Recessions lower (some) mortality rates:: evidence from Germany
Introduction
The effect of short-term economic fluctuations on the state of health in general and mortality in particular has found the interest of scholars since Forbes & McGrego (1975), Forbes & McGrego (1979), Forbes & McGrego (1987), Brenner's (1995), Brenner's (1973) path-breaking work. Brenner found with the help of time-series analysis that recessions are associated with deteriorating health in the United States, England and Wales and Sweden. Others have failed to find analogous evidence in a replication of his work on other countries or time-periods (Forbes & McGregor 1984; Wagstaff 1985; Joyce & Mocan 1993). The major advantage of Ruhm's (2000) analysis of the effect of unemployment rates on mortality rates in US states over the period 1972–1991 is the use of panel data. He finds that ‘state unemployment rates are negatively and significantly related to total mortality and eight of the ten specific causes of fatalities’ (Ruhm 2000, p. 617, emphasis in original). In other words, mortality behaves pro-cyclically as it moves with the business cycle. Panel data analysis has the advantage over time-series analysis that it can control for time-invariant state-specific effects and thus control for a potentially important source of omitted variable bias.
Gerdtham and Ruhm (2002) find similar evidence to Ruhm (2000) in pooling data from OECD countries. Ruhm (2001) himself has repeated his analysis on individual rather than aggregate data from the 1972–1981 US National Health Interview Surveys, coming to very similar conclusions as his analysis with aggregate data. In particular, he finds that the number of medical problems, the prevalence of acute morbidities and the number of reported “bed-days” all fall in economic recessions. Gerdtham and Johannesson (2002), on the other hand, find in their analysis of individual data from Sweden that some aspects of male mortality move counter-cyclically rather than pro-cyclically, whereas the business cycle is unrelated to female mortality. However, Tapia Granados (2003) in his analysis of the relationship between economic fluctuations and aggregate mortality in the 19th and 20th century does find a pro-cyclical movement in Swedish mortality rates. Tapia Granados (2002) finds similar results pooling data from the 50 Spanish provinces over the period 1980–1997.
The objective of this paper is to test the relationship between aggregate mortality and economic fluctuations for a different nation-state, namely Germany over the period 1980–2000. For Germany, no comprehensive individual data are available. Germany is an obvious candidate for an analysis of aggregate data, however, since similar to the US it is a federal nation-state, which allows panel data analysis, and as one nation-state, Germany is also more homogeneous in terms of population characteristics than a pool of nation-states, which helps to reduce omitted variable bias. We will show that economic recessions lower aggregate and some specific mortality rates, thus providing more support for Tapia Granados’ (2002, p. 41) contention that ‘the pro-cyclical character of mortality fluctuations is beginning to be a proven fact’.
Like Ruhm (2000) we also use fixed-effects estimation. At the same time, we improve on his analysis in a number of ways. First, we use standard errors, which are robust towards arbitrary heteroscedasticity and autocorrelation. Second, in sensitivity analysis we additionally make standard errors robust towards clustering such that observations are merely assumed to be independent across states, but not necessarily within states. Third, we use a dynamic model, which does not require the researcher to specify the number of time lags included in the model, a decision, which is always and by necessity somewhat arbitrary. Instead, we include the lagged dependent variable and correct for the correlation of the regressor with the error term with the help of Arellano and Bond's (1991) generalized method of moments (GMM) estimator. Fourth, we look at gender-specific mortality rates to test for gender differences.
Section snippets
The impact of economic fluctuations on health
There are many theories of the impact of economic fluctuations on health conditions. However, one can perhaps distinguish between two main perspectives. As the main objective of this paper is an empirical analysis, we will merely sketch the arguments here. The reader is referred to Brenner and Mooney (1983), Watkins (1985), and Ruhm (2000) and the many references cited therein for a more extensive discussion. One perspective focuses on the social and psychological aspects of the hardship caused
The dependent variables
Like Ruhm (2000) we include total mortality for all age groups taken together as well as total mortality for three specific age groups (20–44, 45–64, older than 65) and ten specific mortality causes: malignant neoplasms (ICD 140–208), cardiovascular diseases (ICD 390–459), pneumonia and influenza (ICD 480–487), chronic liver diseases (ICD 571), motor vehicle accidents (E810–825), intentional self-damage (suicide) (E950–959), murder and manslaughter (homicide) (E960–969), other external effects
Results
We start with the static model with contemporaneous effects only. Table 1 presents results for aggregate mortality rates for all age groups, specific age groups as well as for both sexes. Note that for simplicity, only the coefficients of the unemployment rate is shown, whereas the coefficients of the other control variables and the year-specific time dummies are suppressed. We will come back to the other control variables further below. We can see that the unemployment rate is negatively and
Sensitivity analysis
We will now test the sensitivity of our results with regards to model specification. For reasons of space, we will merely describe but not report detailed results, which are available from the author upon request, however. To start with, we include the Gini coefficient as a further control variable. As mentioned above, this variable is only available from 1985 onwards. Our results are hardly affected. The coefficient of the income inequality variable itself often has a positive sign, but it is
Discussion
We have seen that economic downturns are associated with higher mortality in German states similar to the pattern observed by Ruhm (2000) for the states of the United States of America. The effect is statistically significant for aggregate mortality rates for all age groups taken together, all specific age groups and for both males and females. This is by and large true independent of whether the static or dynamic model is estimated and independent of whether the indicator for economic
Conclusion
All in all, we have found confirmation for Ruhm's (2000) general result in our analysis of German states: recessions tend to lower mortality rates. We find consistent and robust evidence that recessions lower aggregate mortality rates for all age groups taken together as well as all specific age groups. We find less consistent evidence for specific mortality causes than Ruhm (2000) did. For both sexes taken together, we find the predicted effect for cardiovascular diseases, pneumonia and
Acknowledgements
Helpful comments from Christopher Ruhm, José A. Tapia Granados, Stephen Birch and two anonymous referees are gratefully acknowledged. I would like to thank Peter Krause from the German Institute for Economic Research as well as many individuals in the German Federal Statistical Office and as in some state Statistical Offices for providing the data required for the analysis.
References (41)
Relation of economic change to Swedish health and social well-being, 1950–1980
Social Science & Medicine
(1987)- 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 unemploymentA critique of Brenner's time series analysis
The Lancet
(1981) - et al.
Do high levels of unemployment influence the health of those who are not unemployed? A gendered comparison of young men and women during boom and recession
Social Science & Medicine
(2001) - et al.
Does drinking really decrease in bad times?
Journal of Health Economics
(2002) Time series analysis of the relationship between unemployment and mortalityA survey of econometric critiques and replications of Brenner's studies
Social Science & Medicine
(1985)- et al.
Estimation of dynamic models with error components
Journal of the American Statistical Association
(1981) - et al.
Some tests of specification for panel dataMonte Carlo evidence and an application to employment equations
Review of Economic Studies
(1991)
Econometric analysis of panel data
Mental illness and the economy
Trends in alcohol consumption and associated illnessesSome effects of economic changes
The American Journal of Public Health
Political economy and health
Mental health and wellbeing and unemployment
Australian Economic Review
Unemployment and mortality in post-war Scotland
Journal of Health Economics
Are recessions good for your health? Further evidence
Deaths rise in good economic timesevidence from the OECD
Cited by (331)
Does unemployment in sub-Saharan Africa have asymmetric effects on health? A panel quantile approach
2023, Journal of Economic AsymmetriesHow do road infrastructure investments affect Powered Two-Wheelers crash risk?
2023, Transport Policy