What if they take it all? Heterogeneous impact of sickness absence reform

Abstract

I investigate the effect of sickness absence reform in the Czech Republic, which reduced benefits paid during the first three days of sickness absence to zero. I find a substantial decrease in the incidence of sickness absence, which is about 15 percent of the pre‐reform mean. I find that workers in occupations with high flexibility and fewer routine tasks are more likely to reduce their sickness absences.

Sick Pay and Absence from Work: Evidence from Flu Exposure

“…Thus, it is important to understand and measure the effects of sick-pay programs, as they are associated with significant financial and health costs. 2 The literature on workplace absences has so far focused on how individual characteristics affect the probability of being absent from work (e.g., Barmby, 2002;Scoppa, 2010;De Paola, 2010;Amuri, 2011), and analyzed the effects of changes in sick-pay programs (e.g., Ziebarth and Karlsson, 2010;Pettersson-Lidbom and Thoursie, 2013;De Paola et al, 2014;Pertold, 2019). Still, little is known about the relationship between actual sickness rates and absences from work.…”

“…The omission of variables that control for local sickness levels does not necessarily change the explanatory power of the estimated models. However, it may change the size of other estimated coefficients of interest, e.g., the controls for periods after the reforms, which are widely interpreted as the effects of policy reforms in a before-after comparison framework (e.g., De Paola et al, 2014;Pertold, 2019). 4 We first estimate the overall effects of policy reforms using a 'before-after' comparison.…”

“…Applying the difference-in-differences estimation strategy, Csillag (2017) finds that the legislative change caused a small reduction in the incidence of absence form the workplace and a significant decrease in the number of days absent among low-earning employees. Pertold (2019) studies the effects of policy reforms in the Czech sick-pay system during 2008-2009, i.e., the same changes that are addressed in this paper. Working with the Czech the SES data, Pertold (2019) uses the 'before and after' comparison to find that these legislative changes significantly reduced the number of total days absent.…”

“…Working with the Czech the SES data, Pertold (2019) uses the 'before and after' comparison to find that these legislative changes significantly reduced the number of total days absent. Moreover, Pertold (2019) claims that the effects of the policy changes are extremely heterogenous across industries and occupations, and most affect employees working in manufacturing, hotels, and restaurants. Furthermore, Pertold (2019) shows that employees with more routine tasks and lower job-flexibility are much more likely to reduce their hours of absence.…”

Sick pay and absence from work: Evidence from flu exposure

“…The first 3 days of sickness are termed a 'quarantine period' or 'waiting period'.5 This is important as the omission of variables that control for sickness levels may change the size of other estimated coefficients of interest. Prominent examples include the controls for periods after the reforms, which are widely interpreted as the effects of policy reforms in a before-after comparison framework (e.g., DePaola et al, 2014;Pertold, 2019). For a demonstration see TableA8.6 We cannot apply a standard difference-in-differences estimator, because the policy changes affected all employees in the Czech economy.7 In most of the specifications estimated, we use a measure of sickness where we count the number of weeks with influenza epidemic status among children, assuming that children can infect adults but not vice versa.8 In this setup, the policy reforms did not change the need for sickness insurance of workers who were shirking or who suffered from other illnesses than influenza; hence, we strictly focus on changes in absence hours instigated by the 3-day waiting period for sickness benefits.9 By "overall effect" we mean the effect of disease exposure in periods after 2008, that is, the combination of β and γ coefficients from Equation 1 presented in Section 4.10 For many other diseases we find similar substitution effects as in our baseline results, though the results are economically negligible.11 The size of the effects is smaller for sickness-related absences, but of similar magnitude for paid and unpaid leave.One additional week of influenza outbreak in the periods before 2008 caused an increase in sickness-related absences of 1 day per quarter and a decrease in paid and unpaid leave of 1.8 and 1.6 days, respectively.…”