Work Absence in Europe: An Update Lusine Lusinyan Leo Bonato International Monetary Fund Rome, June, Disclaimer: The views expressed herein are those of the authors and should not be attributed to the IMF, its Executive Board, or its management.
Motivation Based on Lusinyan-Bonato (7) Concern about trends in labor supply in Europe Increase in participation Decrease in average hours worked Looking ahead Negative impact of ageing Growth potential will decline under reasonable assumptions on productivity Shrinking tax bases will undermine generosity of welfare systems
Participation and hours worked,,, IE PT, Hours worked,9,8,7,6, ES IT LU BE NL FR DE AT SE FI CH UK NO DK IS,9,8,7,6,, 98 989 99 999 8,,, 6 6 7 7 8 8 9 Labor f orce participation (percent)
Policies to increase labor supply Increase labor force participation Address working time utilization Contractual/usual hours of work Actual hours worked Positive difference: work absence
Plan of the presentation I. Overview of work/sickness absence II. Some facts about sickness absence and its determinants III. Econometric evidence IV. Conclusions
I. Overview Declining hours worked must reflect, to some extent, preferences If anything, taxes have declined and welfare systems have become less generous (Blanchard, ) The gap with the US shows that labor supply may be affected by tax distortions (Prescott, ) or labor market institutions (Alesina and others, ) Social insurance systems play an important role: they are costly, generally not well targeted, and unlikely to be optimal (Treble, ) 6
What determines work/sickness absence? Literature mostly focuses on labor supply: Brown and Sessions (996), Allen (98), Leigh (98), Dunn and Youngblood (986), Drago and Wooden (99), Barmby et al. () Labor force characteristics (age, gender, health) Non-labor income Unemployment/cyclicality Disciplining effect of unemployment More sick-prone people in expansion (Arai and Skogman Thoursie, ) Work arrangements Sickness insurance provisions Ample evidence of moral hazard: Henrekson and Persson (); Johansson and Palme (996, ); Skogman Thoursie (); Andren () Impact of employment protection Ichino and Riphahn () 7
II. Some facts about sickness absence and its determinants 8
Sickness absence varies across countries in Europe (Full-time employees absent due to sickness as a percentage of total employed, e 8) 8) SE NO NL FI UK BE FR AVE. IS DK AT ES CH DE PT Total Public administration IT IE 6 7 9
Sickness absence (Full-time employees absent due to sickness as a percentage of total employed, e 98 8) Belgium 98 986 989 99 99 998 7 Denmark 98 986 989 99 99 998 7 Germany 98 986 989 99 99 998 7 Spain 98 986 989 99 99 998 7 France 98 986 989 99 99 998 7 Ireland 98 986 989 99 99 998 7 Italy 98 986 989 99 99 998 7 Netherlands 8 6 98 986 989 99 99 998 7 Portugal 98 986 989 99 99 998 7 United Kingdom 8 6 98 986 989 99 99 998 7
Sickness absence (cont.) (Full-time employees absent due to sickness as a percentage of total employed, e 99 8) 8 6 8 6 Sweden 99 997 999 7 99 997 999 7 Iceland Austria 99 997 999 7 8 6 8 6 Norway 99 997 999 7 Finland 99 997 999 7 Switzerland 99 997 999 7
Sickness absence in Italy (Full-time employees absent due to sickness as a percentage of total employed, e 99 8) 8) All Sectors 99 99 996 998 6 8 Public Administration 99 99 996 998 6 8
Some support for pro-cyclicality of sickness absence (Vertical axis: sickness absence rate; horizontal axis: unemployment ment gap) 9 Sweden 7 6 7 8 999 998 99 997 996-8 -6 - - 6 8 6. Italy 7 98 98 8 98 99. 99 986 99. 987 989 988 99 997 99 99 996 999 998. - - -
Sickness insurance systems have become less generous % DE. NO. NO. SE.7 FI SE DK IE NL BE IT CH AT FR 9. 88.6 86.9 8. 79.8 76.9 7.6 7. 7.7. Initial level of sickness benefit net replacement rate, 98 99 CH DE AT IE DK BE NL FI FR IT.. 9.9 9.9 9. 8.9 8.8 8. 8. 6. Initial level of sickness insurance system generosity, 98 99 UK 7. UK.9 - - - - - less generous more generous less generous more generous Change in average sickness benefit net replacement rate between 99 and 98 99, percentage points Change in sickness insurance system generosity between 99 and 98 99
Employers sharing the costs of public insurance scheme can create an incentive to reduce sickness absence Persons absent due to sickness, % of total employed, 8 7 6 SE NO FI UK BE DK IS ES CH DE PT IE AT NL FR IT 7 6 6 8 Average (days)*(benefit replacement rate) of sick leave paid by employers, 8
Cost to the public finances (Percent of GDP) 6
Working time arrangements may affect the choice of work effort IS UK AT CH ES AVE. PT DE SE LU IE DK FI BE IT NL FR NO NL CH UK NO DE IE BE IS DK AVE. AT SE IT FR LU FI PT ES 8 7 Usual hours worked per week, 8 Part-time employment, percent of total employment, 8 7
Absence increases with labor force participation and share of female and elderly but no clear relation with health 7 (Vertical axis: sickness absence rate; 8) 8) 7 6 NL SE NO 6 NL SE NO IT BE FR ES IE FI UK AT DE PT DK CH IS IT BE ES IE UK FR AT DE PT FI CH DK IS 6 6 7 7 8 8 9 Labor force participation rate ( 6), percent 7 6 6 7 7 8 8 9 labor force participation rate ( 6), percent 7 6 NL NO SE 6 NL NO SE BE FR AT ES IT IE FI UK IS DE PT DK CH DK PT FI UK BE AT DE IE FR ES IT IS CH 77 78 79 8 8 8 Share of labor force 6 in labor force 6, percent Life expectancy at birth, years 8
Some observations for Italy Sickness absence in Italy has been among the lowest in Europe Absence in public sector is on average over percent higher than in total employment the the difference is among the largest in Europe Sickness absence rate doubled in 6 6 and declined somewhat afterwards Structural break in LFS series Possible cyclical impact Several factors could explain low sickness absence rate Lower labor force participation, especially for female Lower share of employees over Better health status Lower usual hours worked Higher cost to employers (industry structure) Increasing share of temporary contracts with little employment protection 9
III. Econometric evidence Only a few cross-country country comparative studies on work absence: Drago and Wooden (99), using micro database from plants in U.S., Canada, New Zealand, and Australia, find higher absence rates among women, full-time, low-wage, wage, and long- tenure employees. Also, absence is positively correlated with shiftwork, generosity of sick leave entitlements, and better labor market options Barmby et al. () LFS sickness absence data for eight European countries and Canada similar conclusions, also robust positive relationship with usual hours worked Bergendorff et al. () aggregate long-term sickness absence data from LFS for eight European countries similar similar results, and temporary workers less absent, but no clear relationship with health status. This paper: broader country coverage static and dynamic panel data models novel datasets on sickness insurance provisions and costs to employers, also controlling for labor market regulations
The model Combines labor supply and demand Augments the determinants of the labor- leisure choice by a number of institutional characteristics: generosity of paid leave provisions employment protection Looks at differences in the impact of publicly vs. privately financed insurance on absence
Main hypotheses to test Replacement rate (+) Contractual hours (+) Employment protection (+) Unemployment rate ( )( impact depends on employment protection: stronger employment protection smaller impact Unemployment benefit (+) impact depends on employment protection: stronger employment protection smaller impact Privately financed insurance impact depends on model assumptions (elasticity of probability of keeping the job with respect to absence) Utility function characteristics age, health
The data Total sickness absence (short-term term and long-term) Eurostat LFS (NewCronos) Employment and demographic data ILO s s KILM, OECD Health Data Welfare programs Scruggs () Comparative Welfare Entitlements Data Set Labor market institutions WEO (based on Nickell and Nunziata, ), Scruggs () Sickness insurance cost to employers U.S. SSA, Social Security Programs Throughout the World Unbalanced panel of N=8 countries, (98-, 99- )
Empirical strategy Static and dynamic panel data models, controlling for endogeneity and fixed effects. In a general setup: a k i, t = ai, t jβ, j + Xi, tβ + Wi, tβ + ηi + εi, t, j= i =,... N; t =,...,Ti a i, t X i,t W i,t η i ε i, t absence rate for country-gender pair i at time t vector of exogenous covariates vector of predetermined and endogenous covariates (similar to lagged dependent variable) both gender and country specific only country specific unobserved fixed effect disturbance term
Estimation and robustness checks Static panel data models (FE, RE, pooled OLS) Appropriateness of RE and test of serial correlation Adjust for within-group correlation in disturbances Endogeneity tests Robustness to country sample Robustness to the type of absence Dynamic panel data (DPD) models (AB GMM, AH IV, BB) Different sets of instruments Robust s.e. Robustness to country sample Robustness to the type of absence Test for non-stationarity (panel UR tests)
Summary of estimation results Determinants of sickness absence Sign Notes Lagged absence (+) highly significant LFPR (+) positive lagged LFPR Age structure (+) expected sign but insignificant Life expectancy (-) positive lagged impact Part-time time employment (PTE) (-) significant Usual hours (+) highly significant Usual hours*pte (-) but (+) if only PTE included Unemployment gap (-) Sweden Sickness benefit (+) highly significant, Sweden Unemployment benefit (-) or insignificant Employer sick pay (-) significant Union density (+) highly significant Union density*ue gap (+) expected sign but insignificant Union density*employer pay (+) significant but small 6
IV. Conclusions High sickness absence explained by high participation but not only Strong impact of usual hours Reducing contractual hours may not be inconsistent with policies to reduce absence Encouraging flexible work arrangements may be a better policy option Sickness insurance incentives and employment protection matter Interaction between social security and labor market institutions is important 7
Further research Private vs. government financed compensation systems Role of labor market institutions Data labor market institutions, absence data, private vs. public employment 8