Effect of job loss due to plant closure on mortality and hospitalization

https://doi.org/10.1016/j.jhealeco.2012.03.001Get rights and content

Abstract

We investigate whether job loss due to plant closure causes an increased risk of (cause-specific) mortality and hospitalization for male workers having strong labour market attachment. We use administrative data: a panel of all persons in Denmark in the period 1980–2006, containing records on health and work status, and a link from workers to plants. We use propensity score weighting combined with non-parametric duration analysis. We find that job loss increases the risk of overall mortality and mortality caused by circulatory disease; of suicide and suicide attempts; and of death and hospitalization due to traffic accidents, alcohol-related disease, and mental illness.

Introduction

It is well documented that unemployment is associated with poor health; see, for instance, the survey of Kasl and Jones (2000). The determinants of this correlation are far from fully understood, however. Many studies have considered whether there is a causal effect of unemployment on health by using the downsizing or closure of firms or plants as a quasi-experiment; see e.g. Morris and Cook (1991) for a review. In recent years this approach has been applied using large administrative datasets; see Browning et al. (2006), Sullivan and von Wachter (2009), and Eliason and Storrie, 2009a, Eliason and Storrie, 2009b, Eliason and Storrie, 2010. The present paper also follows this approach.

We focus on the effect of displacement due to plant closure on the risk of overall mortality and cause-specific mortality and hospitalization, and we investigate its effects on a wide range of diseases. It is important to study a wide array of diagnoses and causes of death in order to better understand possible negative health effects of job loss. For instance, it is interesting to investigate whether – and to what extent – a possible increase in all-cause mortality is related to circulatory disease, alcohol-related disease or suicide. We take an eclectic view since there is not a single well-established theory on which specific negative health outcomes to expect as a consequence of job loss. It is also interesting to consider categories of serious diagnoses which are unlikely to be much affected by job loss, for instance cancer.

The population studied is full-time male workers with a strong labour market attachment.1 We use administrative data: a sample of all persons in Denmark in the period 1980–2006. We identify workers who lost their job due to closure of plants in the private sector in the period 1986–2002, and we identify a control group of workers. The data contain very full records on demographics, health and work status for each person, and include a link from every working person to a plant. Health outcomes are based on causes of death and diagnoses from somatic and psychiatric hospital departments.

Five new contributions emerge from this study. First, we use a much larger dataset than any previous plant closure study, which is essential when considering rare outcomes such as death or hospitalization due to specific causes. For instance, point estimates in Eliason and Storrie (2009a) of the effect of job loss on the risk of death of male workers from circulatory disease and cancer in the first 4 years after plant closure are large (about 20% and 40%, respectively), but they are imprecise and not significantly different from zero, even though circulatory disease and cancer are the two most important causes of death. As we have a much larger dataset, our estimates are more precise. Second, we do not impose parametric restrictions when analysing the duration to death (or hospitalization), and we are able to analyse the time pattern of effects more precisely than is done in Eliason and Storrie, 2009a, Eliason and Storrie, 2009b, Eliason and Storrie, 2010.2 This is important since a priori we have no hypothesis of a specific time pattern, and there is no reason to expect that the time pattern of effects should be the same for the very different types of diagnoses which we are analysing. Third, we investigate the effects of job loss on mental disorders as a (secondary) cause of death. Fourth, among studies using large administrative datasets this study is the first to analyse the effects of job displacement due to plant closure on hospitalization for many categories of diagnoses, including both fatal and non-fatal events.3 We also study its effects on ‘combined death and hospitalization outcomes’, such as duration to suicide or a suicide attempt. Fifth, we investigate how the effects of job loss depend on local labour market conditions. A priori, the impact of local labour market conditions on the health effects of job loss is uncertain. A high local unemployment rate makes it more difficult to find a new job, and a new job may be less attractive, but at the same time the stigma associated with being unemployed may be less severe. To our knowledge, the impact of local labour market conditions on the health effects of job loss due to plant closure has not been investigated before.4

An earlier paper (Browning et al., 2006), which also used Danish administrative data, estimated the effect of displacement on hospitalization due to stress-related diseases of the circulatory and digestive systems, but found no significant effect. In the present paper we study its effects on cause-specific mortality and hospitalization for many different categories of diagnosis. Other important differences from Browning et al. (2006) are that the present paper focuses exclusively on workers with a strong labour market attachment and on displacements due to plant closure (not just downsizing), and we use a much larger dataset.

Using Swedish data, Eliason and Storrie (2009a) find that in the first 4 years after plant closure males have increased risks of overall mortality and mortality from external causes, including suicide, and from alcohol-related diseases. They find no effects at durations longer than 4 years after displacement. Point estimates indicate rather large increases in risk of death from circulatory disease, and especially from cancer, in the first 4 years after plant closure, but these effects are not statistically significant. Eliason and Storrie (2009b) analyse effects on hospitalization but only for non-fatal events, i.e. their analysis does not include serious events such that patients die in the hospital. Upon restricting the effect to be constant within their follow-up period of 12 years after plant closure, they find no effect on circulatory disease (myocardial infarction and stroke), but significantly increased risk of hospitalization of displaced workers due to alcohol-related diseases, traffic accidents and self-harm. Eliason and Storrie (2010) use the same dataset and methods to study the effects of job loss on hospitalization for mental illness; they find no effects for males. Analysing the effects of job loss due to downsizing on overall mortality in Pennsylvania, Sullivan and von Wachter (2009) find significantly increased risk of death for displaced workers; the effects are largest in the first years after displacement, but even 20 years after displacement the estimated increase in annual death hazards is 10–15%.

In the present analysis we find that job loss increases the risk of overall mortality and death from circulatory disease; suicide and suicide attempts; and death and hospitalization due to traffic accidents, alcohol-related disease and mental illness. We find no effect on mortality or hospitalization due to cancer, and no effect on hospitalization due to circulatory diseases. The risk of overall mortality is 79% higher in the year of displacement, 35% higher 1–4 years after displacement, 17% higher 1–10 years after displacement and 11% higher 1–20 years after displacement. Thus, the effects are largest just after plant closure, but they remain statistically significant even after 20 years, indicating that short-term effects do not ‘just’ represent a speeding-up of deaths that would have happened a few years later anyway. The same pattern applies for deaths from circulatory diseases, but the effects are larger. Job displacement significantly increases the risk of hospitalization and death due to alcohol-related diseases in both the short and long term. The effect on suicide and suicide attempts is very strong in the first 3 years after displacement but is insignificant in the long term, whereas the effects on hospitalization or death due to traffic accidents are smaller in the short term, but significant in the long term. The effect on hospitalization for mental disorders is large in the short term, and remains clearly significant in the long term. There is a very large short-term effect for mental illness as a (secondary) cause of death, but no significant long-term effect. Effects on mortality are larger when the local unemployment rate is high which is consistent with the idea that employability mitigates the impact of job loss. In the concluding section we compare our findings with earlier studies.

In Section 2 we discuss the choice of health outcomes and hypotheses on the effect of job loss on these outcomes. Section 3 considers the econometric methods used: propensity score weighting combined with non-parametric duration analysis. Section 4 describes the data and the identification of displacement and control groups. Sections 5 Estimation procedure and the propensity score, 6 Estimation results present the results of estimation for the propensity score and health outcomes, respectively. Conclusions are stated in Section 7.

Section snippets

Reasons why job displacement may affect health

Involuntary job loss may have effects on health outcomes, both directly and indirectly through negative economic and social consequences of job loss. Job losers may face large declines in earnings in both the short term and the long run (see, e.g., Ruhm, 1991, Jacobson et al., 1993, Kuhn, 2002, Eliason and Storrie, 2006, Hijzen et al., 2010), and this may give rise to stress. Social and psychological consequences of job loss, such as loss of work relationships, of self-esteem, sense of control,

Empirical methods

This paper investigates whether there is a causal effect of undergoing displacement as a result of plant closure on all-cause mortality and on cause-specific mortality, and on hospitalization for different categories of diagnosis. We use propensity score weighting; see Hirano and Imbens (2001) and Wooldridge (2002, ch. 18.3).

Denote displacement status by the dummy variable D, where D = 1 if displaced (treated) and 0 otherwise, and let Y(0) and Y(1) denote potential health outcomes, where 0

Register data

We use Danish administrative register data. In Denmark all residents have a personal number which is used for administrative purposes to record activities such as education, hospitalization, employment status, interactions with the welfare system, income, and residence. This information is collected centrally by Statistics Denmark and the Danish National Board of Health which makes these data available for statistical and research purposes. Our sample comprises all persons in Denmark in the

Estimation procedure and the propensity score

In this section we first discuss the estimation procedure, and then the estimation of the propensity score and balancing properties of the non-weighted and weighted samples.

Estimation results

Most papers in the job displacement literature in economics focus on effects on labour market outcomes such as earnings and unemployment. This is not the focus of the present paper, but since health effects may be related to economic stress, we briefly report a few results. There are indeed significant negative economic consequences of job loss in our sample. Although the majority of displaced workers find a new job immediately, the effect of displacement on unemployment is very significant:

Conclusion

Based on a unique administrative dataset of all persons in Denmark in the period 1980–2006, we have estimated the causal effects of job displacement due to plant closure on mortality and hospitalization outcomes for male workers in private sector firms. The very large dataset makes it possible to estimate effects on rare health outcomes with much more precision than in previous studies.

We find that job loss increases the risk of overall mortality and mortality caused by circulatory disease; of

Acknowledgements

We thank Jeffrey Smith, Nabanita Datta Gupta, Mette Gørtz, Gabriel Pons Rotger, Søren Leth-Petersen, Bo Honore and two anonymous referees, seminar participants at CAM, University of Copenhagen, and University of Aarhus, and participants at the SOLE-EALE Conference in London 2010 and Econometric Society World Congress in Shanghai 2010, for their helpful comments. The Danish Council for Independent Research–Social Sciences is acknowledged with gratitude for its support.

References (40)

  • M. Eliason et al.

    Lasting or latent scars? Swedish evidence on the long-term effects of job displacement

    Journal of Labor Economics

    (2006)
  • M. Eliason et al.

    Does job loss shorten life?

    Journal of Human Resources

    (2009)
  • M. Eliason et al.

    Inpatient psychiatric hospitalization following involuntary job loss

    International Journal of Mental Health

    (2010)
  • European Commission

    Towards Common Principles of Flexicurity: More and Better Jobs Through Flexibility and Security

    (2007)
  • J.P. Godbout et al.

    Stress-induced immune dysregulation: implications for wound healing, infectious disease and cancer

    Journal of Neuroimmune Pharmacology

    (2006)
  • D.S. Hamermesh

    The costs of worker displacement

    Quarterly Journal of Economics

    (1987)
  • A. Hijzen et al.

    The income losses of displaced workers

    Journal of Human Resources

    (2010)
  • K. Hirano et al.

    Estimation of causal effects using propensity score weighting: an application to data on right heart catheterization

    Health Services & Outcomes Research Methodology

    (2001)
  • K. Huttunen et al.

    How destructive is creative destruction? Effects of job loss on job mobility, withdrawal and income

    Journal of the European Economic Association

    (2011)
  • L.S. Jacobson et al.

    Earnings losses of displaced workers

    American Economic Review

    (1993)
  • Cited by (0)

    View full text