The impact of atypical employment on individual wellbeing: evidence from a panel of British workers

Abstract

This study explores the relationship between individual wellbeing and atypical employment, which includes both temporary and part-time employment schemes. Individual wellbeing is measured in terms of subjective indicators of mental health, general health status, life satisfaction, and job satisfaction. It addresses four questions: (1) Are workers on a temporary contract more likely to report poor health and poor life and job satisfaction than those who are employed in permanent jobs? (2) Is this the case for part-time workers compared to those who are in a full-time job? (3) Do changes in employment profiles (e.g., from a fixed-term contract to a permanent job, or from part-time employment to full-time employment) affect individuals’ health and life satisfaction? (4) Are there differences in such relationships between men and women? To answer these questions, logistic regression models were used to analyse a panel of almost 7000 male and female workers from the first 10 waves of the British Household Panel Survey, 1991–2000. Controlling for background characteristics, atypical employment does not appear to be associated with adverse health consequences for either men or women, when both health and employment are measured at the same time. However, there is evidence that job satisfaction is reduced for seasonal/casual workers and is higher for part-timers. Taking account of selection issues does not change the general picture: the chances of poor mental and physical health and low life satisfaction are unaffected by atypical employment and some of the effects of job satisfaction persist. In addition, very few employment transitions appear to be consequential for a worsening in health outcomes, which tends to be observed in the case of job satisfaction. Although the pattern of results suggests that atypical forms of employment do not have durable adverse health consequences on workers, public policies that aim at improving the working conditions of workers in weak bargaining positions should give special attention to equity issues, including the possible health effects of experience of work in atypical employment arrangements.

Introduction

For the past two decades, many commentators have increasingly emphasised the importance of flexibility in the labour market (OECD, 1999). Employers and policy makers have seen labour market flexibility as a means of improving workers’ performance and adaptability in the face of technical change and increasing globalisation. Although a few studies document the widespread prevalence of atypical jobs in Britain (Dex & McCulloch, 1995; Booth & Francesconi, 2003), very little research has been done to analyse whether experience of work in such jobs has a positive or negative impact on individuals’ physical and mental health (Rodriguez, 2002).

The purpose of our paper is to fill this gap by investigating the extent to which specific forms of atypical work influence a broad range of health measures. In particular, we address the following questions: (1) Are workers on a temporary contract more likely to report poor health and poor life and job satisfaction than those who are employed in permanent jobs? (2) Is this the case for part-time workers compared to those who are in a full-time job? (3) Do changes in employment profiles (e.g., from a fixed-term contract to a permanent job, or from part-time employment to full-time employment) affect individuals’ health and life satisfaction? (4) Are there differences in such relationships between men and women? Throughout, and in relation to question (4), we estimate our models separately for men and women, using a large representative sample of employees from the British Household Panel Survey (BHPS), collected annually during the period 1991–2000.¹

In this study we examine two specific forms of atypical (or marginal) employment. The first refers to temporary employment, which distinguishes individuals who are on a fixed-term contract from those who hold a casual or seasonal job. The second refers to part-time employment, and, owing to specific institutional settings that are relevant over our survey period (and discussed below), separately identifies individuals who work 1–15 h a week from individuals who work 16–29 h a week. Not only have these two forms of atypical employment been examined in several previous studies concerning Britain (e.g., Blossfeld & Hakim, 1997; O’Reilly & Fagan, 1998; Dex, 1999; Rodriguez, 2002; Booth and Francesconi (2002), Booth, Francesconi, & Frank (2003); Booth and Francesconi, 2003), but they have also been the focus of recent policy initiatives. Our interest is to determine how such forms of marginal employment are associated with a number of health measures, which have been widely analysed by social and medical researchers (e.g., Clark, Osward, & Warr, 1996; Darity & Goldsmith, 1996; Theodossiou, 1998; Currie & Madrian, 1999; Frey & Stutzer, 2002; Rodriguez, 2002). These include a mental health indicator derived from the General Health Questionnaire (GHQ), a measure of perceived general health status, a subjective measure of life satisfaction, and a subjective indicator of job satisfaction.

Under the European Union (EU) directive 1999/70/EC put into force in the United Kingdom in October 2002, firms are required to offer fixed-term workers the same treatment with regard to pay and benefits (including, for example, holiday pay and maternity benefits) as permanent workers, along with the same rights to be protected against discrimination. Part-time workers are covered by similar EU directives (1997/81/EC and 1998/23/EC).² Temporary employment and part-time employment therefore represent one of the prime targets of recent public policies aimed at equal pay treatment. These policies were partly motivated by the observation that, across Europe, temporary workers and part-timers had to face less favourable pay conditions than their permanent and full-time counterparts. But little is known about the non-monetary situation of workers who are employed in atypical jobs (for exceptions, see Maier, 1991; Sneider, 1995; Hudson, 1999).

Yet knowing how such jobs affect individuals’ health (which is arguably one key non-monetary facet of workers’ life) is also important both for social analysts and health practitioners as well as for policy makers and employers. Indeed, this knowledge is likely to provide some insights into key equity considerations, which would complement the efficiency arguments advocated by those who encourage a greater flexibility in the labour market. In addition, a better understanding of the relationship between health and atypical employment is necessary to inform public policy debates over questions such as the appropriateness of work-related disability transfers, or the desirability of mandating minimum insurance coverage for psychiatric services in highly unstable jobs. Certain private sector decisions (e.g., the funding of employee assistance programs) also depend on the degree to which ill health impairs workplace performance, relative to the cost of implementing such programs.

In theoretical terms, however, the effect of atypical employment on health outcomes is ambiguous a priori. Arguably, it depends on individuals’ preferences, expectations, and financial constraints. Some people may prefer ‘stable’ jobs, which guarantee a certain flow of income but require a full-time attachment to the labour market. Others (such as the young, or the old, or women with young children) may prefer more flexible work arrangements, even if these pay less or are less secure. To the extent that individuals voluntarily select marginal jobs, we may expect that atypical employment does not have adverse health consequences, or it may even have positive health effects. For example, Hakim (1997) argues that most part-time workers (especially women) prefer to work part time rather than full time. Other social researchers have however stressed the involuntary nature of part-time employment and its costs (e.g., Burchell, Dale, & Joshi, 1997), as well as the costs associated with temporary work (Booth et al., 2002). Therefore, from a theoretical viewpoint, we cannot unequivocally say whether the health impact of atypical jobs is detrimental, or beneficial, or neutral. It is precisely this impact that will be the focus of our empirical investigation.

Fig. 1 plots the trends over the ten years between 1991 and 2000 in the four types of atypical employment that are of interest in this study for men (panel (a)) and women (panel (b)).³ The values in each graph are proportions expressed as a percentage of all employees (including permanent and full-time workers). Compared to men, women are slightly more clustered in both types of temporary employment (about 9% versus 6%), and substantially more concentrated in the two types of part-time employment (around 35% versus 5%).⁴ Interestingly, we cannot detect any remarkable trend for men (except for a modest increase in the proportion of workers on fixed-term contracts in the middle of the 1990s, perhaps as a delayed response to the recession that affected the economy few years earlier). But we do observe a steady decline in the proportion of women working 1–15 h a week, which is mirrored by a stable increase in the proportion of women working 16–29 h a week (panel (b)). This trend emerges over the entire period and particularly after the introduction of the Working Families Tax Credit and other benefits in 1998–99, which were intended to provide support for low wage families with children who worked at least 16 h per week (Blundell, 2001).

These are some of the salient institutional and macroeconomic features that render the background to this research. We now turn to some of the previous studies upon which we develop our contribution.

There is a huge amount of work that has documented the link between personal wellbeing and unemployment (see, among others, the excellent surveys by Jin, Shah, & Svoboda, 1995; Darity & Goldsmith, 1996, and Machin & Manning, 1999). Both cross-sectional and panel data analyses reveal that unemployment is systematically associated with a lower level and deterioration of physical and mental wellbeing. An issue that the research in this area has to face is that of reverse causation, whereby it is not the case that the experience of an unemployment spell leads to detrimental health outcomes but rather the opposite, that is, people with poorer health are more likely to work less and exit out of the labour market (Hamilton, Merrigan, & Dufresne, 1997; Kerkhofs & Lindeboom, 1997; Theodossiou 1998; Wadsworth, Montgomery, & Bartley, 1999; Strandh, 2000). Indeed, a well-established literature is concerned with the impact of health on employment, productivity and earnings (Bartel & Taubman, 1986; Ettner, Frank, & Kessler, 1997), as well as the impact of mental health on criminal activity (Link, Andrews, & Cullen, 1992) and divorce (Bartel & Taubman, 1986). (For up-to-date reviews, see Currie & Madrian, 1999; Frank & McGuire, 2000.) Similar considerations may apply to the case of atypical employment.

For atypical employment, however, the empirical evidence related to its health effects is scant. Fernie, Shipley, Marmot, Martikainen, and Stanfeld (2001); Fernie, Shipley, Marmot, Stanfeld, and Smith (1998) document a positive association between job insecurity and adverse psychological changes for a cohort of white-collar British civil servants. Analysing a sample of managers and employees from twenty UK firms, Burchell, Ladipo, and Wilkinson (2001) report a strong link between feelings of job insecurity and stress, and find that, for employees, such link gets even stronger as their exposure to insecurity increases. One problem with these studies is that job insecurity may be felt not only by workers in temporary or part-time jobs but also by those on full-time permanent contracts, if the fear of instability has become an endemic feature of the labour market. In addition, we do not know how these findings can be generalised to workers in other sectors and firms of the British economy. In the case of part-time employment, a great deal is known about the potential effect of health on hours of work and wages (see the review in Currie & Madrian, 1999), but our understanding of the opposite effect (that of part-time employment on health) is comparatively limited.

The paper most closely related to ours is that of Rodriguez (2002). This study finds a small and insignificant association of poor health status with marginal employment in Britain (but larger and more precisely measured effects in the case of Germany). Despite the fact that Germany is not examined here, we extend Rodriguez's work for Britain in a number of ways. First, although the data source (the BHPS) is the same, we use more waves of data: this should provide a more up-to-date picture of the relationship between health and atypical employment. Second, our regressions include a greater number of controls. Some of the variables we use (e.g., occupation and industry) are meant to ‘soak up’ part of the correlation between health measures and atypical employment that would be otherwise only spurious. Other variables (e.g., the unemployment/vacancy ratio) are intended to approximate the local labour market conditions in which workers operate and that can affect their health status independently of the specific type of job they hold. Finally, using the employment and job history files collected by the BHPS in 1992 and 1993 respectively, we also include people's entire work histories summarised by their total work experience in full-time and part-time employment. Several studies have shown that individuals’ wellbeing is strongly related to their labour force histories (Goldsmith, Veum, & Darity, 1996; Clark, Georgellis, & Sanfey, 2001), and failure to control for such trajectories may lead to biased estimates of the effect of atypical employment.

A third important extension to Rodriguez's (2002) study is that we analyse four measures of health status rather than one. To the extent that these measures cover different aspects of individual wellbeing (i.e., their correlations are less than perfect), we are likely to capture more diverse components of workers’ health and have a more sophisticated understanding of how atypical employment influences workers’ life conditions (Currie & Madrian, 1999). Fourth, our analysis tries to deal with the endogeneity (or selection) problems that plague most studies relating labour market activity and health more extensively than previous work did. The issue of reverse causality mentioned above in relation to the health/unemployment research is an example of such problems. Another example is given by the mutual association that individuals’ health and atypical employment patterns may share with some unmeasurable factors. For instance, individuals who have relatively low job aspirations may prefer a temporary or part-time commitment to the labour market while being able to do other things. Therefore, the differences in health outcomes across workers may simply reflect differences in their unobserved (or unobservable) job aspirations rather than differences in the type of their employment contract or hours of work. We address such issues with five different methods (described below), which, albeit not perfect, are meant to provide us with sensible robustness checks of our benchmark results.