Does sickness absence increase the risk of unemployment?

Abstract

This paper examines whether a worker's sickness absence behaviour influences the risk of becoming unemployed. Swedish panel data are used to estimate the relationship between the incidence and the duration of sick leave and subsequent unemployment. The results indicate that an increase in the number of sick leaves as well as an increase in the duration of sick spells are associated with a higher risk of unemployment. An implication of the results is that less absence-prone workers are more likely to remain employed in a recession, which may partly explain the pro-cyclical pattern of aggregate Swedish sickness absence rates.

Introduction

The purpose of this paper is to test if sickness absence is associated with an increased risk of future unemployment. Swedish panel data are used to estimate the effect on the risk of unemployment of previous sick-leave behaviour. The transition into unemployment is studied from January 1st 1992 to December 31st 1998 and the absence record is observed from January 1st 1989 to February 28th 1991.

The everyday choice for the worker of whether to be present at work, will be affected by present and future repercussions of the choice. For the majority of workers in some European countries, the compensation level for sickness absence has been close to the normal wage during the last decades (Nyman et al., 2002). A smaller income loss during a sick-leave spell may imply a low incentive to attend work. However, some countries with relatively generous compensation schemes have a relatively low absence rate¹; Germany is a case in point. This suggests that the everyday attendance choice is also affected by other factors.

The utilisation of sick leave may yield a productivity signalling effect; a less absent worker is more attractive from the employer's perspective, especially if the employer directly pays part of the costs associated with a sick spell. A worker with an extensive absence record may be tagged as expensive and less productive which, in turn, may lead to dismissal (Yaniv, 1991). The penalty may exist, even though the individual is obliged to use a sick spell to recover from illness and restore productivity, since the employer may have problems monitoring the health among employees. If the disutility of being sick is disregarded, the cost for the individual can be divided into a direct cost of the sick leave, the possible income loss, and an indirect cost which can be seen as the possible effect on future labour market outcomes. In the analysis of the economic incentives of calling in sick, it should be kept in mind that some sort of non-economic work moral effect is likely.

During the last decade, economists have given some attention to sickness absence, especially in Sweden where the absence rate in 1990 was one of the highest among the industrialized countries and thereafter showed a rapid decline followed by an increase at the end of the 1990s. There were many changes in the Swedish sick-leave insurance system during the same time period, which resulted in large differences in the sick-leave compensation rate over time. Johansson and Palme, 1996, Johansson and Palme, 2005 have shown that economic incentives seem to be of importance for the incidence of sick leave. They also find that the variation in the sick-leave rate cannot solely be explained by higher costs of absence; the findings show a significant inverse relationship between the unemployment rate and the incidence of sickness absence. These findings suggest a pro-cyclical pattern of the sick-leave rate, which has later also been shown by, e.g. Arai and Skogman-Thoursie (2005), Askildsen et al., 2000, Askildsen et al., 2002 and Henrekson and Persson (2004).

Leigh (1985) discusses the theoretical pro-cyclical relationship between the sick-leave rate and the unemployment rate. He claims that besides an eventual disciplining effect of the unemployment rate on the sick-leave rate, a labour force composition effect is likely if absence-prone workers have a higher incidence of unemployment in an economic downturn. A fear of being laid off that reduces the number of annual absence days is credible, if a positive relationship is found between sick-leave utilisation and the risk of becoming unemployed.

The main result of the estimated models in this essay is that both the incidence and the duration of a sick leave are associated with an increased risk of unemployment. Some evidence shows long sick-leave spells to be associated with a higher unemployment risk for women than for men. Older male workers seem to have a higher risk of unemployment than young male workers, given equal work experience, which may partially be explained by health differences over age. Comparing the effects of incidence and duration of sickness absence for three different absence length categories indicates that non-short-term sickness absence yields a penalty (increased risk of unemployment) that to a high extent is associated with the incidence rather than the duration.

The next section discusses the relationship between sickness absence and unemployment. Section 3 presents the data, how the sample was selected and defines some key variables. Section 4 discusses sample issues, describes the estimation method and the model to be estimated. Section 5 presents the results and finally, Section 6 concludes.