Article Text

Download PDFPDF

Welfare regimes, labour policies and unhealthy psychosocial working conditions: a comparative study with 9917 older employees from 12 European countries
  1. Nico Dragano1,
  2. Johannes Siegrist1,
  3. Morten Wahrendorf1
  1. 1Department of Medical Sociology, University of Duesseldorf, Duesseldorf, Germany
  1. Correspondence to Dr Nico Dragano, Department of Medical Sociology, University Clinic Duesseldorf, University of Duesseldorf, PO Box 10 10 07, D-40001 Duesseldorf, Germany; dragano{at}uni-duesseldorf.de

Abstract

Background Recent analyses explored associations of welfare state regimes with population health, with particular interest in differences between social protection-oriented versus more liberal regimes. Little is known about such associations with work-related health. The aims of this contribution are (1) to study variations of quality of work according to type of welfare regime and (2) to analyse differences in the size of effects of quality of work on workers' health according to type of welfare regime.

Methods The authors use cross-sectional and longitudinal data from two studies (‘Survey of Health, Ageing and Retirement in Europe’ and the ‘English Longitudinal Study on Ageing’) with 9917 employed men and women (aged 50 to 64) in 12 European countries. Psychosocial quality of work is measured by low control and effort–reward imbalance at work. Depressive symptoms are introduced as a health indicator. Linear multilevel models and logistic regression analyses are performed to test the hypotheses. In addition to the welfare regime typology, the authors introduce labour policy and economy-related macro indicators.

Results Between-country variations in quality of work are largely explained by macro indicators and welfare regimes, with poorer quality of work in countries with less emphasis on older workers' protection. Moreover, in the Liberal and Southern welfare regime, effects of quality of work on depressive symptoms are relatively strongest (adjusted ORs varying from 1.45 to 2.64).

Conclusion Active labour policies and reliable social protection measures (eg, Scandinavian welfare regime) exert beneficial effects on the health and well-being of older workers. More emphasis on improving quality of work among this group is warranted.

  • Occupational health
  • occupational stress
  • elderly
  • social policy
  • multilevel modelling
View Full Text

Statistics from Altmetric.com

Introduction

The world of work is a highly political sphere. Systems of social protection as well as national or international economic and labour market policies set the frame of working life in modern states. These macro-level characteristics are obviously linked to the micro-level of everyday working conditions through pathways such as occupational safety legislation, dismissal protection laws or minimum wage policies. The broader political context of labour is of interest for occupational health practitioners and researchers because it may impact on occupational health by enhancing healthy or unhealthy work environments.1 2 Building on recent comparative investigations of associations between welfare state regimes, political systems and population health,3–12 we set out to study the following two more specific questions focussing on the psychosocial work environment: (1) Does the prevalence of a health-adverse psychosocial work environment vary between different types of welfare states? In addition, does it vary according to distinct indicators of national labour policies? (2) Does the strength of the effect on health produced by an adverse psychosocial work environment differ according to type of welfare state regimes? Specifically, we are interested to know whether the impact of a stressful work environment on the health of older employees is less pronounced in social protection-oriented welfare regimes compared with more liberal regimes.

Exposure to an adverse psychosocial work environment was shown to increase the risk of physical and mental disorders, sickness absence, work disability and mortality.13–17 A substantial part of this research is based on measuring the following stressful aspects of work and employment: jobs with high demands and low control,18 work characterised by an imbalance between efforts spent and rewards received,19 organisational injustice20 and job insecurity.21 Although rooted in everyday individual working experiences, these aspects are related to macro-level processes and structures.1 2 22 23 For instance, the risk of experiencing job insecurity is determined by contextual factors such as economic cycles or the presence and extent of dismissal-protection laws.23–25 Other important macroeconomic and political dimensions with impact on working conditions are labour market structures and policies as well as specific institutional environments.1 Systematic empirical explorations of links between the micro and macro level are largely missing, but previous studies observed variations in the prevalence of a stressful work environment between countries and economic sectors.2 24 26–29 We hypothesise that this variation is the result, at least in part, of underlying differences in macroeconomic and political structures.

Moreover, contextual factors may also modify the effects exerted on workers' health by an adverse psychosocial work environment. While this assumption has not yet been tested in a rigorous way, indirect evidence indicates that workers' health in case of job loss is less affected in the presence of extended national unemployment protection measures.3 30 31 Thus, social protection might function as personal resources which facilitate coping with stressful life events and material deprivation.32

Against this background, using data from two European ageing studies, we examine links between macro-level factors, stressful psychosocial work environments and older employees' health in terms of depressive symptoms. To this aim, we apply a comparative design with data from 12 countries and introduce two analytical strategies to account for macro-level factors: (1) grouping countries by types of welfare regimes and (2) using distinct work and employment-related macro-indicators. The first approach refers to welfare-state research and the identification of ideal types of welfare regimes. These types differ in the selection of, and emphasis on, social policy dimensions such as the degree of redistribution, coverage, family–state–market relations or labour-marked policies and employment protection.33–36 Employment and work are important facets of all typologies, since welfare states are designed to compensate loss of income from employment in cases of unemployment, disability or retirement. Furthermore, labour policy in itself is an inherent part of welfare-state activities—for example full-employment policies.35 To distinguish welfare states, we rely on the influential typology of Esping-Andersen who identified Liberal, Conservative and Scandinavian welfare-state regimes.35 Additionally a fourth type—the Southern regime—was added in order to adequately represent southern-European countries in our study.34

As welfare-state typologies are rather general and incorporate other fields of social policy (eg, family), a second complementary approach was chosen. Countries were characterised by means of specific work and employment-related indicators at the national level. These indicators (eg, old age employment rate) should allow us to estimate the impact of macroeconomic structures and labour policies on work and health in older employees more precisely.

Methods

Data and study design

Data were obtained from two European Ageing studies: the ‘Survey of Health, Ageing and Retirement in Europe’ (SHARE)37 and the ‘English Longitudinal Study on Ageing’ (ELSA).38 Cross-sectional data are available from both studies from the year 2004. Longitudinal data can be analysed with the second SHARE wave in 2006. Here, we test the effects of a stressful psychosocial work environment (2004) on future risk of depressive symptoms (2006). The design of SHARE and ELSA was developed in close coordination, thus enabling us to compare data from 12 European countries (ELSA: England; SHARE: Sweden; Denmark; Germany; Netherlands; Belgium; France, Switzerland; Austria; Italy; Spain, Greece). In this analysis we restricted the sample to all men and women aged 50–64 years reporting to do any paid work (employed or self-employed). This restriction resulted in a maximal cross-sectional sample size of 9917 participants (5383 men; 4534 women). The longitudinal sample is restricted to countries from the SHARE study and respondents that participated at both waves (n=4755; 2611 men; 2144 women). Studies were approved by the institutional review boards (SHARE: University of Mannheim, Germany; ELSA: Multi-Centre Research Ethics Committees in England).

Measurement

Individual-level data were collected by CAPI interviews and self-administered questionnaires. An adverse psychosocial work environment was measured by a short battery of items derived from (a) the Job Content Questionnaire measuring the demand-control model39 and (b) the effort–reward imbalance (ERI) model questionnaire.29 Items were selected on the basis of factor loadings on core scales of original measures. With regard to the first model, measurement was restricted to the control dimension (sum score of two Likert-scaled items ranging from 2 to 8, with higher scores indicating lower control at work). To measure effort–reward imbalance, two items measuring ‘effort’ and five items assessing ‘reward’ were included. ‘Effort–reward imbalance’ was defined by a ratio of the sum score of ‘effort’ items (nominator) and ‘reward’ items (ranging from 0.25 to 4.0, with higher values representing a higher imbalance).

As health indicators, we used two binary measures of depressive symptoms, the Centre for Epidemiologic Studies Depression (CES-D) scale40 and the EURO-D depression scale.41 Short forms of the CES-D were available both in the SHARE (11 items) and in the ELSA (eight items) study. According to predefined cut-points, the presence of depressive symptoms was identified. As EURO-D (12 items) was included in the baseline and follow-up examination of SHARE, this measure was chosen as the health outcome in longitudinal analyses.

Age, gender, income, education and employment characteristics were included as covariates. Total annual household income, composed of the sum of different income components, was adjusted for household size. In accordance with the OECD equivalent-scale household income was categorised into country-specific tertiles based on all available cases in each country. Education was measured according to ISCED-97 categorised into ‘low education’ (preprimary, primary or lower secondary), ‘medium education’ (secondary or postsecondary) and ‘high education’ (first and second stage of tertiary). Two variables related to employment were analysed: employment status (self-employed vs employed) and work time (full vs part time).

Welfare regimes and macro indicators

Table 1 contains a short description of the four types of welfare regimes, the Liberal, Conservative, Scandinavian and Southern type. The first three types are based on Esping-Andersens typology.35 The Southern type was added, because most Southern European countries share unique features and could best be treated as a distinct regime3 34 Our classification of countries follows that proposed by Bambra and Eikemo3 (table 1).

Table 1

Welfare regime types and macro level indicators

In addition to the welfare-state typology, three macro indicators at the national level for the year 2004 were identified, using the Eurostat database which draws on the EU Labour Force Survey42: old age employment rate, old age long-term unemployment rate and extent of further education among adults (table 1).

Analyses

According to our research questions, we performed two sets of analyses. The first involves studying the effects of macro factors on stress-related working conditions. To this aim, we first present scatter diagrams with associations between work stress measures (country mean) and the four macro variables at aggregate level. Moreover, following a recent approach to quantify country differences7 multilevel models were conducted to explain work stress with individuals (level 1) nested in countries (level 2). In particular, we tested to what extent macro indicators could explain the between-country variations of work stress.43 For each work stress measure, three models were calculated. Model 1, an ‘empty’ model, was estimated to disentangle the between-country variation from the within-country variation. In model 2, individual-level variables were introduced to account for socio-demographic differences at individual level. Third, the macro variables were integrated, and the reduction in the between-country variation was assessed. Given the small number of participating countries, for each macro variable a separate model was calculated (Model 3a–3d). In respective tables, we present the estimated between-country variations together with model fit statistics (log likelihood, Akaike Information Criterion and Bayesian Information Criterion). Moreover, the proportional reduction in between-country variance (R22) was calculated.44

In a second set of analyses, we studied whether the effect of work stress on depressive symptoms varied between welfare regimes using logistic regression models. Adjusted models were estimated for the total sample and separately for each welfare type. Low control and effort–reward imbalance were defined according to the criterion of scoring in the upper country-specific tertile of the respective scale distribution.45 In addition to cross-sectional analyses, we calculated the effects of work stress on prospective depressive symptoms with the longitudinal sample. In this later case, results were adjusted for baseline depressive symptoms. In doing so, the estimated coefficients allow us to explore to what extent work stress measures are associated with changes in depressive symptoms between baseline and follow-up. In the tables, ORs and 95% CIs are given. All calculations were performed using the STATA 10 statistical package (STATA, College Station, Texas).

Results

The dataset contains information on 9917 men and women from 12 countries, with the highest number of respondents in England and the smallest number in Austria (table 2). Slightly more than half of the population under study were men, mostly working as employees in full-time employment.

Table 2

Sample characteristics (N=9917)

With regard to the first research question, figure 1 gives a preliminary answer. Concerning the three continuous macro indicators, we observe that the psychosocial work environment is worse (low control, high effort–reward imbalance) in countries with lower rates of old age employment and in countries with higher rates of long-term unemployment among older people. Similarly, work stress is higher in countries with low proportions of people engaged in continued education. With regard to welfare regimes, results differ between the two work stress models, but in either case, work stress is highest in countries of the Southern type.

Figure 1

Correlation between work stress measures (mean country scores) and macro indicators/welfare regimes. For country abbreviations, see table 1. ERI ratio high values: high imbalance; control scale high values: low control; dashed line: regression line; for welfare regimes, 95% CIs are shown.

Next, we present results from multilevel models with work stress scales as dependent variables (table 3). In the empty model, significant between-country variations are observed, but the interclass correlation is relatively small for both measures (ERI: ICC=0.034; low control: ICC=0.050), indicating that the largest variation exists within countries. In model 2, individual characteristics are included, but the reduction in the between-country variation is absent in case of low control and minimal (13%) in the case of ERI. This indicates that variations of work stress at country level are only modestly explained by differences in the composition of national samples. In contrast, in models 3a–d, where the macro factors are introduced, the between-country variation is substantially reduced, specifically so with regard to low control where a large part of level 2 variance is explained. For instance, in case of welfare regimes the proportional reduction in variance is 73.7% (R22).

Table 3

Reduction in the between-country differences in work stress (low control and effort–reward ratio): results of multilevel models

The second study question concerns variations in the strength of the association between an adverse psychosocial work environment and depressive symptoms in relation to the type of welfare regime. Adjusted ORs derived from logistic regression analyses indicate significantly elevated prevalences of depressive symptoms among those who score high on the work stress measures. However, the size of these odds varies between the different welfare types. In the cross-sectional analysis (upper part of table 4) ORs are highest in the Liberal welfare type and lowest in the Scandinavian welfare type. In the longitudinal analysis (lower part of table 4) where the Liberal welfare type is not included, highest ORs are observed in the Southern and the Conservative welfare type, whereas in the Scandinavian type, they are again found to be lowest. Effects of work stress on depressive symptoms are slightly stronger in case of effort–reward imbalance than in the case of low control, and as ORs of depressive symptoms are additionally adjusted for baseline depression they tend to be generally lower compared with the cross-sectional analysis.

Table 4

Associations between work stress and depressive symptoms at baseline and at follow-up

Discussion

This study provides evidence on two relevant aspects of how national labour policies and welfare state regimes impact on an adverse psychosocial work environment and work-related health of older employees. First, in countries with an active labour policy manifested by a high old age employment rate, by a low prevalence of long-term unemployment, and by a culture of further adult education mean levels of psychosocial stress at work are lower than in countries with a less pronounced labour marked integration of older workers. This finding is of interest given the documented associations of stressful work with health in general, and with depressive symptoms in particular.14 15 46 Furthermore, our second result suggests that the effect of work-related stress on health (depressive symptoms) is less pronounced in older workers living in countries with social protection-oriented welfare regimes, compared with those living in more liberal welfare states.

Our first finding is open to different interpretations. For instance, the association could be the result of direct influences of legislation on working conditions—for example by occupational safety and health regulations, dismissal protection or minimum wage policies. Alternatively, welfare regimes could coincide with specific organisational structures and cultures like leadership practices, educational systems or employee participation.1 Our observation that investments into continued education correlate with better working conditions supports this latter interpretation. Yet, the observed associations can also result from a generally more favourable appraisal of working conditions by people who experience the benefits of worker-friendly policies.2 Moreover, general cultural differences or global macroeconomic characteristics of a nations' economy like the sectoral mix or the degree of globalisation in business may play a role.24

The second main result indicates that health effects of work stress vary by type of welfare regime. In particular, strong associations between stressful work and depressive symptoms are evident in countries of the Southern welfare regime (longitudinal data) and in the country representing the Liberal welfare state (cross-sectional data). In the Liberal welfare regime, social protection regulations are relatively weak, and labour policies follow principles of neo-liberalism,10 33 thus leaving a burden of coping with adverse work to individual workers. The Southern type is characterised by a highly fragmented social protection system with ‘macro-scopic gaps of protection.’34 These gaps concern crucial issues, such as unemployment protection and regulations in favour of employment security. It is likely that threats to job stability and related stressful features of work are more often perceived by older workers under these conditions.

Our study has several strengths. First, it represents information drawn from a large sample of older men and women from 12 European countries. The survey was conducted according to a rigorously controlled study protocol, including standard procedures of translating the measures into different languages and of collecting and controlling data. Second, as results from cross-sectional and longitudinal data analysis are included, the methodological problem of common method variance is limited to some extent. Third, to our knowledge, this is the first report documenting a variation of the size of effect of work stress on reduced health (depressive symptoms) according to type of welfare state regime. In this approach, the measurement of work stress was based on two established theoretical models.

These strengths have to be weighted against limitations. For instance, we were not able to apply the full original scales measuring high demand/low control and high effort/low reward jobs, due to constraints in data collection, or to include additional measures of an adverse psychosocial work environment. Moreover, no objective measure of the psychosocial work environment was available to validate the self reports.

Although the overall sample is large and represents the population of the respective age groups within countries quite well,37 we cannot rule out an unobserved selection bias in particular during follow-up. Yet, we applied Heckman's model of sample selection bias,47 and no relevant changes in results were apparent.

A further limitation concerns the lack of specificity of welfare-state typologies with respect to employment conditions. We tried to compensate for this lack by additionally analysing quantitative indicators of national labour policies with relevance to job conditions, keeping in mind the complexity of their meaning in different contexts. For instance, a low old-age unemployment rate could be the result of an active labour market policy but also of retirement policies pushing elder employees to disability or early retirement. Finally, the number of countries should be extended, thus, offering opportunities of testing our hypothesis more strictly, of increasing the statistical robustness and of exploring additional variations—for example by gender.

In conclusion, keeping in mind these limitations, it can be stated that an active labour policy for older workers and reliable social protection measures exert beneficial effects on occupational health and well-being.

What is already known on this subject

  • Health adverse working environments have been shown to vary between different countries.

  • Little is known about the impact of country-specific labour and welfare policies on these differences.

What this study adds

  • Results show that indicators of specific labour policies (eg, old age employment rate) and the type of welfare regime are related to levels of psychosocial stress at work. We found a better quality of work in countries with an active labour policy for older workers and reliable social protection measures.

  • Associations between work stress and health (depressive symptoms) are strongest in the Southern welfare states and weak in the Scandinavian regime type.

References

View Abstract

Footnotes

  • Funding This paper uses data from SHARE waves 1 and 2, as of December 2008. SHARE data collection in 2004–2007 was primarily funded by the European Commission through its 5th and 6th framework programmes (project numbers QLK6-CT-2001- 00360; RII-CT- 2006-062193; CIT5-CT-2005-028857). Additional funding by the US National Institute on Ageing (grant numbers U01 AG09740-13S2; P01 AG005842; P01 AG08291; P30 AG12815; Y1-AG-4553-01; OGHA 04-064; R21 AG025169) as well as by various national sources is gratefully acknowledged (see http://www.share-project.org for a full list of funding institutions). Data from ELSA were made available through the UK Data Archive. ELSA has been supported by grants 2RO1AG7644-01A1 and 2RO1AG017644 from the NIA and a consortium of UK government departments coordinated by the Office for National Statistics. The authors are grateful to A Börsch-Supan and to M Marmot, the directors SHARE and ELSA respectively, for their support. Moreover, we would like to thank the Hans-Boeckler-Foundation, Duesseldorf, Germany, for continued funding of the presented comparative analyses (project no S-2007-997-4 and S-2009-311-4).

  • Competing interests None.

  • Ethics approval Ethics approval was provided by the SHARE/ELSA: respective national ethics/data safety authorities.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.