Good times make you sick
Introduction
Strong evidence has recently been provided that mortality increases when the economy temporarily improves. Using aggregate data for a panel of the 50 states and District of Columbia over a 20-year period (1972–1991), Ruhm (2000) indicates that state unemployment rates are negatively and significantly related to the total death rate and 8 of 10 specific causes of fatalities, with suicides the important exception. For instance, a 1 percentage point fall in unemployment is associated with 0.5, 3.0, 0.7, and 0.4% increases in mortality from all causes, motor vehicle fatalities, influenza/pneumonia, and cardiovascular disease. Compared to earlier research, this study has the advantage of estimating fixed-effect (FE) models that exploit within-state changes and so automatically control for time-invariant factors that are spuriously correlated with economic conditions across locations.1 Using similar methods, other research documents a procyclical fluctuation in fatalities using aggregate data for 50 Spanish provinces over the 1980–1997 period (Tapia Granados, 2002), 16 German states from 1980–2000 (Neumayer, 2002), and 23 OECD countries between 1960-1997 (Gerdtham and Ruhm, 2002).
However, at least two important questions remain unanswered. First, mortality could be procyclical while other aspects of poor health are not. For example, accident fatalities exhibit a particularly strong variation but may be only weakly related to previous health status.2 Second, individual-level relationships cannot be ascertained using aggregate data. Thus, the effects could vary by age, sex, ethnicity, or employment status. These issues are addressed below. Microdata for persons aged 30 and over from the 1972–1981 years of the National Health Interview Survey (NHIS) are used to examine how health fluctuates with state economic conditions after controlling for personal characteristics, time-invariant factors, general time effects, and (usually) state-specific time trends. The proxies for health include the prevalence of medical conditions, specific morbidities, activity limitations, and the utilization of medical care.
The major finding is that most measures of health decline as the economy improves. In the preferred specifications, a 1 percentage point drop in the state unemployment rate is associated with a statistically significant 1.5% rise in the probability that respondents report one or more medical problems and a 1.2% (1.6%) growth in the probability of one or more “restricted-activity days” (“bed-days”) during the previous 2 weeks. The cyclical variation is especially pronounced for males, employed persons, and those of prime-working age.3 The negative health effects of sustained economic expansions persist or accumulate over at least a 3-year period and occur despite a possible procyclical variation in the use of medical care.
The fluctuation in health is more pronounced for acute than chronic problems. Thus, the one point decline in unemployment is estimated to raise the percentage of individuals reporting at least one acute (but no chronic) condition by a statistically significant 3.9%, versus a statistically insignificant 1.1% growth in the prevalence of chronic ailments. However, there is considerable variation across types of chronic morbidity. For instance, the drop in joblessness is correlated with a 4.3% (8.7%) rise in ischemic heart disease (intervertebral disk problems) but a 7.3% decrease in non-psychotic mental disorders.
One concern is that worse measured health during good times might reflect better access to medical care (e.g. due to higher incomes or enhanced insurance coverage) resulting in improved identification of existing problems, rather than a deterioration in actual status. The evidence does not support this possibility. The counter-cyclical variation is generally stronger for employed persons under the age of 65 (for whom access to medical care is less of an issue) than for the full sample, while income has a protective effect on measured health.
Section snippets
Why might good economic times make you sick?
Many researchers hypothesize that cyclical upturns benefit health by reducing the stress associated with economic insecurity (e.g. Brenner and Mooney, 1983, Catalano and Dooley, 1983, Fenwick and Tausig, 1994). However, there are at least three reasons why health might instead worsen. First, non-market “leisure” time decreases making it more costly for individuals to undertake time-intensive health-producing activities such as exercise.
Econometric methods
The basic regression specification estimated below is:where H measures the health or medical care utilization of individual i in state j at time t, X is a vector of personal characteristics, E the macroeconomic variable (typically the unemployment rate), α a year-specific intercept, S a state fixed-effect, and ε a disturbance term.
The year effect holds constant determinants of health that vary uniformly across states over time; the fixed-effect accounts for those that
Data
Data are from the 1972–1981 years of the National Health Interview Survey (NHIS), which is conducted by the National Center for Health Statistics at the Centers for Disease Control and Prevention. The NHIS is designed to assess the amount and distribution of illness, disability, and chronic impairment of the civilian non-institutionalized population of the US, as well as the type and duration of health services received by them.
Economic conditions and health status
Table 1 summarizes the results of regression models estimating the relationship between economic conditions and health status. The first three columns refer to the full sample, the next three columns to 30–64-year-olds employed at the survey date, and the last three to working men aged 30–55. For each group, the first column displays (weighted) mean values of the dependent variables, while the second and third show predicted effects of a 1 percentage point increase in the state unemployment
Medical care
Table 2 provides evidence that the use of medical services may increase when the economy strengthens, although large standard errors require caution in interpreting the results. A 1 percentage point decline in unemployment is associated with an imprecisely estimated 0.11 point (0.9%) rise in the probability of hospitalization during the previous year and a 0.3 point (0.4%) growth in the likelihood of having visited a doctor.
Two factors make it doubtful that the higher use of medical care
Subsamples
Table 3 summarizes findings for subsamples stratified by age, sex, and race. The first column restricts analysis to 30–55-year-old respondents who are of “prime working age”; the next four provide results for males, females, whites, and blacks.35 Small sample sizes often lead to imprecise estimates, limiting our ability to make comparisons across groups. Nevertheless, the macroeconomic effects appear to be
Specific chronic conditions
Specific limiting chronic morbidities are examined in Table 4. For expositional convenience, the predicted effects of a ten percentage (rather than one percentage) point rise in the state unemployment rate are shown. The most striking finding is the strong procyclical pattern of heart disease, the leading cause of mortality. This is completely accounted for by changes in ischemic heart problems, where a one point fall in unemployment is associated with 4.3, 13.3, and 12.8% increases in
Income effects
Permanent income growth is expected to improve most aspects of health (e.g. by allowing purchases of more advanced medical care and health-preserving consumption goods, such as safer cars). These effects could be muted or reversed, however, when considering transitory changes in incomes. Moreover, even a protective effect of earnings growth could be overwhelmed during cyclical upturns by heightened job stress or greater time costs of health-enhancing activities. The role of income is addressed
Adjustment paths
Macroeconomic conditions have been assumed to have only a contemporaneous impact to this point. There are at least two reasons why this might not be so. The impact of health investments and job-related stress are likely to accumulate over time, if these represent flows that gradually alter the stock of health capital (Grossman, 1972), implying larger medium-term than short-run effects. On the other hand, agents have greater flexibility in making consumption, time allocation, and production
Discussion
Most aspects of health worsen when the economy temporarily improves. In the preferred models, a 1 percentage point fall in unemployment is estimated to raise the prevalence of medical problems, acute morbidities, restricted-activity days, bed-days, ischemic heart disease, and intervertebral disk disorders by 1.5, 3.9, 1.2, 1.6, 4.3, and 8.7%. The deterioration in health is particularly strong for persons of prime working age, employed individuals under the age of 65, and men. It occurs despite
Acknowledgements
I thank McKinley Blackburn, Donna Gilleskie, Mark Loewenstein, Ken Snowden, and seminar participants at the Southern Economic Association Meetings, International Health Economics Association Third International Conference, Conference on Social Insurance and Pension Research, University of South Carolina, and the Duke-UNC Health Workshop for helpful comments and William Black for research assistance. Financial support from the National Science Foundation (SES9876511) is gratefully acknowledged.
References (51)
Unemployment and health selection
The Lancet
(1996)- et al.
Unemployment and health in the context of economic change
Social Science Medicine
(1983) New evidence on the relationship between income and health
Journal of Health Economics
(1996)A note on economic conditions and alcohol problems
Journal of Health Economics
(1999)- et al.
Mortality and unemployment: a critique of Brenner’s Time Series Analysis
The Lancet
(1981) - et al.
Psychiatric illness and risk factors for suicide in Denmark
The Lancet
(2000) Economic conditions and alcohol problems
Journal of Health Economics
(1995)- et al.
Does drinking really decrease in bad times?
Journal of Health Economics
(2002) Time series analysis of the relationship between unemployment and mortality: a survey of econometric critiques and replications of Brenner’s studies
Social Science and Medicine
(1985)- et al.
Adjusting for multiple testing when reporting research results: the Bonferroni versus Holm methods
American Journal of Public Health
(1996)
The study of stress at work
Annual Review of Public Health
Trends in alcohol consumption and associated illnesses: some effects of economic changes
The American Journal of Public Health
Mortality and the national economy
The Lancet
The health effects of economic insecurity
American Journal of Public Health
Health effects of economic instability: a test of the economic stress hypothesis
Journal of Health and Social Behavior
Homicide and economic conditions: a replication and critique of M. Harvey Brenner’s new report to Congress
Journal of Quantitative Criminology
Preferences over inflation and unemployment: evidence from surveys of happiness
American Economic Review
Personal and aggregate unemployment and psychological problems
Journal of Social Issues
Traffic safety and the business cycle
Alcohol Drugs and Driving
Cited by (388)
“Deaths of despair” over the business cycle: New estimates from a shift-share instrumental variables approach
2024, Economics and Human BiologyWhen the economy falters, hearts suffer: Economic recessions as a social determinant of health in cardiovascular emergencies
2024, American Journal of Emergency MedicineThe health consequences of returning to work after retirement: Evidence from a Japanese longitudinal survey
2024, Economics and Human BiologyEconomic conditions and health: Local effects, national effect and local area heterogeneity
2023, Journal of Economic Behavior and OrganizationThe long-term impact of the COVID-19 unemployment shock on life expectancy and mortality rates
2023, Journal of Economic Dynamics and ControlProtective effects of health insurance against disasters: an insight from the Great East Japan Earthquake
2022, Journal of Pension Economics and Finance