Elsevier

Labour Economics

Volume 43, December 2016, Pages 171-185
Labour Economics

The impact of acute health shocks on the labour supply of older workers: Evidence from sixteen European countries

https://doi.org/10.1016/j.labeco.2016.04.002Get rights and content

Highlights

  • Experiencing a first acute health shock doubles the risk of labour market exit.

  • Conditional on remaining in work, men increase hours worked, women do not.

  • Stroke causes the largest LMP response, followed by cancer, and then infarction.

  • Men's response is driven by impairment, women's by preferences and finances.

  • Access to disability benefits drives cross-country heterogeneity in LMP response.

Abstract

We investigate the consequences of experiencing an acute health shock, namely the first onset of myocardial infarction, stroke or cancer, on the labour supply of older workers in Europe. Despite its policy relevance to social security sustainability, the question has not yet been empirically addressed in the European context. We combine data from the English Longitudinal Study of Ageing and the Survey of Health, Ageing and Retirement in Europe and cover sixteen European countries, representative of different institutional settings, in the years spanning from 2002 to 2013. The empirical strategy builds on the availability of an extremely rich set of health and labour market information as well as of panel data. To remove the potential confounding bias, a selection on observables strategy is adopted, while the longitudinal dimension of data allows controlling for time invariant unobservables. Implementation is based on a combination of stratification and propensity score matching methods. Results reveal that experiencing an acute health shock on average doubles the risk of an older worker leaving the labour market, and is accompanied by a deterioration in physical functioning and mental health, as well as by a reduction in perceived life expectancy. Men's labour market response appears driven by the onset of impairment acting as a barrier to work. In the case of women, preferences for leisure and financial constraints seem to play a prominent role. Heterogeneity in behavioural responses across countries – with the largest labour supply reductions observed in the Nordic and Eastern countries, and England – are suggestive of a relevant role played by social security generosity.

Introduction

The influence of health on labour supply behaviour has long been recognised in the economic literature, mostly with respect to older workers approaching retirement age (Currie and Madrian, 1999, Lumsdaine and Mitchell, 1999). Poor health and health deteriorations have been shown to increase the hazard of early labour market exit in a variety of countries and institutional settings (for example, Bound et al., 1999, Disney et al., 2006, Kalwij and Vermeulen, 2008, Zucchelli et al., 2010). Multiple reasons explain why the bulk of existing literature focuses on older workers. First, older workers face a higher morbidity risk due to longer exposure to risk factors and insults that accumulate through the life course (Ben-Shlomo and Kud, 2002). Indeed, wide epidemiological evidence shows that the chance of experiencing chronic and acute health conditions increases dramatically at older ages (Ward and Schiller, 2013, Feigin et al., 2003, International Agency for Research on Cancer, 2012). Second, older workers are at the core of the current social security reform agenda. In the event of health shock, while facing lower incentives to invest in re-training towards less physically demanding jobs, (Charles, 2003, Newmark and Song, 2012), older workers typically benefit from wider labour market exit options. The widespread take-up of early retirement or other paths of exit from the labour market, such as disability benefits, has been indeed well documented (for example by Blundell et al., 2004, Autor and Duggan, 2006, Euwals et al., 2010, Koning and Lindeboom, 2015).

Economic interest in the topic is motivated, first, by a concern that early labour market exit might bear severe and enduring consequences on individuals' financial wellbeing in later life, though reduced contributions histories resulting in lower pension entitlements (Angelini et al., 2009). No less important comes the concern that early exits might threaten the fiscal sustainability of social security systems, already challenged by population ageing. Beyond financial concerns, working is identified in the psychological literature as a positive contributor to personal and social wellbeing (Spelten et al., 2002, Hackett et al., 2012, Vestling et al., 2013), as it typically fosters individuals' self-esteem and sense of purpose while providing a daily routine and opportunities for social interactions. Not surprisingly, in clinical studies return to work is regarded as an indicator for the success of recovery after the onset of major health conditions (Daniel et al., 2009, Trygged et al., 2011).

Institutional settings represent the first levers governments can activate to promote the labour market inclusion of older and disabled workers. Institutions might influence labour supply adjustments to health shocks, not only through social security generosity (Börsch-Supan, 2008, Gruber and Wise, 1999, Gruber and Wise, 2004, García-Gómez, 2011) but also in terms of labour market structure (Cai et al., 2014) and, last but not least, healthcare provision arrangements. For example, the job lock literature developed in the US illustrates how individuals under employment-contingent health insurance might increase their labour supply after an adverse health shock because of the envisaged rise in future healthcare costs (Madrian, 1994, Kapur, 1998, Bradley et al., 2013). The same pattern could headily be expected though in most European countries, where a national health service is in place and out of pocket expenditures play a limited role in the funding of healthcare (WHO, 2015).

The design of potentially effective policy interventions requires appropriate and up to date evidence on individuals' response to health shocks, and on the mechanisms lying behind the observed adjustments. Multiple mechanisms could be in place. First, poor health might result in the onset of physical and mental impairment limiting the ability to work, especially in the case of heavy manual occupations, and in contexts where job accommodations are not available (Hogelund and Holm, 2014). This would call for interventions aimed at supporting workers' reallocation to less strenuous jobs, or mandatory requirements for on-the-job accommodations. Second, irrespective of impairment-related barriers to work, or even in the absence of impairment, health deteriorations might reduce the optimal level of labour supply, as individuals update expectations about their remaining lifespan, or because of increased preferences for leisure, or more impatient time preferences (Becker and Mulligan, 1997). Were these the prevailing mechanism, as opposed to impairment acting as a barrier to work, clearly different types of policy interventions would be required.

While labour market policy targets are becoming increasingly integrated across European countries (Turrini et al., 2014), there is a lack of systematic and up to date evidence on the labour market impact of health deteriorations for older workers in Europe. Several works have investigated the issue in specific national settings (see, among others, Riphahn, 1999 and Shurer, 2014 for Germany; García-Gómez et al., 2010 for the UK; Datta Gupta et al., 2011 for Denmark, compared to the US; García-Gómez et al., 2013 for Netherlands; García-Gómez and López Nicolás, 2006 for Spain). The only work that, to our knowledge, has approached the subject adopting a systematic framework for a set of European countries is García-Gómez (2011).1 She studies the impact of health shocks, measured as drop in self-assessed health and onset of illness, in nine European countries between 1994 and 2001, and stresses the role of institutional differences as explanatory mechanisms. However, the time span considered dates back more than a decade, well before the onset of the Great Recession that has posed further challenges to older workers employability (Eichhorst et al., 2013, Johnson, 2012, Newmark and Button, 2014).

This paper contributes to the literature by providing novel and up to date evidence on older workers' labour supply response to health shocks in a larger set of European countries, over the period spanning from 2002 to 2013. This represented a lively time of social security reforms, prompted also by the crisis that hit the continent since 2008.2 The analysis covers sixteen countries, showcasing different institutional models including the Nordic (Sweden, Denmark, Netherlands); the Anglo-Saxon (England); the Continental (Austria, Germany, France, Belgium, Switzerland); the Mediterranean (Spain, Italy, Greece); and the Eastern European (Czech Republic, Poland, Estonia and Slovenia) one.

We build on the availability of similar longitudinal surveys of people aged 50 and older, namely the English Longitudinal Study of Ageing (ELSA) and the Survey of Ageing, Retirement and Health in Europe (SHARE), illustrated in Section 2. Following a few previous authors (Smith, 1999, Smith, 2005, Coile, 2004, Datta Gupta et al., 2011), our identification approach, described in Section 3, exploits the chance of individuals being hit by an acute health shock, i.e. myocardial infarction, stroke or cancer. The focus on these specific types of health events is motivated by their severity and still somehow unanticipated onset: even in the case individuals might envisage experiencing a similar health shock at some point in life, for example because of awareness about genetic or behavioural risk factors, uncertainty remains, if not on occurrence, on the timing of potential occurrence. Moreover, although typically self-reported in population surveys, the same nature of these health shocks makes their onset less exposed to reporting bias (Baker et al., 2004) than milder or progressive diseases, as well as plausibly less exposed to the chance of justification bias (Benitez-Silva et al., 2004) than measures based on self-assessed health. With respect to previous works, we improve the identification strategy in two respects. First, we focus on the first acute health shock the individual experiences, rather than using any new onset, as the first onset is plausibly more unanticipated than possible subsequent ones. Second, rather than adopting parametric methods, we employ semi-parametric matching techniques to remove bias stemming from observable confounders, as in García-Gómez (2011) and other works cited there, leading to estimates that are less model dependant (Ho et al., 2007). While identification relies on a standard conditional independence assumption, the longitudinal dimension of data allows controlling for time invariant unobservables too.

Estimation results are presented in Section 4. In terms of outcomes, we consider labour market participation (LMP) and the number of hours worked. We also investigate the demographic, socioeconomic and disability gradient in labour supply responses. To gain insight on the mechanisms explaining the observed labour supply adjustments, we then measure the impact of acute health shocks on other outcomes: physical functioning and mental health, indicative of work ability impairment, and perceived life expectancy, indicative of other factors that might alter the optimal labour supply choice, irrespective of shock-induced impairment. Evidence of heterogeneity across different European countries, and an analysis of the driving institutional differences, are then offered. Results are discussed, in the light of previous literature, in Section 5, before conclusions are drawn in Section 6.

Section snippets

Data

The ELSA and SHARE surveys, both modelled after the US Health and Retirement Study, present extensive similarities with respect to survey design, longitudinal dimension, interview target population, and topics coverage (Taylor et al., 2003, Börsch-Supan and Jürges, 2005). Since 2002, the ELSA collects panel data from a representative sample of the English population aged 50 or older, and younger partners. The ELSA sample is drawn from respondents to the Health Survey for England and covers

Design

To overcome the endogeneity of health10 with respect to labour market behaviour (Haan and Myck, 2009), the identification strategy exploits

Labour supply

Given the remarkable gender difference observed in labour market attachment and trajectories at older ages (see e.g. Aaberge et al., 1999), we present results separately for men and women. Table 8 reports the estimated ATT for LMP and hours, conditional on working. The onset of a first acute health shock results in a significant reduction in the LMP of older European workers, similar in size across genders, or slightly higher for women (10 and 13 percentage points for men and women

Discussion

That acute health shocks increase the risk of leaving the labour market does not come as news to health and labour economists. Smith (2005) found major health events to cause a 15 percentage points reduction in LMP for older workers in the US. Coile (2004) found older US workers hit by an acute health shock to decrease participation (by 16 and 9 percentage points for men and women respectively) approximately doubling the baseline exit probability, with a more sizeable adjustment of those

Conclusions

The current European social security reform agenda aims at fostering older and disabled workers' inclusion in the labour market. If effective policy interventions are to be devised, rigorous evidence on how older workers respond to health deteriorations, and on the mechanisms lying behind the observed adjustments, is required. Although the issue of labour supply responses to health deteriorations represents a previously studied subject, so far up to date and cross-country comparable evidence

Acknowledgements

We are grateful to the Editors and two anonymous Referees, to Ana Rute Cardoso, Axel Börsch-Supan, Michael Hurd, Andrew Jones, Stefania Maggi, Elena Meschi, Pedro Mira, Kenneth Nelson, Marianna Noale, Nigel Rice, Thomas Siedler, Peter Simmons, James Smith, Gugliemo Weber, Jefferey Wenger and to participants in the RAND ‘Comparative International Research Based on the HRS Family of Data’ Conference, the International Health Economics Association 11th World Congress, the XIX Health Economics

References (84)

  • M. Halla et al.

    The effect of health on earnings: quasi-experimental evidence from commuting accidents

    Labour Econ.

    (2013)
  • E. Heinesen et al.

    Effects of breast and colorectal cancer on labour market outcomes — average effects and educational gradients

    J. Health Econ.

    (2013)
  • R. Lumsdaine et al.

    New developments in the economic analysis of retirement

  • J. Moran et al.

    Long term employment effects of surviving cancer

    J. Health Econ.

    (2011)
  • T. Taskila-Abrandt et al.

    The impact of education and occupation on the employment status of cancer survivors

    Eur. J. Cancer

    (2004)
  • R. Aaberge et al.

    Labor supply in Italy: an empirical analysis of joint household decisions, with taxes and quantity constraints

    J. Appl. Econ.

    (1999)
  • V. Angelini et al.

    Ageing and unused capacity in Europe: is there an early retirement trap?

    Econ. Policy

    (2009)
  • Applica et al.

    Study of Compilation of Disability Statistical Data from the Administrative Registers of the Member States

    (2007)
  • O. Attanasio et al.

    The Consequences of Financial Hardship (and Recessions) on Income and Welfare

  • D.H. Autor et al.

    The growth in the social security disability rolls: a fiscal crisis unfolding

    J. Econ. Perspect.

    (2006)
  • M. Baker et al.

    What do self-reported, objective, measures of health measure?

    J. Hum. Resour.

    (2004)
  • G. Becker et al.

    The endogenous determination of time preferences

    Q. J. Econ.

    (1997)
  • H. Benitez-Silva et al.

    How large is the bias in self reported disability?

    J. Appl. Econ.

    (2004)
  • Y. Ben-Shlomo et al.

    A life course approach to chronic disease epidemiology: conceptual models, empirical challenges and interdisciplinary perspectives

    Int. J. Epidemiol.

    (2002)
  • M. Blekesaune et al.

    Working conditions and early retirement — a prospective study of retirement behavior

    Res. Aging

    (2005)
  • R. Blundell et al.

    A life-cycle consistent empirical model of family labour supply using cross-section data

    Rev. Econ. Stud.

    (1986)
  • R. Blundell et al.

    Pension Incentives and the Patterns of Retirement in the United Kingdom

  • A. Börsch-Supan

    Changes in Health Status and Work Disability

  • A. Börsch-Supan et al.

    The Survey of Health, Ageing and Retirement in Europe — Methodology

    (2005)
  • L. Cai et al.

    The effects of health and health shocks on hours worked

    Health Econ.

    (2014)
  • K. Charles

    The longitudinal structure of earnings losses among work-limited disabled workers

    J. Hum. Resour.

    (2003)
  • C. Coile

    Health Shocks and Couples' Labour Supply Decisions

  • K. Daniel et al.

    What are the social consequences of stroke for working-aged adults? A systematic review

    Stroke

    (2009)
  • A.M. Dano

    Road injuries and long-run effects on income and employment

    Health Econ.

    (2005)
  • N. Datta Gupta et al.

    The Effect of an Acute Health Shock on Work Behavior: Evidence from Different Health Care Regimes

  • G. De Luca et al.

    Estimating Engel curves under unit and item nonresponse

    J. Appl. Econ.

    (2012)
  • W. Eichhorst et al.

    Combining the Entry of Young People in the Labour Market with the Retention of Older Workers

  • European Commission

    Compendium — Good Practice in Employment of People with Disabilities

    (1999)
  • R. Euwals et al.

    Early retirement behaviour in the Netherlands

    De Economist

    (2010)
  • P. Farley Short et al.

    Long-term effects of cancer survivorship on the employment of older workers

    Health Serv. Res.

    (2008)
  • P. Frijters et al.

    Mental Health and Labour Market Participation: Evidence from IV Panel Data Models

  • M. Fuchs

    Quota System for Disabled Persons: Parameters, Aspects, Effectivity

  • Cited by (0)

    View full text