Killer Debt: The Impact of Debt on Mortality

Debt has increasingly become a central component of modern household and individual finance in the United States.  As of 2011, approximately 69 percent of all U.S. households held some form of debt with median household debt of $70,000 in 2011, up from $50,971 in real dollars in 2000 (Vornovytskyy, Gottschlack and Smith, 2015).  In tough economic times, changes in the composition and severity of debt within a household are a powerful measure of how that household is impacted, especially when compared to other broad measures of economic distress such as unemployment, which does not necessarily capture the nuance of an individual’s financial situation.

A number of studies have found a negative correlation between debt and health (Drentea and Lavrakas, 2000; Lyons and Yilmazer, 2005; Keese and Schmitz, 2010; Lau and Leung, 2011; Averett and Smith, 2014), but there are reasons to be concerned that these associations may not be causal. First, poor health, through a reduction in work and earnings and an accumulation of health care expenses, can cause debt and unobservable characteristics, such as a risk-taking personality or impulsivity, may be correlated both with debt and health.

There are, however, plausible reasons to expect a causal relationship between debt and poor health. High levels of debt could lead to an increase in stress and stress-related behaviors such as poor nutrition and substance use that can have negative health consequences.  Stress has been linked to poor physical and mental health and financial insecurity may reduce health-care use and adherence to medical treatment plans leading to worsening health.

We use the individual panel component of the Federal Reserve’s Consumer Credit Panel (CCP). The CCP is a nationally representative 5% random sample of US consumers and their household members with information in the consumer credit data system. The panel follows these individuals from 1999 to the present. The data contains credit balances and delinquencies (bad debt) for different categories of debt.  The CCP includes an indicator of the quarter of death for individuals who died while part of the panel.  The longitudinal data allows us to track financial well-being and health outcomes in a single dataset.  We track an individual’s debt accumulation and delinquency, by type of debt, leading up to the time of death.   

To address reverse causality (bad health causing debt) we identify the impact of an adverse economic event on debt, in our case, the housing crisis of 2006-2010.  This information is then directly interacted with individual debt to see how debt and mortality change based on differential shocks. By identifying adverse economic conditions in advance of individual debt and our health outcome, the ITT estimator removes the impact of endogenous migration.  We also look at the direct impact of fluctuations local area economic conditions as measured by aggregate debt (jack-knifing out the individual).

Our results thus far are consistent with a negative association between individual debt and health.  In our initial models we find that individuals  with lower riskscores (better credit ratings) have a lower probability of mortality.  Addressing issues of reverse causality by using lagged local area averages of debt default confirms our initial finding.