UC Berkeley

Americans in the twenty-first century are dying earlier in life and at higher rates from preventable causes than in nearly any other developed economy. Understanding of the root determinants of the recent reversal in life expectancy and identifying policy approaches to combat the rise in midlife mortality is a national public health and economic imperative.This dissertation focuses on the well-documented increase in fatal drug overdose, suicide, and alcohol-related mortality—a collection of causes of death often referred to as the “deaths of despair”—and examines the potential economic determinants of the acceleration in these causes of death over the past several decades. Building upon extensive literature examining macroeconomic and labor market conditions as upstream factors shaping population health, the following chapters consist of two empirical analyses intended to estimate the causal effect of short- to medium-term changes in local employment rates on these causes of death among working-age adults during the 2003-2017 period. These studies are of increasing importance as the United States continues to experience widespread employment uncertainty and prolonged economic distress in the wake of the COVID-19 pandemic.

The first study presented in this dissertation focuses on the effects of county-level employment conditions on “deaths of despair” using a Bartik-style shift-share instrument to isolate demand-driven variation in county-level employment rates. In line with most existing studies that document countercyclical variation in suicide, I estimate that a one percentage point increase in the current-year employment-to-population ratio decreases non-drug suicide rates by one to two percent. On the other hand, my causal models suggest that rates of fatal drug overdose increase by a similar magnitude as the economy improves, and I find no evidence of changes in alcohol-related mortality in response to short-term employment shocks. I conduct a simulation exercise based on these point estimates to show that in general, and especially for accidental drug overdose, these estimated effects are small relative to the increases in cause-specific mortality over the 2003—2017 period.

Motivated by the procyclical variation in accidental drug overdose uncovered in the first study, the second analysis examines the extent to which county employment rates affect the demand for prescription opioid medication among a population of commercially insured adults. This study draws on de-identified, individual-level pharmacy and medical claims from 2003—2017 aggregated to the county level to test the hypothesis that county-level employment fluctuations differentially affect the demand for prescription opioids that place individuals at higher (versus lower) risk for abuse and dependence. Unlike existing studies, I find no evidence of an effect of employment conditions on the demand for prescription opioids overall or differential effects between high- and low-risk prescriptions.

The relatively small magnitude of the estimated effects, suggestive evidence of heterogeneity across demographic groups, and mixed findings on the cyclicality of these causes of death over various time horizons all point to a more complex set of factors underlying the rising rates of “deaths of despair” that is not explained by local employment rates alone. Developing a more nuanced understanding of these trends—particularly along key dimensions such as race/ethnicity and socioeconomic status—will be critically important in designing equitable policies to help the country recover from the COVID-19 pandemic and to reverse the disconcerting trends of increasing midlife mortality in the years to come.