Background The objective of this study was to establish the direction of causality in the relationship between labour market status and psychological well-being by investigating how transitions between secure employment, insecure employment, unemployment, permanent sickness and other economic inactivity predict changes in psychological well-being over a 16-year period.
Method This study used data from the British Household Panel Survey (1991–2007). Psychological well-being was measured using the 12-item General Health Questionnaire (GHQ-12). Fixed effects models were utilised to investigate how transitions between labour market statuses predicted GHQ-12 score, adjusting for current labour market status and a range of covariates.
Results After taking account of the contemporaneous effects of joblessness on psychological well-being, and the impact of a range of confounding factors, experiencing a transition from employment to joblessness was significantly predictive of poorer psychological well-being. Transitions into employment were not found to have equal and opposite effects: the positive effects of moving into work from unemployment were not as large as the negative effects of job loss. Transitions between secure and insecure employment did not independently predict changes in psychological well-being.
Conclusions A causal relationship between labour market status and psychological well-being is indicated.
- LONGITUDINAL STUDIES
- MENTAL HEALTH
Statistics from Altmetric.com
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.
Against a backdrop of sustained global economic uncertainty, it is of vital importance that we gain a greater understanding of how joblessness and insecure employment affect the psychological well-being of populations. This study examines the temporal sequencing of changing labour market status and variation in psychological well-being through 16 years of the lives of individuals in the UK, in order to establish the direction of causality in this relationship.
Ecological and cross-sectional studies have predominantly found a strong association between joblessness and poor psychological well-being.1–3 Workplace closure studies have shown that psychological well-being declines following redundancy4 ,5 and longitudinal studies have generally found that transitions from employment to unemployment are associated with a decline in psychological well-being, whereas transitions from unemployment to employment predict improvement.6–12 However, with regard to re-employment, some studies have suggested that selection effects may operate,13 ,14 although this is disputed.15 ,16
A limitation of much of the research in this field has been an overemphasis on unemployment and employment, a false dichotomy which fails to capture the various ways in which individuals engage with the formal labour market. In the UK, a growing gulf between declining unemployment rates and rising male inactivity rates characterised the 1990s and 2000s. These decades also saw an increase in casual and fixed-term contractual working arrangements, building higher levels of job insecurity into the labour market.17 Studies on the effects of job insecurity on psychological well-being have indicated a causal association9 ,18–21 but such research has not been widely integrated into studies on the links between labour market status and health. Unusually in the literature, Strandh22 made the distinction between secure and insecure employment, showing that in a Swedish sample, those exiting unemployment to secure employment experienced greater improvements in psychological well-being than those who gained insecure employment.22 Research by Thomas et al6 is a rare example of work in the field which addressed the effects of transitions between employment and permanent sickness or other inactivity, finding that transitions from employment to permanent sickness were associated with increased 12-item General Health Questionnaire (GHQ-12) scores, and that moving into employment from permanent sickness was associated with improved psychological well-being.6 Little further investigation of the effects of switching between employment and permanent sickness or between jobless states has been undertaken. In order to construct any hypothesis about this relationship, it is first necessary to appreciate that the experience of economic inactivity due to permanent sickness is a varied one. The onset of a physical health problem is known to have a substantial negative effect on mental well-being23 so it seems likely that moving from employment to permanent sickness would be associated with significant declines in psychological well-being.
The objectives of the present study were as follows. (1) To investigate the association between changing labour market status and psychological well-being variation within the lives of individuals over time. It is hypothesised that a significant association will be observed between labour market status change and psychological well-being change after allowing for potential confounding factors. (2) The second objective was to disentangle the processes surrounding labour market status transitions and psychological well-being decline or improvement. It is hypothesised that, controlling for current labour market status and confounders, (1) making a transition from secure to insecure employment and from either form of employment to unemployment or permanent sickness will be predictive of a decrease in psychological well-being; (2) the reverse transitions will be predictive of improvements in psychological well-being; (3) gaining an insecure job will not be as beneficial as gaining a secure job and (4) not all transitions from one labour market status to another will have an equal and opposite effect to transitions in the other direction.
Data and sample
This study uses 17 consecutive waves of data from the British Household Panel Survey (BHPS), 1991–2007. The BHPS began in 1991 with a nationally representative sample of 10 264 adults drawn from 5511 households recruited using a clustered, stratified random sampling method. Children of sample members are added to the main sample on reaching 16 years of age. Adults joining the households of sample members are included in the survey on a temporary basis, as long as they reside with original sample members. These original and temporary sample members are resurveyed annually. A detailed overview of the BHPS's methodology is given elsewhere.24 The sample used in this study was restricted to broadly working-age (16–65 years) original sample members who had complete data for the analytic variables and had contributed at least three consecutive waves to the study. These criteria yielded an analytic sample of 88 895 observations on 10 494 unique individuals over 16 waves.
The outcome of interest was psychological well-being, measured at each wave of the BHPS using the GHQ-12.25 ,26 Likert-scoring was used, producing a near normally distributed continuous outcome variable with a possible range of 0–36. Higher scores indicate lower levels of psychological well-being. To measure the exposure of interest a five-category labour market status variable was derived. These categories were: securely employed; insecurely employed; unemployed; permanently sick; other inactive. The latter was a heterogeneous category comprised of all those who were economically inactive but not permanently sick. The employed were split according to whether the respondent had reported feeling satisfaction with their level of job security or not. In order to assess the psychological effects of changing labour market status, it was necessary to derive ‘transition’ variables to capture how individuals switched from one labour market status category to another, over time. One-year lagged versions of labour market status (ie, t−1) were derived, describing the individual's labour market status in the preceding numerically consecutive wave. The lagged (t−1) and current (t) labour market status variables were combined to create 25 binary transition variables, for each possible combination of previous (t−1) and current (t) labour market states.
A range of factors hypothesised to confound the relationship between labour market status and psychological well-being were identified: age (mean-centred)1 ,27; education (highest academic qualification attained: higher educational qualification; A-levels; General Certificate of Secondary Education; none of these)28; physical health problems (one or more physical health problems)29 amount of joblessness (one or more spells of joblessness between waves)1 ,30; social housing tenure31; spousal joblessness; spousal GHQ-12 caseness32 and marital status.33 ,34 Household income was conceptualised as a mediator of the relationship.35 ,36 This was log-transformed and equivalised (using the Organisation for Economic Development and Co-operation (OECD)-modified scale).37
In order to assess the extent to which changes in labour market status predicted changes in GHQ-12 score, fixed effects (FE) regression models were used. Exploiting the inherently hierarchical structure of longitudinal data, the FE approach uses only the within-individual effects to estimate models in terms of deviations from individual-specific means.38 ,39 FE models show the extent to which a change in the exposure variable predicts a change in the outcome, within an individual, over time; thus allowing stronger causal inference. The use of FE models also permits one to be somewhat more confident about having addressed residual confounding, as by its very nature the approach adjusts for unobservable time-invariant differences between individuals.38 ,39
First, a bivariate FE model was fitted in order to establish the nature of the association between current labour market status and current GHQ-12 score. The hypothesised confounders and a hypothesised mediating factor (household income) outlined above were then added as covariates to assess whether adjustment for these factors attenuated the central relationship. Investigation of the nature of the causal relationship between labour market status and psychological well-being was then further developed by looking at the extent to which specific labour market status transitions predicted changes in GHQ-12 scores. Twenty separate FE models were run in order to establish the association between each of the binary transition variables and the GHQ-12 score outcome, controlling for current labour market status and confounders. The five stable ‘transition’ types (ie, in which the individual had remained in the same labour market status for two consecutive waves) were not included in these analyses. The transition variables which were found to be significantly associated with GHQ-12 score, independently of current labour market status and confounders, were then included in a multivariate analysis. When interpreted in conjunction with the coefficients for current labour market status, the coefficients for each of the transition variables show whether making the transition is associated with higher or lower current GHQ-12 score, over and above the contemporaneous effects of current labour market status on psychological well-being. In this way, analysis of specific labour market status transitions allows fuller understanding of the complex processes and mechanisms at work.
The characteristics of the sample are shown in table 1. Of the 88 895 person-years of data in the sample, 67% of labour market status observations were for secure employment, with a further 10% for insecure employment. Only 4% of observations were for unemployment. Just over half of all observations for physical health status identified one or more physical health problems and just 0.4% of observations indicated problems with alcohol or drugs.
The bivariate FE model showed a spell of insecure employment, unemployment, permanent sickness or other inactivity predicted a significant increase in GHQ-12 score, compared with that individual's mean GHQ-12 score during their time in secure employment (table 2). The multivariate model showed that adjustment for covariates did not greatly attenuate the association between labour market status and GHQ-12 score (table 2). Compared with when they were securely employed, an individual experiencing a spell of insecure employment had a GHQ-12 score elevated by 1.2 units. An unemployed spell was associated with a GHQ-12 increase of 2.2 units. Experiencing permanent sickness was associated with a 2.8 unit increase in GHQ-12, while a spell of other inactivity predicted a 0.4 unit increase in GHQ-12 score, compared with when the individual was securely employed. The direction of the change between secure employment and a different labour market status category cannot be inferred from these FE models; the coefficients just show the expected change in an individual's GHQ-12 value if she/he switched from being, for example, unemployed to being securely employed, or vice versa.
But how does making a transition in employment status affect psychological well-being, over and above the effects of current labour market status? Table 3 describes the prevalence of each possible labour market status transition between two consecutive years (t−1 to t).
Table 4 summarises the results of the 20 separate FE models which tested the association between the binary transition variable and GHQ-12 score. The results from the final multivariate model, showing the association between significant transitions and GHQ-12 score controlling for current labour market status and confounding covariates are shown in figure 1. When interpreting the results of the models, it should be noted that irrespective of the transition, a current insecurely employed or jobless labour market status was associated with a higher GHQ-12 score within individuals, compared with a spell of secure employment.
Having made a transition into unemployment from either secure or insecure employment was associated with decreased levels of psychological well-being, in addition to the contemporaneous negative effects of current unemployment. Moving into insecure or secure employment from unemployment was significantly associated with improved psychological well-being, but the benefits to mental health of finding a job did not appear to be as great as the negative effects of losing a job. Overall, moving from active labour market engagement, whether employed or seeking work, into permanent sickness was associated with a significantly decreased level of psychological well-being. This was over and above the negative effect of current permanent sickness on its own. Moving back into employment from permanent sickness was significantly associated with improved psychological well-being, but there was no significant change in GHQ-12 score associated with a move from permanent sickness to unemployment.
Whereas moving from ‘other inactivity’ into secure employment had a positive effect on psychological well-being, moving from employment into other inactivity did not affect GHQ-12 scores. With this exception, the securely and insecurely employed categories seem undifferentiated with regards to transition into or out of employment.
It has been shown that when controlling for a range of covariates, a spell of insecure employment, unemployment, permanent sickness or other inactivity was predictive of a significant increase in GHQ-12 score, compared with that individual's mean GHQ-12 score during their time in secure employment. These results support the study's first hypothesis.
Transitions from employment to unemployment or permanent sickness were predictive of increased GHQ-12 score, independently of the contemporaneous effects of current labour market status. This supports the findings of Thomas et al6 and is in line with the general literature on the subject. The positive effects of moving into work from unemployment were not as large as the negative effects of job loss, possibly because an experience of unemployment instils a fear of future job loss. This was also true for the transition from permanent sickness into work, which did not affect GHQ-12 scores as greatly as leaving work for permanent sickness. If there are individuals who have not recovered from their illness, but have returned to work anyway, perhaps because the conditions for receipt of sickness benefits were changed, the return to work may have a negative effect on psychological well-being, reducing the net positive effect of returning to work. Among individuals experiencing a transition from employment to permanent sickness the distress caused by anticipation and experience of redundancy could have resulted in the worsening of a pre-existing health condition, or been instrumental in the development of a mental illness, causing the individual to become ‘permanently sick’ instead of ‘unemployed’. In sum, either through the development or worsening of a medical condition or through job loss and subsequent receipt of incapacity benefits instead of unemployment benefits, a transition from employment to permanent sickness is likely to be a near-universally negative experience, predictive of a substantial and significant increase in GHQ-12 scores.
The transition analysis shows that controlling for the contemporaneous negative effects of being permanently sick having moved into permanent sickness from ‘other inactivity’ in the previous wave was associated with an increase in psychological well-being. This result is at odds with the others, which generally show that a transition into permanent sickness from any other labour market status predicted a decrease in psychological well-being. Since ‘other inactivity’ was the labour market status in which psychological well-being was most similar to secure employment, this is a counterintuitive result. In 92% of the observations reporting a transition from permanent sickness to ‘other inactivity’, the individual had reported suffering from one or more physical problem in the wave prior to becoming permanently sick, that is, while they were inactive. Therefore, it seems likely that this result is due to the psychosocial benefits that may arise from accepting the sick role40 or receiving financial or practical assistance to help with their health condition.
The initial FE model showed that a period of insecure employment was associated with a significantly higher GHQ-12 score than it would have been when the individual was securely employed. However, there appears to be no differentiation between secure and insecure employment with regard to transitions, when controlling for the effects of current labour market status. Moving from secure to insecure employment (or vice versa) was not predictive of any change in psychological well-being. Contrary to the findings of Strandh,22 there was no significant difference between the beneficial effects of moving from worklessness into a secure job compared with moving into an insecure job: both transitions were similarly predictive of increased psychological well-being.22 These results do not support the study's third hypothesis.
Strengths and limitations
A key strength of this research is the exploration of both transitions into joblessness and transitions into employment using a large panel study conducted over many years. The modelling strategy allowed isolation of the effects of the labour market status transition itself net of individual differences. This makes the causal interpretation of observational data more robust.
A subjective assessment of job security is theoretically more suitable than an objective measure (such as type of employment contract), since this study is mainly concerned with how the personal experience of feeling insecure at work affects mental health. In the context of recession, many people employed under permanent contracts who formally have a secure employment contract may feel increasingly insecure in their role.20 These individuals would not be classed as insecurely employed if an objective measure such as contract type were used. However, while the subjective experience of insecurity is of paramount interest, there is a potential question over causality. An individual with a pessimistic outlook on life may be more likely to have a high GHQ-12 score. It is possible that an employment situation which may be considered reasonably secure by a more optimistic individual could be considered insecure by someone who is more pessimistic. In addition, it is important to note that the variable used to derive the ‘insecurely employed’ category did not simply ask respondents to assess their level of job security, but asked instead the extent to which they were satisfied with their level of job security. Conceivably, a respondent could report a high level of satisfaction with their level of job security, despite being engaged in very insecure employment. It should be noted that the ‘other inactive’ group was very heterogeneous, comprising early-retirees, students, home-makers and others. It is therefore important not to overinterpret results pertaining to this group.
Previous studies have shown that the effects of labour market status on psychological distress vary by gender.6 Gender was not included as a covariate in the models presented here as it is inappropriate to adjust FE models for time-invariant factors. Stratification was considered as an analytical approach; however, some of the labour market status transitions were uncommon (table 3; eg, permanent sickness to insecure employment: n=21) rendering gender stratification inadvisable. Using a larger dataset, an investigation into differential effects of labour market status transitions on the psychological well-being of men and women would be an interesting avenue for further research. While age was adjusted for, this does not adequately address the question of whether the relationship varies according to an individual's stage in the lifecourse. Stratification of these analyses by lifecourse stage would be an interesting potential development of this work, but a larger sample or narrower focus on fewer labour market status categories would be required.
The present study offers persuasive evidence for a causal relationship between labour market status and psychological well-being by allowing us to state that, after taking account of the contemporaneous effects of joblessness on psychological well-being, and the impact of a range of confounding factors, experiencing a transition from employment to joblessness in the last year is significantly predictive of poorer psychological well-being. Moving from employment into unemployment or permanent sickness was significantly associated with lower levels of psychological well-being, even when accounting for the contemporaneous effects of unemployment or permanent sickness and a range of confounding factors. Conversely, moving from unemployment into employment was significantly associated with lower levels of psychological distress. Transitions between employment and joblessness were not found to be equal and opposite: the positive effects of moving into work from unemployment were not as large as the negative effects of job loss. This study reveals a more complex relationship between psychological well-being and labour market status transitions than hitherto shown.
What is already known on this subject
Previous research has consistently shown a strong association between joblessness and low psychological well-being. However, studies which have attempted to define the direction of causality are low in number, and the statistical methods used have generally allowed for only tentative conclusions.
What this study adds
This study employs longitudinal data analysis methods to investigate the effects of labour market status transitions on the psychological well-being of individuals in a large sample over a 17-year period. This research also separates different forms of joblessness and makes the important distinction between secure and insecure employment. The results provide strong evidence for a causal interpretation of the relationship between labour market status and psychological well-being.
Data from the British Household Panel Survey were produced by the Institute for Social and Economic Research (ISER) at the University of Essex, sponsored by the ESRC, and supplied by the Secure Data Service at the UK Data Archive. The data are the copyright of ISER. The use of the data in this work does not imply the endorsement of ISER, ESRC or the Secure Data Service at the UK Data Archive in relation to the interpretation or analysis of the data.
Contributors This study was originated by EF and AS. EF undertook the literature review, conducted the data analysis and drafted the manuscript. AS, MB and NS interpreted the results and contributed to the drafting of the manuscript.
Funding This work was supported by the UK Economic and Social Research Council (ESRC) grant number RES-596-28-0001.
Competing interests None.
Ethics approval The British Household Panel Survey has adopted, in full, the ethical guidelines of the Social Research Association. No ethical approval was required for this secondary analysis.
Provenance and peer review Not commissioned; externally peer reviewed.