, a $1,000 increase in competing Chinese imports per worker corresponds to a 7.8 percent increase in the morbidity of poor mental health, adding approximately 3 days of poor mental health per year for the average adult (Lang, McManus, and Schaur, 2019). Additionally, firms particularly susceptible to import competition may make decisions that put workers’ health at greater risk. Injury risk, for example, increases by 13 percent at the smallest plants (McManus and Schaur, 2016). Thus it is important to look at the effects of economic
shocks, which may be sudden and geographically concentrated, on those most acutely affected.
Taking advantage of differential exposures to trade liberalization due to Congress’s granting of Permanent Normal Trade Relations (PNTR) to China in 2000, Pierce and Schott (2016) found an increase in mortality due to “deaths of despair.” Counties in the 75th percentile of exposure to PNTR had a 0.42–0.63 higher suicide mortality rate per 100,000 population compared with those in the 25th percentile, accounting for 4–6 percent of the average age-adjusted suicide mortality rate across counties. Likewise, shifting a county from the 25th to the 75th percentile of exposure to PNTR was associated with an almost 30 percent higher mortality rate from accidental poisonings, which include drug overdoses. This increase in drug-related mortality was observed across a large portion of the working-age population (most age bins in the 20–54 age group). Importantly, this association was observed only among Whites, consistent with the notion that non-Whites are less affected by job loss (see Chapter 8). Similarly, focusing on young adults ages 18–39, Autor, Dorn, and Hanson (2019) found an increase in mortality due to increased import exposure. According to these authors, the average decade-level rise in import exposure induced an additional 64.4 male relative to female deaths per 100,000 population (of each gender) per decade (Autor, Dorn, and Hanson, 2019). Furthermore, manufacturing trade shocks were found to cause significant increases in male mortality due to drug and alcohol poisoning, HIV/AIDS, and homicide (Autor, Dorn, and Hanson, 2019).
Despite clear evidence relating economic disadvantage to mortality and demonstrating the deleterious economic consequences of job loss and trade competition, market-level economic fluctuations due to the business cycle are not strongly associated with increased mortality, including that related to opioids, at the population level. Although one might expect worse economic conditions to lead to worse health outcomes, a large body of literature suggests a potentially counterintuitive effect—that increases in the unemployment rate actually lead to a decrease in mortality. Ruhm (2000), for example, estimates that a 1 percentage point increase in the state unemployment rate decreased total mortality by 0.5 percent. Similar patterns appear to hold in other industrialized countries, so this effect apparently is not unique to the United States (Gerdtham and Ruhm, 2006). These results appear to be explained in part by a reduction in the motor vehicle deaths and other accidents that occur among those younger than 45
during economic upswings. In addition, nursing homes experience severe shortages of aides when the economy is strong, explaining why higher mortality among the elderly is expected (Stevens et al., 2015). Despite finding decreases in mortality during economic downturns, however, Ruhm (2000) did find an exception for suicides, estimated to rise by 1.3 percent with a 1 percentage point increase in the state unemployment rate. Effects on mental and behavioral health and physical health thus appear to differ.
The evidence relating fluctuations in economic conditions and drug-related mortality is mixed. Research from the United Kingdom has shown that a 1 percentage point increase in unemployment leads to an increase of 18–25 percent in opioid doses prescribed per capita (Vandoros, Gong, and Kawachi, 2020). Relatedly, another study found that as the annual county unemployment rate increased by 1 percentage point, the opioid death rate increased by 3.6 percent, and the rate of emergency department visits due to opioid overdose increased by 7.0 percent (Hollingsworth, Ruhm, and Simon, 2017). Quasi-experimental studies using somewhat longer periods of analysis suggest a weaker effect. Indeed, Ruhm (2018b) found that after controlling for various demographic and geographic variables, changes in economic conditions (including unemployment rates and import exposure) explained less than one-tenth of the observed increase in drug deaths that occurred from 1999 to 2015 and even less of the growth in mortality involving opioid analgesics or illicit opioids. Ruhm’s analysis differs from those of other authors cited earlier, such as Hollingsworth, Ruhm, and Simon (2017), because of the nature and length of the economic changes studied. The latter studies link changes in mortality to short-term economic changes (often in the local unemployment rate). By contrast, Ruhm (2018b) looks at medium-term economic changes that incorporate changes in a 3-year average poverty rate, in a 3-year average of median household income in 2015 dollars, in a 4-year average of median home prices, and in a 3-year average of the unemployment rate, as well as a measure of import competition.
The evidence that excess mortality is concentrated in disadvantaged populations but that fluctuations in economic conditions are related only weakly to mortality may appear contradictory. Yet these findings are consistent with the view that economic disadvantage makes populations susceptible to developments that threaten health. Over the past decades, the most significant of such developments was the introduction and diffusion of new opioids. Because disadvantaged populations suffer more from pain relative to the more advantaged, they would have been prescribed more pain medications during this period and may have been more susceptible to the opioid crisis as a result (Poleshuck and Green, 2008).
The relation between income inequality and health is complex, as it can involve several different causal and noncausal processes. A number of studies have shown that areas (e.g., countries, regions, cities, neighborhoods) with higher income inequality tend to have worse population health (e.g., higher mortality) (Backlund, Sorlie, and Johnson, 2007; Beckfield, 2004; Ben-Shlomo, White, and Marmot, 1996; Kaplan et al., 1996; Kennedy, Kawachi, and Prothrow-Stith, 1996; Kondo et al., 2009; Wilkinson, 1992; Wilkinson and Pickett, 2010). At least three different processes could contribute to these associations.
First, it is well established that income (at the individual or family level) is strongly related to health in a graded fashion (although a threshold effect and possibly declining benefit at high incomes have been documented in some studies) (Braveman, Egerter, and Williams, 2011). Because areas with higher income inequality often have larger proportions of the population at lower income levels, areas with more inequality will have worse health simply because of differences in the distribution of income.
Second, it has been hypothesized that income inequality may itself be related to worse health (even after accounting for absolute income) because of the psychosocial consequences of living in an unequal society. Living in a society that is more unequal could worsen health because the experience of inequality generates stress and reduced social cohesion and trust (Daniels, Kennedy, and Kawachi, 2000; Hastings, 2018; Subramanian and Kawachi, 2004; Wilkinson, 1996). Variants of this hypothesis posit that these adverse psychosocial impacts of income inequality are more pronounced (or only present) in persons of lower income, who therefore experience more adverse effects from living in a more unequal society.
Third, it has been hypothesized that income inequality may be related to worse health (even after accounting for absolute income) because societies or areas with less income inequality tend to have more egalitarian social policies or investments that benefit population health (Kawachi and Kennedy, 1999; Kawachi et al., 1997). According to this hypothesis, it is not the inequality that is the problem but the lack of public response to it. As with the hypothesis focused on inequality as a stressor, it has sometimes been posited that these effects are more pronounced in persons of lower income, for whom the social safety net is more important. Of course, an interesting question is whether lower income inequality is the cause or the consequence of these more egalitarian policies.
Research on whether there is or is not an income inequality effect on mortality (or other health outcomes) that is independent of absolute income has generated substantial debate (Lynch et al., 2004; Mellor and Milyo,
2001). Much of the debate hinges on whether analyses properly account for other population-level factors. For example, Deaton and Lubotsky (2003) argue that studies demonstrating a state income inequality effect did not properly control for race or other population-level characteristics that are differentially distributed across states and correlated with inequality. Deaton (2003) suggests that inequality may not be the driving factor so much as the correlate of more fundamental driving factors (Deaton, 2003). Others, however, have presented evidence that disputes these arguments (Ram, 2005; Wolfson and Beall, 2017).
Regardless of whether there is or is not an identifiable independent effect of income inequality on health (after individual-level income is accounted for), what is clear is that societies with larger income inequality tend to have worse health.
The relationship between economic conditions and mortality is complex. Descriptive analysis of trends suggests that disadvantaged White populations relative to other racial/ethnic groups have experienced a disproportionate increase in mortality, but recent data suggest that other disadvantaged populations have experienced worse mortality trends as well.
The evidence related to the impact of a change in economic conditions on mortality is less conclusive. Quasi-experimental analyses of job loss, plant closings, and disruption from foreign trade support the notion that economic disruption increases mortality. But these studies focus on assessing changes in economic well-being due to a narrow set of causes, such as plant closings and trade disruptions, as opposed to such broader causes as technical change and general economic trends related to productivity, including automation and a host of related processes. Furthermore, not all findings from these analyses are robust across specifications. Other analyses examining changes in county-level economic conditions and mortality have found little effect of economic factors on drug mortality (a major component of increased working-age mortality), but a major impact of opioid availability.
The best interpretation of the relationship between economic well-being and mortality suggests that economic hardship is associated with higher mortality, especially in the context of widespread availability of potent and life-threatening medications. However, the overall impact of the direct economic shocks that have been examined (i.e., short-term changes in economic circumstances) may be modest, and more work is needed to quantify the magnitude of this impact. A more important connection between economic deprivation and mortality may arise as people of lower socioeconomic status (SES) are exposed to a series of such shocks, whose cumulative
effect is large even if each individual shock is small. These individuals also may be more susceptible to adverse events (e.g., the introduction of opioids) when they occur. For example, deterioration in economic conditions may have led to some excess mortality, but the greater effect is that as opioids were introduced, lower-SES individuals were more affected. This pattern may be causal and specific to opioids because lower-SES individuals were more likely to have pain and thus more likely to be prescribed opioids, or it may be a general phenomenon whereby lower-SES populations are more affected when events increase overall mortality.
Much more work remains to be done to investigate the relationship between economic factors and mortality (see Recommendation 11-6 in Chapter 11). Economic research highlights several economic factors that influence income and may thus influence health: technological change, including automation; expansion of global trade to countries with lower wages; and policy changes, including making unionization more difficult and increasing the pay of chief executive officers (Autor, 2010). There is significant economic debate about the importance of each of these factors. The general sense is that technological change is likely the most important issue in differential wage trends by demographic group, but there is debate around this point.
For understanding health trends, the key issue is how these changes influence economic outcomes and health among different population groups, directly or indirectly. As technology changes, it—along with other factors—affects the income distribution and returns to education. More broadly, technological changes affect the nature of work, which in turn affects the stability of employment relationships and the overall financial risk faced by individuals over the course of their lifetime. Understanding the impact of this risk, as well as the effects of frequent job transitions on health, is important.
A second area of needed research is understanding how economic conditions interact with other factors in affecting mortality. For example, changes in medium-term economic conditions in themselves do not appear to have led to a very large increase in opioid deaths, but the increase in opioid availability may have contributed to a large increase in mortality because of a reservoir of susceptible individuals resulting from economic issues.
A third important set of questions involves understanding the relationship between the duration of economic hardship and mortality. Short-term or transient economic shocks may be less salient than longer-term declines in economic well-being. Similarly, the variability of economic status itself may have important consequences for health that are not well understood.
A fourth issue that deserves more analysis is how public programs to alleviate economic deprivation affect mortality (see Chapter 11 for a discussion of specific policies designed to improve economic well-being, such as the Earned Income Tax Credit and an increase in the minimum wage). The benefits of employment may extend beyond the associated economic rewards, and replacing income earned from wages with public support may not lead to the same outcomes.
Another important set of research questions is related to how economic well-being is measured, questions that are important both substantively and methodologically. Taking the example of how to measure economic hardship, many studies in this area focus on unemployment rates. Yet unemployment is measured relative to the labor force, and adverse economic conditions may cause people to leave the labor force. Conceptually, therefore, it may be better to measure the employment rate among working-age adults. Similarly, many measures of economic well-being focus on income levels, but relative income may also matter. Individuals with a given level of income may be worse off if they previously were wealthier than if they always had that income. Similarly, although research has examined the role of economic inequality in health, more work could be done to examine whether measuring absolute income or income relative to others in the community is a better predictor of health outcomes. Research is needed to illuminate how income should be measured—absolute or relative—how to count access to social programs, or whether to use point-in-time versus lifetime income. Finally, it is important to assess the extent to which individual-level versus community-level income (or wealth) may be important because the latter may influence a range of factors that affect people beyond their individual incomes, such as area traits (e.g., availability of fresh food) and tax revenue that can be used to fund social supports.
This page intentionally left blank.