A Temporal Comparison of the Effects of Unemployment and Job Insecurity on Wellbeing
by Brendan Burchell
University of Cambridge
Sociological Research Online, 16 (1) 9
Received: 15 Jun 2010 Accepted: 17 Jan 2011 Published: 28 Feb 2011
Analyses of individuals' working lives make a variety of assumptions about the relationship between time, wellbeing and economic stress. Some assume that stress will accumulate in adverse environments, leading to chronic effects of, for instance, long-term unemployment or job insecurity. Other studies emphasize the acute effects of changes per se, and assume adaptation. This paper examines how employees respond both to acute and chronic job insecurity.
This paper will use two datasets. The first is from a survey of over 300 UK employees employed in 26 companies; this dataset included both qualitative and quantitative data, at both employer and employee levels. The second dataset consisted of longitudinal data from the British Household Panel Survey. It was found that the unexpected announcement of job insecurity can cause a sudden and marked spike in psychological symptoms. Looking at longer-term effect for prolonged periods of job insecurity, wellbeing (i.e. symptoms of anxiety and depression) continues to deteriorate for at least a year, with no sign of leveling off or recovery. This is in contrast to the findings on long-term unemployment, where there is evidence of adaptation and slight improvements in psychological wellbeing after six months. The reasons for these opposing patterns between job insecurity and unemployment are discussed in terms of the challenge for individuals attempting to cope with perceived future uncertainty during the prolonged recovery from the current recession.
Keywords: Job Insecurity, Recession, Wellbeing, Unemployment, Chronic Stress
Introduction1.1 Sociologists and other social scientists have accumulated a large body of knowledge regarding the social and economic processes that accompanied the major recessions in the early1980s and early 1990s. For instance, the Social Change and Economic Life Initiative (SCELI) was explicitly commissioned by the Economic and Social Research Council (ESRC) to study the effects of the 1980s recession on employees, the unemployed, employers and their families (See, for instance, Gallie et al, 1994). By coincidence, the British Household Panel Survey (BHPS) had its first wave in 1991, and so was well placed to study the social and economic impacts of the last major recession.
1.2 There have been a number of recent attempts in the UK to marshal some of this evidence to understand the nature of labour markets under recession. This has come from both research councils (Vaitilingam 2010) and from the New Labour Government's Cabinet Office (2009), but both with the explicit objective of informing policy. Furthermore Crow, Hatton, Lyon and Strangleman (2009) argue for the usefulness of sociological comparisons of the current recession with previous recessions. Here I hope to take up that challenge, by using data from the 1990s to shed light on the effects of chronic job insecurity on individual wellbeing. Current predictions are that the recovery from the current recession may be slow and prolonged, with many groups of employees living under the threat of job loss for several years to come. Thus the longer-term effect of job insecurity is a particularly pertinent issue.
1.3 This paper starts with a brief review of some of the existing literature on the effects of job insecurity on wellbeing, contrasting it to the larger literature on the effects of unemployment. Particular attention is paid to temporal aspects of the relationship between job insecurity and employee wellbeing. This is similar to the debates about changing psychological health with the increasing duration of a spell of unemployment (although, as will become apparent, the findings are quite different). The paper then presents two separate sets of empirical analysis. The first draws upon case studies of two organisations with high levels of job insecurity, one chronic, one acute. The second set of analyses is from the early waves of the British Household Panel Survey, examining the effects of enduring job insecurity during the recovery from the UK recession in the early 1990s. Conclusions are drawn from these studies of the likely effects over time of the prolonged effects of job insecurity that many employees are likely to experience during the recovery from the current recession.
Background to Study2.1 Job insecurity has received much attention from social scientists over the past two decades, and much is already known about it. To provide a context for this paper, some important points from the literature will be summarised, although a full review of the literature would be redundant given several recent comprehensive reviews (e.g. Sverke Hellgren and Näswall, 2002; De Witte, 2005: Burchell, 2005; Cheng and Chan, 2008).
2.2 Firstly, it is important to be clear what we mean by job insecurity, as there is some inconsistency and confusion on this point in the research and policy literatures. For this paper, I am defining job insecurity as an employee's perception of the likelihood of the losing their job involuntarily in, say, the next six or twelve months. This is clearly not an objective measure, and indeed there is much evidence that, in representative surveys, many more employees are worried about losing their jobs than will actually lose them (Dickerson and Green, 2006). However, if we are concerned with their subjective wellbeing, then their perceptions of the risk of job loss are important per se, even if those fears are exaggerated or unfounded. For instance, their anxiety and the challenge to their sense of identity will be a function of their own assessment of the risk of losing their jobs. The same is true of the effect of perceived job insecurity on their job search behaviour, work motivation, and on their longer-term planning of decisions concerning matters such as housing and family formation. Thus, for the purposes of this paper, no attempt is made to measure the objective likelihood of job loss. Nor will job insecurity be measured by proxy through measures of average job tenure or turnover rates (and, in any case, these measures correlate poorly at best with subjective job insecurity, as involuntary job losses make up only a small minority of quits). Finally, job contracts have also been used in some studies as a measure of job insecurity, although this is again unsatisfactory. There is little or no relationship at the country level between the proportion of employees on temporary contracts and the proportion of employees who state that their jobs are insecure. Furthermore, Origo and Pagani's analysis of Eurobarometer data (2009) and Lewchuk, Clarke and de Wolff's (2008) analysis of Canadian data both find that wellbeing is more strongly linked to perceived job security than to the nature of the job contract.
2.3 Secondly, job insecurity is not only studied because of what might follow – unemployment. Numerous studies have shown that the very perception that one is likely to lose one's job is itself sufficient to cause symptoms of anxiety and depression. The magnitude of this effect is not trivial; typically the difference on measures of wellbeing (for instance symptoms of anxiety and depression) between secure and insecure employees is about the same size as the difference between employees and the unemployed (Burchell, 1994; Burgard, Brand and House (2009)). This finding has been replicated consistently across a number of surveys, both cross-sectional and longitudinal (see, for instance, Burchell, 2005 or Cheng and Chan, 2008 for recent reviews). Furthermore, qualitative studies have provided rich descriptions of the nature of employees' concerns about losing their jobs, worrying about maintaining their lifestyles, servicing mortgages and planning for the future (Nolan, 2002, 2009). These negative effects are not limited to the employed individuals, but extend to their spouses and families (Nolan, 2002).
2.4 The term 'wellbeing' is used widely in the social sciences to mean a number of different aspects of individual or societal functioning. In practice it is usually measured at the individual level and usually represents some aspect of affective, or emotional, wellbeing. In some cases this is related to the absence of sub-clinical symptoms of anxiety and depression; for instance the General Health Questionnaire (GHQ) is widely used in employment and stress research. It consists of a battery of questions asking about recent experiences of positive and negative affective states. It could be claimed that the linking of unemployment and job insecurity to symptoms of poor mental health is part of a wider move to medicalise individuals' life experiences, as Holmquist (2009) finds to be widespread in Swedish agencies dealing with unemployment. Alternatives to this approach ask individuals to rate their level of happiness or life satisfaction on a scale (Halpern, 2004).
2.5 Some sociologists have critiqued these approaches to wellbeing. Hughes (2006) suggests that many aspects of our lives (such as having children) can be better understood by our striving for meaning than our striving for happiness or the minimising of stress. Indeed, when Giddens (1991) addresses issues of insecurity, he is more interested in looking at the level of 'ontological insecurity', or how we have become anxious about the very nature of our selves in 'late modernity'. The reflexive project of the self, having a continued sense of the nature and meaning of one's life, can be threatened by job insecurity if, as is common, one's employment and membership of a particular occupation are central to one's self-identity. In this framework, the main threat to individual wellbeing is, in the first instance, via the threat to sense-making concerning one's identity, and the anxiety or other negative affective states that arise from job insecurity are a secondary consequence of this. Tulloch and Lupton's sociological investigations into risk in everyday life of British and Australian employees (2003) do suggest, indeed, that job insecurity is a common and salient risk for many employees. Beck's 'risk society' (1992, 2000) has been influential in framing the sociological discussions of insecurity, with labour market insecurity taken as just one form that this takes in contemporary society. Others have criticised Beck's perspective on job insecurity; for instance, Fevre (2007) demonstrated that there has not been any evidence of an increase in job insecurity in the period from the 1980s to the mid-2000s. Furthermore, unlike some of the ubiquitous environmental, medical or industrial risks that Beck describes, job insecurity is experienced by a minority of the population in most advanced industrialised countries with the majority of employees reporting that there is little or no risk to them of job loss (Green, 2009).
2.6 The widely accepted "chronology of unemployment" is that job loss is accompanied by a rapid deterioration in affective wellbeing for the first few months (Warr, 1987, Jackson and Warr, 1985). However, after about three to six months the low-point is reached, and thereafter there is a levelling off or even slight recovery in wellbeing (illustrated in Figure 1) as the unemployed individuals start to cope and adapt to their challenging circumstances. More recently, other variables have also been found to be effected by the duration of unemployment in a similar pattern, such as cortisol levels (a physiological measure of stress that does not rely on self-report) and the capacity for physical work (Maier et al, 2006). Brenner and Levi (1987) provide the most detailed data for two separate outcome measures, GHQ scores and cortisol levels, measured seven times between one month before the onset and two years after the onset of unemployment for Swedish women. The mean GHQ scores were recovering from four months after the onset of unemployment. The cortisol levels increased for the first 12 months of unemployment before starting to recover. It is also of noteworthy in their data that the period just before unemployment, when it was anticipated but had not yet been experienced, showed the worst GHQ scores and the second worst cortisol levels, and that the transition from this job insecurity to unemployment was associated by a marked (albeit short-lived) improvement on both measures. This provides further evidence that job insecurity can have a more extreme effect on individual wellbeing than unemployment.
2.7 A number of studies have proposed models of the qualitative development of reactions to unemployment over time, such that unemployed individuals go through a sequence of distinct phases such as shock, optimism, pessimism and fatalism in their mental health (e.g. Jahoda Lazersfeld and Zeisel, 1972) or morale (e.g. Harrison, 1976). Despite their popularity and appeal, mimicking prominent stage theories such as Piaget's account of cognitive development, Fryer (1992) argued that there is, in fact, no good empirical evidence for any such set of phases, and furthermore a conceptual analysis suggests that for all unemployed individuals to go through a single set of stages in the same order is highly implausible. This is supported by Brenner and Levi's data, where the mean trends are statistically significant, but individual trajectories show marked heterogeneity (1987). These arguments can also be applied equally to job insecurity. Although any one individual is likely to experience a number of different phases as they experience job insecurity, it is unlikely that there is anything resembling a general pattern in the order or duration of qualitatively different phases.
2.8 Researching the temporal changes in wellbeing that accompany stressors poses a number of methodological challenges. Frese and Zapf (1988) suggest designs that can be used to track the effects of a stressor before, during and after its presence. However, by their own admission, these designs are challenging and few studies achieve their optimal criteria. There are still some published studies (e.g. Stankunas et al 2006) that purport to measure the effects of short- vs long-term unemployment using cross-sectional designs, taking no account of differential attrition, i.e. the greater propensity of healthier and more optimistic unemployed individuals to regain employment (Claussen, et al, 1993), thus confounding duration effects with survival effects. Even with longitudinal designs, the time lag between measurements can be critical in determining the exact nature of an effect, i.e. whether the effect peaks during or after the stressor is present. They give several examples of plausible relationships between the onset of a stressor and its effect on sociological or psychological functioning.
2.9 Very little has been known about the changing relationship between job insecurity and wellbeing over time, perhaps, as Baert (2000) has argued, reflecting a wider neglect of temporal issues in sociology. There have been some isolated studies that have already investigated a small number of aspects of this relationship. Longitudinal studies serve not only to confirm cross-sectional evidence concerning the causal direction linking job insecurity to poor wellbeing, but also to investigate the effects of chronic job insecurity, where the insecurity is experienced over a prolonged period of time. One argument has been that harmful stress is most likely where there has been a sudden and unexpected increase in job insecurity, whereas employees might adjust over time to insecure work with only minor consequences for health and wellbeing. An alternative view, however, is that chronic levels of job insecurity are more likely to result in adverse reactions. This second view is consistent with many models of stress and coping that suggest that an individual's capacity to cope with the stressor becomes exhausted over time (Lazarus and Folkman, 1984).
2.10 There is some empirical evidence for both of these scenarios. Highlighting the initial shock hypothesis, some studies have examined workers when they have only just been made aware of reduced security (for instance, the announcement of a proposed downsizing) but these studies typically do not have a control group or longitudinal component that permit the effect over time to be charted. A good example is Jacobson (1987) who managed to gain access to Israeli civil servants just days after it was announced that many of their jobs, previously assumed to be secure, were to be shed in a redundancy programme. Many of the descriptions of their feelings, such as 'Bewildering' and 'Disrupting' suggest a short-term shock reaction, but the study did not pursue them beyond that initial stage.
2.11 Other studies indicate that the prolonged experience of job insecurity leads to increasingly impaired wellbeing of employees. Heaney et al. (1994) examined the effects of job insecurity in a mainly male sample of 207 automobile manufacturing workers in the US and found that chronic job insecurity was predictive of changes in job satisfaction and physical symptoms over time, over and above the effects of job insecurity at any one point in time. This effect has been replicated by Burgard, Brand and House (2009) using larger and representative samples from the US in the three and ten year periods 1986-1989 and 1995-2005 respectively. In both cases they found that, where respondents had already reported that their jobs were insecure at time 1, if they were still insecure at time 2, then their wellbeing had deteriorated compared to the averages for other groups. This effect remained significant even after rigorous statistical controls were introduced into the models, providing strong support for the increasing negative effect of job security over time.
2.12 A problem with studies that chart changes in the reaction to job insecurity over time is that, while the starting point of a period of unemployment is relatively easy to pin-point, the point at which an individual starts to be concerned over the security of their job might be more difficult to place. This paper is then fortunate in being able to use two very different studies to examine the wellbeing of insecure employees at very different points in time. Study 1 concentrates on two companies out of a sample of 25 case studies. In one of these companies, purely by chance, a matter of days or weeks before the employees were interviewed a major merger was announced which transformed a formerly secure workforce to being very insecure. In the second establishment employees had been experiencing insecurity and a gradual downsizing for several years. Study 2 uses an annual panel survey to examine the effects of a transition from secure employment to chronic insecurity over a period of three annual waves.
Study 1: The shock of job insecurity3.1 This research was conducted as part of a larger project into job insecurity, funded by the Joseph Rowntree Foundation (see Burchell et al, 1999, 2002). Twenty-five case-studies were undertaken, and senior managers were interviewed as well as a number of employees proportional to the size of each establishment. In total over 300 employees were interviewed across all 25 organisations. The cases were chosen to cover the main sectors of British employment, including the public and private sectors and manufacturing and service industries. Others cases were chosen for their innovatory uses of either secure employment practices (for instance, two organisations had negotiated zero-redundancy policies with trade unions), and several organisations which had come to rely extensively on temporary and agency employment. Taken as a whole, they were not representative of employment in the UK, but did give a diverse and theoretically interesting sample.
3.2 Interviewees within each of the cases were chosen to represent some of the main occupations in that organisation, but were otherwise selected to be as representative as possible. The interviews typically lasted about 45 minutes face-to-face, and employees also filled out self-completion forms that included the GHQ-12, a widely used measure of wellbeing as gauged by symptoms of mild anxiety and depression.
3.3 Rather than looking in detail at all of the 25 cases, each with their own histories and idiosyncratic employment practices, two cases will be selected for particular scrutiny. These are the two cases with the highest levels of job insecurity, as measured by taking the mean value of all the employees when asked to rate their own job on a five-point scale from "very secure" to "very insecure" (Figure 2).
Case 1: InsureCo
3.4 One of those cases was a large employer in the financial services sector with a head-office in the City of London and a network of branches around the country (it had a presence in several other countries too). The first interview at this company was with the UK Managing Director and the UK head of HRM, and was conducted in a large and opulent office overlooking the City of London and River Thames, on the top floor of the company's iconic London headquarters. One of the recurring themes of that interview in early 1998 was that the company had managed to avoid the large-scale redundancies which had been prevalent in the rest of the financial services industry over the preceding decade. The two senior managers prided themselves in the fact that they were even prepared to keep on some staff where they were temporarily surplus to requirements, because "it was not their fault that we decided to restructure" (MD, UK). The HRM director made it clear that he had an informal contract with staff that, as far as possible, everyone's job was safe; this he saw as important to making the staff feel valued. The company was not the highest payer in their sector, but claimed some pride in the supportive working environment provided for their employees including many innovative equal-opportunities schemes. Their only cautionary note on the topic of job security was to say that, in the case of a major merger, "all bets are off". They kindly made arrangements for the research team to visit two of their regional offices and to interview a total of 24 employees in addition to the branch managers.
3.5 About six weeks after this first interview, it was announced that the company would merge with another finance company of a similar size and competing for the same market. It was abundantly clear to employees that there would have to be a large number of redundancies in both companies. Our interviews followed this announcement by about a week and a half in one branch, and by three weeks in the second. This unexpected announcement was still the main talking point at both of these branches which had been characterised by predictability but were now brimming with uncertainty over what the future would hold in terms of selective redundancies or even the complete closure of some branches.
Case 2: CopperCableCo
3.6 The other company that had equally high levels of employee insecurity, studied later in 1998, was a manufacturer of copper cable that had been trading for 80 years. Over the last few years it had become clear that this company was competing against several bigger ones, and that the world market was over-capacity as copper cable was being replaced by fibre optic technologies. Between interviews the research team had the opportunity to walk about in the factory. It was working well below capacity, with many machines and employees standing idle. The only hope was that it might be acquired by a larger company in that field to increase its market share or invest in newer technologies, but even then its future would have been far from secure. Nineteen employees were interviewed, ranging from machine operators to managers. It was clear from those interviews that their fears for the future had not emerged recently, but had existed for several years of gradual decline in output.
Comparing the Cases
3.7 A GHQ score was created for each employee by summing the 12 items scored on a scale from 1 to 4, logged to remove an upward straggle and then averaged for each establishment. A mean insecurity score was also created for each establishment by averaging each employee's score on the five-point measure of job insecurity. As can be seen from the scatterplot in Figure 2, these two cases had, by a considerable margin, the two highest levels of employee job insecurity in the sample of 25 establishments. The difference in job insecurity between the two sets of employees in the finance and manufacturing companies was negligible (T=0.3, df=41, p=.78).
3.8 However, when one looks at the stress being experienced by employees in these two cases, one notices a very different picture; the finance company, with its very recent increase in job insecurity had the highest aggregate level of stress of all the establishments. However, the employees in the manufacturing case, equally insecure, had much lower levels of stress, as measured by the GHQ-12 (T=3.1, df=40, p<0.005).
3.9 The comparison between these two companies is far from a perfect "natural experiment"; they differed in many other ways apart from the recency of job insecurity, such as industry, (financial services vs. manufacturing) and socio-economic status (predominantly manual vs. non-manual workers). Although many of the features of the finance company made it a better working environment and the pay was higher, employees also reported that there was a lot of pressure from the sheer volume of work. Yet the conclusion one might tentatively draw is not a surprising one; when a group of relatively secure employees experience an unexpected deterioration in their perception of their job security, they experience a sudden and large increase in symptoms of anxiety and depression.
Study 2. The longer term.4.1 The second part of the empirical results in this paper uses data extracted from the British Household Panel Survey (BHPS) that has interviewed the same sample, representative of the British population, at yearly intervals from 1991. The sample achieves approximately 5,000 employees in each wave, making it a powerful resource to study changes in the supply side of the British labour market. In each wave respondents are asked how satisfied they are with their job security, and the GHQ-12 is administered. Furher details of the survey can be found in Taylor at al (1996).
4.2 While the survey did not include a direct measure of perceived job security, employees were asked to rate their satisfaction with several aspects of their job, including job security, on a seven-point scale from one (Not satisfied at all) to seven (Completely satisfied). Although not semantically the same as asking directly about the level of job security, in practice this measure behaved in a similar manner to more direct measures, and has been used by other researchers as a measure of job security (e.g. OECD, 1998).
4.3 The satisfaction with job security variable showed a notable upward straggle, with six and seven being the modal categories. For the purposes of the analyses that follow, the variable was categorised into 3 groups, insecure (1-3, 18%), medium (4-5, 28%) and secure (6-7, 55%) satisfaction. The percentage figures are the proportions at Wave 1; they changed very little, on aggregate, over the five waves, 1991-1995. As before, the GHQ-12 is also included in the BHPS interview schedule as a measure of employees' wellbeing.
A Longitudinal analyses between two waves.
4.4 Here we explore the relationship between changes in JS (job security) and changes in GHQ scores. For instance, if one takes only those who felt secure at Wave one, and investigated the fate of the employees who dropped to medium or insecure on the JS measure at Wave two, do their GHQ scores also change?
4.5 The results clearly demonstrate (using repeated-measures MANOVAs) that changes in the level of satisfaction with job security co-occur with changes in wellbeing for the four sets of adjacent-pair waves. Although the numbers changing between JS categories between waves were in some cases small, the results were consistently in the predicted direction, even if they were not always statistically significant. However pooled, the overall pattern is that the level of change in GHQ scores is statistically significant and directly proportionate to the level and direction of change in JS.
4.6 To aggregate these data, the transitions across all four pairs of waves were pooled. Figure 3 shows what became of those who were insecure at Time 1, broken down into three groups: became secure, became medium and stayed insecure at Time 2. The left-hand bar of each pair show the GHQ scores at Time 1, the right-hand bars show the GHQ scores for the same respondents a year later at Time 2 (in this dataset the GHQ scores were best normalised from the raw 12-48 scores with a reciprocal square root). Importantly, those who became more secure showed an improvement in their GHQ scores, but those who stayed insecure showed a decline in their GHQ scores, suggesting that chronic insecurity was accompanied by a deterioration in wellbeing (this is demonstrated even more convincingly later in the analyses, when changes over three waves are considered).
4.7 Figure 4 shows those who were medium JS at Time 1, and Figure 5 shows those who were low JS at Time 1. The conclusions to be drawn reinforce those from Figure 3. In each case an increase in job security is matched with improved GHQ scores, and a reduction in job security was associated with a deterioration in GHQ scores. All three show highly significant two-way interactions between Wave and Job Security at Time 2 and in the expected directions (all ps<0.001).
4.8 One factor that makes these results a little equivocal, however, is the fact that there are sometimes already differences in the groups at "Time 1", while they all have the same rated level of job insecurity. This might be a function of residual heterogeneity within groups. Figure 5 is an example of this. Clearly, there are large differences at Time 2, such that those who have stayed insecure deteriorated slightly in their GHQ scores, those who became medium on JS showed some improvement, and those who became secure improved the most. Note however that there were some smaller differences already at Time 1 when all groups were secure.
B Longitudinal analyses between three waves.5.1 For the purposes of this paper, the most interesting analyses examine the chronic effect of job insecurity by looking at patterns across more than two waves. This can be explored most directly by selecting out those employees who were secure at Time 1, but insecure at Times 2 and 3. Because of the small numbers involved, these cases were combined with those individuals who showed the same "Secure-Insecure-Insecure" pattern over waves 1,2 and 3: 2,3 and 4 and 3,4 and 5 (i.e. in the period 1991-1995, during the recovery from the early 1990s recession).
5.2 The results of this pooled dataset are shown in Figure 6. If these employees habituated to job insecurity and learned coping mechanisms to deal with it, we might expect to see a recovery in their GHQ scores between the T2 and T3. The opposite is clearly the case, with a continued, marked and statistically significant deterioration in GHQ scores between T2 and T3, This supports the "chronic effects" hypothesis of continued deterioration in wellbeing as the duration of job insecurity continues.
Discussion6.1 The empirical findings in this paper show markedly different effects for the initial onset of job insecurity and for the chronic effect of prolonged job insecurity. The evidence suggests that the variation of employee wellbeing over time might look something like that in Figure 7 with an initial "shock" phase, followed by a recovery and then a more gradual, prolonged decline. This is suggested as one possible time series that is consistent with the studies presented here, and is certainly not based on anything like a definitive study. Evidence from this paper and other studies has demonstrated the existence of an initial shock phase and a continued decline phase. The intervening recovery stage has not been directly observed, but is inferred. A healthy skepticism is therefore recommended until more carefully controlled analyses are available to chart the changing wellbeing of insecure workers over time. Presumably this line will level off at some point, but the data presented here shows that the decline continues, and possibly even accelerates, for at least the first one to two years after the switch from job security to job insecurity. More convincing evidence is likely to come from a dataset explicitly collected to examine the effects of job insecurity over time; as Frese and Zapf (1988) argue, the provision of unequivocal evidence of the precise mapping of the effects of stressors over time requires very specific methodological designs, and these requirements are unlikely to be satisfied through the secondary analysis of existing datasets; monthly rather than annual measurements are probably needed for this.
6.2 This pattern is in marked contrast to the relationship generally found between wellbeing and time for long-term unemployment in two important respects. Firstly, the initial reaction to the onsets of job insecurity and unemployment would seem to be diametrically opposed to each other. With unemployment, particularly when it comes after a period of job insecurity, there is evidence from quantitative studies that there can be an initial period of relief with improved wellbeing in the first few weeks (Brenner and Levi, 1987; Cobb and Kasl, 1977), and this theme is supported by many qualitative studies (see Fryer, 1992). The present study suggests that the onset of perceived job insecurity, following a period of perceived security, causes a sudden plunge in wellbeing. While the evidence, based on the comparison of just two employers, is far from compelling, it is supported by evidence from at least one qualitative study (Jacobson, 1987).
6.3 The second marked differences between these two trajectories concerns the longer term. Evidence from several studies suggests that individuals tend to acclimatise to unemployment, and start to recover in their personal wellbeing after a few months. The evidence presented in this paper goes beyond previous studies that have found chronic job insecurity to be more problematic than short-term spells of job insecurity, and shows a continued decline in wellbeing over time (if anything, accelerating rather than decelerating).
6.4 One suggestion for an explanation for the differential effects of job insecurity and unemployment is not to be found in the past or present but the future. As Fryer (1992) argues, a high level of uncertainty over the future is widely held to create a stressful state. The early months of a period of unemployment are characterized by uncertainty over many aspects of the individual's life, including uncertainty about future re-employment, uncertainty about budgeting, an inability to make plans for the coming weeks and months. However for those individuals who remain unemployed for longer periods of time, some predictability returns to their life. They will start to plan, probably on the basis that they will continue to be unemployed in the medium term; the longer an individual is unemployed for, the more predictable life becomes. The stress may be ameliorated by the knowledge that "things can only get better". On the other hand, for the employee who remains insecure, there may be no return to greater certainty as time progresses, as they see the continued downsizing of their organization or decline of their industry. Experiencing colleagues being made redundant can also be stressful and cause feelings of guilt in the 'survivors' (Hallier and Lyon, 1996). The longer that their future remains uncertain, the more their ability to cope with that insecurity declines over time.
6.5 An alternative or complementary explanation for the results presented in this paper might be that this relationship between chronic job insecurity and affective wellbeing is mediated by another form of wellbeing, as described by the concept of 'ontological insecurity'. The prolonged periods of uncertainty intensify feelings of existential anxiety. Cole (2007) argues that sociologists studying unemployment have, for many decades, accepted the findings of social psychologists in the tradition of Marie Jahoda. This has, then, been accompanied by an uncritical acceptance of assumptions about human nature and the role of work. Unfortunately there does not yet exist a coherent and comprehensive sociological literature that can provide an alternative explanation of the empirically observed relationships between work and wellbeing. Perhaps detailed (and probably qualitative) elucidation of the importance of job security for wellbeing (in its wider forms) will be an important step along this road.
6.6 Before pondering for too long on the explanations for these effects, one should find some corroborating evidence for the changes in wellbeing levels over time. The evidence presented here is far from conclusive. For instance, the measurement of wellbeing is open to artifact; the GHQ measure, widely used in organizational psychology and sociological research, particularly focuses respondents on change "over the past few weeks". It is therefore possible that it will exaggerate the effect of the initial sudden onset of job insecurity. By the same token this makes the chronic effects even more worrying, as it probably underestimates those long-duration effects. It should also be noted that the empirical work in this paper is concerned with psychological affective wellbeing, partly because measures of this are widely available. If other forms of wellbeing were considered, for instance physical wellbeing or existential wellbeing, it is likely that these might show different patterns of changes over time. And, even with panel data, multivariate models controlling for potentially spurious time-varying effects might be superior to the simpler models presented in this paper, if appropriate data were available.
6.7 One should also be cautious about over-generalising these findings. For instance, the effects and characteristics of job insecurity may be dependent on aggregate factors that exist in the region at the time of the research, such as the economic climate, active labour-market policies, generosity of unemployment benefits and stigma of unemployment. Therefore one should be wary of generalizing these findings between countries or time periods with economic and employment environments. The labour markets in the US and developing countries might well be markedly different to the UK. Having said that, most of the other accumulated knowledge that has been gained from European-wide surveys over the past two decades has shown that the effects of job insecurity are remarkably consistent across European countries (Burchell, 2009). There is also the possibility that the effects of job insecurity are mediated by demographic variables such as age, sex, socio-economic status and tenure, although the literature suggests that the effects of job insecurity are pervasive, affecting many different demographic groups with surprising consistency (Chen and Chang, 2008).
6.8 These findings provide cold comfort at the time of writing, in a recession that most economic forecasts predict will be long with a slow recovery. We have seen the initial shock as many employees from previously secure jobs have realized that their employers will be making redundancies or even closing completely. The prolonged, but less visible, misery is only just starting as the duration of job insecurity drags on for many employees. As governments and employers plan for the social costs of recession, the findings in this paper suggest that those costs might continue to increase for some considerable time after unemployment has peaked, as the duration of spells of job insecurity lengthen.
6.9 These findings also suggest that the labour market will remain a polarizing force throughout the recession. Like unemployment, the experience of job insecurity is concentrated in certain individuals and groups. Perhaps then we are not experiencing a 'risk society' but a 'society polarized by chronic risk'.
AcknowledgmentsAcknowledgments to Stephanie Auge who conducted the data manipulation and analyses for the BHPS findings, Michelle Sequeira and Nurjk Agloni for assistance in the preparation of the graphics and manuscript and to the rest of the team involved in the collaborative JRF project, and the Joseph Rowntree Foundation for funding that covered part of the research for this paper. I am also grateful for the comments of two anonymous referees.
Notes1MANOVA is a variation of the Analysis of Variance (ANOVA) statistical procedure that permits the analysis of the mean scores on the dependent variable (i.e. GHQ) by variables that compare differences between respondents (i.e level of job security at time 2) and variation within the scores for each respondent (i.e. time 1 vs time 2). These analyses were conducted in SPSS).
BAERT, P.(ed.) (2000) Time in Contemporary Intellectual Thought. New York: Elsevier North-Holland.
BECK, U. (1992) Risk Society: Towards a New Modernity. Cambridge: Polity Press.
BECK, U. (2000) The Brave New World of Work. Cambridge: Polity Press.
BURCHELL, B.J. (1994). 'Who is Affected by Unemployment? Job Insecurity and Labour Market influences on Psychological Health.' in Gallie, D., Marsh, C. and Vogler, C. (eds.) Social Change and the Experience of Unemployment. Oxford: Oxford University Press.
BURCHELL, B.J. (2005) The Welfare Costs of Job Insecurity: Psychological Wellbeing and Family Life. Trends in Social Cohesion Special Issue: Reconciling Labour Flexibility with Social Cohesion – Facing the Challenge 15: 71-108.
BURCHELL, BJ., Day, D., Hudson, M., Ladipo, D., Mankelow, R. Nolan, J., Reed, H., Wichert, I., and Wilkinson, F. (1999) Job Insecurity and Work Intensification; Flexibility and the Changing Boundaries of Work. Joseph Rowntree Foundation report.
BURCHELL, BJ., Ladipo D. and Wilkinson, F. (Eds) (2002) Job Insecurity and Work Intensification. London: Routledge.
BURCHELL, BJ (2009) Flexicurity as a moderator of the relationship between job insecurity and Psychological well-being Cambridge Journal of Regions, Economy and Society. Advance Access published on August 6, 2009.
CABINET OFFICE (2009). Learning from the past: working together to tackle the social consequences of the recession: Evidence pack. Social Exclusion Task Force.
CLAUSSEN, B., Bjørndal A., and Hjort P.F. (1993) Health and Re-Employment in a two year follow up of Long Term Unemployed. Journal of Epidemiology and Community Health. 47(1):14-8. [doi:10.1136/jech.47.1.14]
COBB, S. and Kasl, S.V. (1977). The consequences of job loss. N.I.O.S.H. Research Report, U.S. Department of Health, Education and Welfare, publication no. 77-224.
DE WITTE, H. (2005). Job insecurity: Review of the International Literature on Definitions, Prevalence, Antecedents and Consequences. SA Journal of Industrial Psychology, 31(4): 1-6.
DICKERSON, A. and Green, F. (2006) How should we measure the fear of job loss? Paper prepared for Fifth IZA/SOLE Transatlantic Meeting of Labour Economists, 18–21 May 2006, Buch-Ammersee, Germany.
FRESE, M. and Zapf, D. (1988) Methodological Issues in the Study of Work Stress: Objective vs Subjective Measurement of Work Stress and the Question of Longitudinal Studies. In Cooper, C.L. and Payne, R. (Eds) Causes, Coping and Consequences of Stress at Work. Chichester: John Wiley and Sons.
FRYER, D. M. (1992). Psychological or Material Deprivation: Why Does Employment Have Mental Health Consequences? In McLaughlin, E. (ed) Understanding Unemployment: New Perspectives on Active Labour Market Policies. London. Routledge. [doi:10.4324/9780203061459.ch6]
GALLIE, D., Marsh, C. and Vogler, C. (eds.) (1994) Social Change and the experience of Unemployment. Oxford: Oxford University Press.
GIDDENS, A. (1991), Modernity and self-identity. Self and society in the late modern age. Cambridge: Polity Press.
GREEN, F. (2009) Subjective employment insecurity around the world. Cambridge Journal of Regions, Economy and Society. 2: 343-363
HALLIER, J., and Lyon, P. (1996) 'Job insecurity and employee commitment: managers' reactions to the threat and outcomes of redundancy selection', British Journal of Management, 7(1): 107-23. [doi:10.1111/j.1467-8551.1996.tb00109.x]
HALPERN, D. (2004) Social Capital. Cambridge: Polity Press.
HARRISON, R. (1976) The demoralising experience of prolonged unemployment, Department of Employment Gazette, Vol. 84, pp. 339 – 348.
HEANEY, C.A., Israel, B.A. and House, J.S. (1994) Chronic Job Insecurity among Automobile Workers Effects on Job Satisfaction and Health. Social Science and Medicine, 38(10): 1431-1437. [doi:10.1016/0277-9536(94)90281-X]
JAHODA, M. Lazarsfeld, P. and Zeisel, H. (1972) Marienthal: The Sociography of an Unemployed Community (London ,1972).
JACKSON,P. and Warr, P. (1984) Unemployment and Psychological Ill Health: the Moderating Role of Duration and Age. Psychological Medicine, 14: 605-614.
JACOBSON, D. (1987) A Personological Study of the Job Insecurity Experience, Social Behaviour, 2:143-55.
LAZARUS, R.S., and Folkman, S. (1984). Stress Appraisal and Coping. New York: Springer.
MAIER, R., Egger, A., Barth, A., Winker, R., Osterode, W., Kundi, M., Wolf, C. and Ruediger, H. (2006) Effects of Short- and Long-Term Unemployment on Physical Work Capacity and on Serum Cortisol. International Archives of Occupational and Environmental Health, 79: 193-198. [doi:10.1007/s00420-005-0052-9]
NOLAN, J. (2002) 'The Intensification of Everyday Life'in Burchell, B. and Ladipo, D. (eds) Job Insecurity and Work Intensification, London, Routledge.
NOLAN, J. (2009) Working to live, not living to work: an exploratory study of the relationship between men's work orientation and job insecurity in the UK. Gender, Work and Organization, 16: 179–197. [doi:10.1111/j.1468-0432.2008.00426.x]
OECD (1998) Economic Outlook. Paris, OECD.
ORIGO F and Pagani, L. (2009) Flexicurity and job satisfaction in Europe: The importance of perceived and actual job stability for well-being at work. Labour Economics, 16, 547–555. [doi:10.1016/j.labeco.2009.02.003]
STANKUNAS, M., Kalediene, R., Starkuviene, S. and Kapustinskiene, V. (2006) Duration of Unemployment and Depression: a Cross-Sectional Survey in Lithuania. Bmc Public Health, 6:174. [doi:10.1186/1471-2458-6-174]
SVERKE, M., Hellgren, J., and Näswall, K. (2002). No Security: A Meta-Analysis and Review of Job Insecurity and its Consequences. Journal of Occupational Health Psychology, 7: 242-264. [doi:10.1037/1076-89126.96.36.199]
TAYLOR, M. F. (ed). with Brice J., Buck N. and Prentice E. (1996). British Household Panel Survey User Manual Volume A: Introduction, Technical Report and Appendices. Colchester: University of Essex.
VAITILINGAM. R. (2010). Recession Britain: Findings from economic and social research. Swindon: ESRC.
WARR, P. (1987) Work, Unemployment and Mental Health, Oxford: Oxford University Press.