Elsevier

Journal of Health Economics

Volume 25, Issue 6, November 2006, Pages 1069-1093
Journal of Health Economics

Are recessions good for workplace safety?

https://doi.org/10.1016/j.jhealeco.2006.03.002Get rights and content

Abstract

Empirical evidence shows that in recessions the rate of workplace accidents goes down. This paper presents a theory and an empirical investigation to explain this phenomenon. The theory is based on the idea that reporting an accident affects the reputation of a worker and raises the probability that he is fired. If the unemployment rate is high, a worker faces a big loss when fired and fewer workplace accidents are reported. The empirical investigation concerns workplace accidents in 16 OECD countries. We conclude that cyclical fluctuations in workplace accident rates have to do with reporting behavior of workers and not with changes in workplace safety.

Introduction

Workplace accident rates are pro-cyclical, which could indicate that workplace safety increases in recessions. If so, this would be in line with evidence presented by Ruhm (2000) indicating that booms are unhealthy. Ruhm finds a strong relationship between macroeconomic conditions and mortality, which he attributes to hazardous working conditions, the physical exertion of employment, and job-related stress when job hours are extended during short-lasting economic expansions.3 This explanation for the pro-cyclicality of workplace accident rates has to do with working conditions and concerns workplace safety. In booms the effort level of workers is higher. When employers require high effort levels from their workers, these workers become less careful. Hence the rate of workplace accidents increases. When workers exert less effort, they can take more care and are therefore involved in less accidents. If effort is pro-cyclical, so is the rate of workplace accidents. Consequently, the workplace accident rate is negatively correlated with the unemployment rate. Also in booms more workers are hired than in slumps. If newly hired workers are less experienced and are therefore more likely to be involved in workplace accidents there is an additional explanation a higher accident rate in a boom.

The change in workplace safety is not the only explanation for the cyclicality in workplace accident rates. An alternative explanation is related to the well-known pro-cyclical variation in workers’ absenteeism. If unemployment is high, workers fear to loose their job, so conditional on a specific state of their health they are less inclined to stay away from work. If unemployment is low, employers will be more reluctant to fire a worker even in case of frequent absenteeism. So, absence-prone workers are more likely to be dismissed in an economic downturn (Leigh, 1985).

In terms of workplace accidents the idea is that workers are reluctant to report these accidents because they fear (correctly or incorrectly) that employers will hold this against them.4 In particular, we assume that workers think that if the firm has to lay off some workers at some future date (say, because of a labor saving innovation or because demand for the industry's products goes down) it will tend to fire workers that have reported accidents in the past. When unemployment is low, being fired does not involve a big loss because it is relatively easy to find a new job. Hence workers are more likely to report an accident when unemployment is low. In contrast, a high unemployment rate makes it harder for a fired worker to find a new job and hence a worker is less likely to report an accident. So we expect that the reporting rate of accidents is low if the unemployment rate is high.

For concreteness, consider the following example. A worker in a factory caries a heavy object. He stumbles and almost drops the object, which causes him to strain his back. If the back pain is so severe that he cannot continue to work, he has no choice but to report the accident. He can then go home to recover for a couple of days. If the worker can continue work but the back pain is rather uncomfortable he decides whether to inform his employer or not. Our conjecture is that he is more likely to report the accident if the unemployment rate is low. If his employer would use the accident against him, in the sense that it becomes more likely that he will be fired if the firm needs to shed labor, at least the consequences are not too severe. Similarly, a secretary who feels her fingers tingling faces the choice of stopping to work and go home or continuing and raise the risk of a mouse arm. Our conjecture, again, is that she will continue to work if the unemployment rate is high, until it becomes clear that she does have a problem with her arm and needs rest.

This paper focuses on the question whether a high workplace accident rate in a boom is a true workplace safety phenomenon related to stress or inexperienced workers and thus an indication of workers’ health being harmed in prosperous times or whether alternatively a high workplace accident rate in a boom is a spurious phenomenon caused by changes in workers reporting behavior. The two explanations for the cyclicality of workplace accidents are tested empirically. Part of the evidence presented uses the distinction between non-fatal and fatal accidents (see the appendix for more details). The workplace safety explanation predicts that the unemployment rate has a negative effect on workplace accidents. Furthermore, it predicts that working hours have a positive effect on workplace accidents because working longer increases the probability of an accident. This can happen for two reasons. First, working more hours may make workers tired and fatigue raises the probability of injury per hour (or per task). Second, even if the injury rate per hour (or task) is unchanged, working more hours increases the workers’ exposure to this risk of accidents at work. This raises the accident rate (per worker). Also, the change in the employment rate is positively related to the workplace accident rate since an increase in employment coincides with a lot of new hirings while a decrease in employment is related to few new hirings. New hirings tend to be less experienced and hence more prone to have or cause an accident. If workplace safety is the driving force not only the non-fatal accident rate but also the fatal accident rate will exhibit cyclicality. The ‘reporting’ explanation predicts that the unemployment rate is negatively related to the non-fatal workplace accident rate but not to the fatal accident rate because fatal accidents are always reported (if there is a reporting decision at all in this case). Further, as unemployment benefits go up one should expect a smaller effect of the unemployment rate on the non-fatal accident rate.

In our empirical analysis it appears that the unemployment rate is negatively related to the workplace accident rate while working hours and the change in the employment rate have no effect on the workplace accident rate. Furthermore, cyclicality is found only for non-fatal workplace accidents. From this evidence and additional sensitivity analyses we conclude that the pro-cyclical behavior of workplace accidents is most likely related to reporting rates and not to actual fluctuations in accidents. From a policy perspective this result implies that when in cyclical upturns the workplace accidents rate goes up there is no urgent need to start worrying about workplace safety nor to start a work safety campaign. Instead, in times of high unemployment safety auditors should encourage people to report (minor) accidents.

The paper is set up as follows. In Section 2 we set the stage by giving an overview of previous studies on absenteeism and workplace accidents. In Section 3 we present a theoretical model that explains the cyclicality in reported workplace accidents. Section 4 describes the data on workplace accidents from 16 OECD countries, discusses the statistical model and presents estimation results. Section 5 concludes.

Section snippets

Previous studies

There are several studies that investigate the cyclicality of workplace absenteeism. According to Barmby et al. (1994) the effect of absence behavior on the probability of being fired may act as a worker discipline device. Brown and Sessions (1996) distinguishes between absence that occurs for valid (i.e. ‘sickness’) and invalid (i.e. ‘shirking’) reasons. According to them little attention has been paid in the economic literature to either the causes or the effects of absenteeism. They analyze

Workplace accidents-theory

This section introduces a model to formalize the relation between cyclical variations in unemployment and workplace accidents. The idea is that reporting accidents affects a worker's reputation and raises the probability that he will be fired. This is especially disadvantageous for the worker if the value of being unemployed is low. Hence in such circumstances one expects fewer accidents to be reported. The value of being unemployed varies with the business cycle. In the empirical analysis we

Workplace accidents across the OECD

According to the European Statistics on Accidents at Work an accident at work is defined as a “discrete occurrence in the course of work, which leads to physical or mental harm”.9 Workplace

Conclusions

In recessions the rate of workplace accidents is lower than it is in booms. We distinguish between two possible explanations for this phenomenon: workplace safety and reporting behavior. The workplace safety explanation predicts that the unemployment rate is negatively related to workplace accidents because effort is negatively related to unemployment and high effort makes accidents more likely. Furthermore, it predicts that working hours are positively related to workplace accidents because

Acknowledgements

The current paper is the result of a substantial revision of Boone and van Ours (2002). The authors are grateful to Joost Lebens for excellent research assistance and to Maria Guadalupe for comments on a previous version of the paper.

References (24)

  • M. Arai et al.

    Incentives and selection in cyclical absenteeism

    Labour Economics

    (2005)
  • P. Johansson et al.

    Do economic incentives affect work absence? Empirical evidence using Swedish micro data

    Journal of Public Economics

    (1996)
  • J.P. Leigh

    The effects of unemployment and the business cycle on absenteeism

    Journal of Economics and Business

    (1985)
  • Askildsen, J.E., Bratberg, E., Nilsen, O.A., 2002. Unemployment, labor force composition and sickness absence; a panel...
  • T.A. Barmby et al.

    Absenteeism, efficiency wages and shirking

    Scandinavian Journal of Economics

    (1994)
  • J. Biddle et al.

    What percentage of workers with work-related illness receive workers’ compensation benefits?

    Journal of Occupational and Environmental Medicine

    (1998)
  • Boone, J., van Ours, J.C., 2002. Cyclical fluctuations in workplace accidents, CEPR DP No....
  • Boone, J., van Ours, J.C., 2004. Effective active labor market policies, CEPR DP No....
  • S. Brown et al.

    The economics of absence: theory and evidence

    Journal of Economic Surveys

    (1996)
  • European Agency for Safety and Health at Work, 2001. Fact 19,...
  • D. Fairris

    Institutional change in shopfloor governance and the trajectory of postwar injury rates in U.S. manufacturing, 1946–1970

    Industrial and Labor Relations Review

    (1998)
  • J. Glazner et al.

    Construction injury rates may exceed national estimates: evidence from the construction of Denver International Airport

    American Journal of Industrial Medicine

    (1998)
  • Cited by (0)

    Financial support from the Dutch National Science Foundation (grant-numbers 016.025.024 and 453.03.606) is gratefully acknowledged.

    1

    Department of Economics, University of Tilburg, Tilec, Netspar, CentER, ENCORE, UvA, IZA and CEPR.

    2

    Department of Economics, Tilburg University, CentER, IZA and CEPR.

    View full text