Beyond Job Strain: Employment Strain and the Health Effects of Precarious Employment

Transcription

1 Labour Studies Programme Beyond Job Strain: Employment Strain and the Health Effects of Precarious Employment by Wayne Lewchuk, Labour Studies & Economics, McMaster University Alice de Wolff, Alliance on Contingent Employment, York University Andy King, United Steel Workers of America Michael Polanyi, KAIROS: Canadian Ecumenical Justice Initiatives 1 Work in a Global Society Working Paper Series Published by: Labour Studies Programme McMaster University

2 Work in a Global Society Labour Studies Programme Welcome to the Work in a Global Society (WIGS) working paper series! It has been developed so that original research can be available to the academic and activist community before submission for publication. This series is edited and graduate student papers are peer reviewed. Labour Studies is an internationally recognized interdisciplinary program housed in the Social Sciences at McMaster. Established in 1976, we offer certificate, undergraduate and graduate programs. We conduct research in work that includes family life, unions, community relations, gender and equality, and state policy. We feel that Work & Society is broadly defined to include those in paid and unpaid work, and in traditional and non-traditional workplaces. Printed copies of these papers may be purchased by contacting us and electronic copies are available on our website. For more information, please feel free to contact us at: Labour Studies Programme McMaster University 1280 Main Street West Hamilton, Ontario L8S 4M4 Canada (905) extension To quote one of our students: Labour Studies is intensely political, constantly controversial and refreshingly socially-conscious. I hope that you find our working paper series as stimulating, Charlotte Yates, Director and Editor, WIGS Working Paper Series

3 BEYOND JOB STRAIN: EMPLOYMENT STRAIN AND THE HEALTH EFFECTS OF PRECARIOUS EMPLOYMENT October 2004 Contact author: Wayne Lewchuk, Labour Studies Program, McMaster University Abstract This article develops a new conceptual approach to measuring the characteristics of precarious employment and its effect on health. Our starting point is the Karasek job strain model. We argue that job strain focuses on the health effects of work once people are employed. It is less effective in capturing the health effects associated with the employment relationship, the process by which workers acquire work, keep work and negotiate its terms and conditions. We develop a new construct, employment strain to measure these aspects of work organization. Employment strain is a multi-dimensional measure taking into account the effort expended finding work, keeping work and balancing multiple employers; the degree of uncertainty regarding future earnings and working conditions; the uncertainty associated with having access to the resources needed to sustain a household; and the nature of support when employment relationships are temporary. Evidence is presented based on a sample of 800 individuals in various forms of precarious employment.

4 Introduction Across Europe and North America, the 1980s and 1990s was a time of change in the nature of employment relationships (Vosko 2000; Cranford, Vosko, Zukewich & Cranford 2003; Goudswaard & Andries 2002; Pedersen, Hansen & Mahler 2003). The standard employment relationship based on permanent full-time employment became less important. Precarious employment relationships grew in importance including permanent part-time, temporary, and own account self-employment. This study explores the nature of precarious employment relationships, with a particular focus on understanding their impact on health outcomes. A number of studies have focused on how precarious employment relationships alter exposure to physical risks such as noise and toxins, or the psychosocial work environment including control over work and workload (Benach et.al. 2002b; Goudswaard & Andries 2002; Saloniemi et.al. 2004; D Souza et.al. 2003). This paper takes a different approach and suggests that the employment relationship itself is a source of exposures that can affect health. To explore this hypothesis we introduce a new concept of employment strain. Employment strain is a multifactor measure of the control, workload and support characteristics of an employment relationship. This paper will focus on the development of this new construct and report results from a first wave of fieldwork. Our point of departure is the Karasek Job Demand-Control (JD-C) model and the concept of job strain (Karasek 1979; Karasek and Theorell 1990). Job strain is found when jobs provide low levels of control over how work is done while at the same time requiring high expenditures of psychosocial effort to complete assigned tasks. Studies show that workers exposed to job strain are more likely to be exhausted, depressed and dissatisfied with their job, and they are more likely to have stress-related illness and cardiovascular disease (Belkic et al. 2000; Karasek and Theorell 1990; Marmot 2000). Other studies have shown that chronic exposure to job strain increases blood pressure (Schnall et al. 1992; Schnall et al. 1998; Landsbergis et al. 1994; Laflamme et al. 1998). A variant of the JD-C model, the ISO-strain model explores how the health effects of "job strain" are mediated through social support (Johnson 1991).

5 Karasek developed his analysis in the early 1970s in a world where, at least for most white men, the standard employment relationship (full-time, permanent employment, contractually defined terms and conditions of work) was the norm. Under the standard employment relationship, how people acquired work, how they kept work, and how they negotiated the terms and conditions of work was determined by a dense web of labour market institutions and regulations. For those lucky enough to be working under these conditions, the employment relationship itself was less of a contested process and hence the employment relationship itself, did not play an important role in the Job Demand-Control model. A central argument of this paper is that Karasek s job strain captures only one dimension of the control-demand-support trilogy at work, that being the dimension associated with the work process and the production of goods and services. We argue that the control-demand-support trilogy associated with the employment relationship is equally important, and that this trilogy has become more relevant with the spread of precarious employment relationships. How might the employment relationship affect the control, effort and support characteristics of work? The employment relationship shapes the level of control workers have over access to employment itself and the setting of its terms and conditions. It influences the amount of effort required searching for employment, the effort expended to keep employment, and the effort required balancing demands related to multiple job holdings at multiple sites. The employment relationship shapes support in three ways. First, the temporary nature of employment relationships influences the relationship between worker and employer, and between workers and levels of support at work. Second, it influences household relations and the ability to satisfy domestic needs. Pay rates are often low, benefits virtually non-existent, and the variability of earnings makes planning difficult. Third, the variability of work demands and irregular schedules can affect the relationship between workers and their community. Those working within the "job strain" framework have become increasingly aware of the limitations of the original control and workload constructs proposed by Karasek and the need to consider the employment relationship as a factor affecting health outcomes (Wall, T., et.al Work in a Global Society, McMaster University page 2

6 & 1996; Sparks & Cooper 1999; de Jonge, J., et al & 2000; NIOSH 2002; Benach et.al. 2002a). In particular, models need to incorporate what Cooper has called the shift to a shortterm contract culture or the spread of precarious employment (Cooper 2002). Quinlan, Mayhew, and Bohle (2001) suggest adding precarious dimensions to the understanding of job control and using this to explore the relationship between precarious work and health outcomes. Benach et.al. (2002a) suggest the need to develop other conceptual alternatives based on the social structure of work, such as precarious work, which might be considered a multidimensional construct defined according to a number of dimensions, such as temporality, powerlessness, or lack of social benefits. Benach et.al. (2002b) point towards the importance of including the employment relationship in any analysis of work and health. Their work based on European Union data concludes that the employment relationship may have an independent effect on health outcomes, regardless of differences in working conditions. The lack of discussion until recently on how the employment relationship itself alters the control/workload/support characteristics of work reflects a broader failure to appreciate how the spread of precarious employment relationships is affecting workers and the power they enjoy. It is not unusual to hear popular expressions suggesting that temporary workers or the selfemployed have more control over their work activities based on a greater freedom to refuse less desirable work assignments compared to those in permanent full-time jobs. Or that those working at home under contract may be freer to organize how they work as management moves away from direct supervision. This may be the case for the minority of contract workers with highly sought after skills (Smith 2001; Casey & Alach 2004). However, to suggest that workers in general have more control under this new form of employment is to misunderstand the real gains workers made under the standard employment relationship and the shift from casual labour markets at the beginning of the 20th century (Scott 2004). Under the standard employment relationship, the rights of workers to access employment were enhanced while the rights of employers not to continue employing a worker were significantly curtailed. The freedom of employers to respond to changing market contexts by adjusting the Work in a Global Society, McMaster University page 3

7 number of workers or the terms and conditions of employment were dramatically reduced. The standard employment relationship represented a clear transfer of control and power to workers over whether they have more work, the terms and conditions of work, where they work, when they work, and in many cases with whom. Employers were constrained by the employment contract, and the legislative and legal framework, limiting their right to hire and fire at will. For many precarious workers, the right to choose whether to work, which could be interpreted as an increase in control, may in fact represent quite the opposite. Under the precarious employment relationship, workers have lost contractual and regulatory control over whether they have a job, the terms and conditions of work, where they work, when they work, and with whom they work relative to workers in standard employment relationships. The increase in choice, and control, made possible by the freer application of competitive market rules pales next to the loss of control as a result of the gutting of the employment relationship. Workers have become free agents to be replaced by employers as needed, in time and space. Our challenge was to design a study that captured the unique employment relationship characteristics of precarious employment. To do this we needed to find indicators that went beyond those useful in the study of work organization at individual workplaces, where employment is full-time, where workers have a permanent and ongoing employment relationship, and where the terms and conditions of employment are either contractually defined or based on well established customs and norms. To incorporate the employment relationship into work organization models we define a number of new indicators of control, workload and support. Together they make up a new work organization construct employment strain. Employment strain is not a substitute for Karasek s job strain construct. Rather, employment strain captures a dimension of work organization that has an independent health effect regardless of the level of job strain. Other studies have shown that job strain and high job insecurity are independently associated with a number of mental health outcomes (D Souza et.al. 2003). Details of this new construct and how we propose to measure it are provided in what follows. Work in a Global Society, McMaster University page 4

8 The paper is divided into three sections. In Section One we discuss the different forms of precarious employment, and trace the history of precarious employment in the Canadian economy. In Section Two we examine the literature linking precarious employment relationships and health outcomes. In Section Three we develop the concept of employment strain and present initial findings. Section One: Precarious Employment in Canada This section traces the historical evolution of precarious employment using Canada as a case study. The experience in Canada is similar to that recorded for much of Europe (Rodgers 1989; Goudswaard & Andries 2002; Pedersen, Hansen & Mahler 2003). Precarious employment is not new to the Canadian labour market. Early in the 20 th century many dock workers, construction, agricultural, food processing, forestry, factory and garment workers were hired at plant gates, street corners or hiring halls for a day, a week or a month at a time. Employment was precarious, at the whim of the employer and could be terminated without warning. These conditions galvanized unions in the early part of the century to force employers to create more stable employment. While few workers received life long commitments of employment, the employment relationship did become less casual for many. Under the emerging standard employment relationship, the terms of employment were governed by legally enforceable contracts that defined rights to continuing employment, the terms and conditions of work, seniority based job rights, and a process for renegotiating terms and conditions. This represented a significant victory for workers giving them some control over key aspects of the employment relationship. The gains were heavily gendered and constructed around a nuclear family with a male wage earner and an unpaid female domestic partner. For a period of forty years after World War II, most men in the workforce had relatively stable employment relationships: they worked fulltime, their salary and benefits kept pace with inflation, and they worked for long periods in one job, often for the same employer. The male dominated, permanent workforce was supplemented Work in a Global Society, McMaster University page 5

9 by a number of temporary or part-time jobs. These jobs were generally held by women, recent immigrants and by young workers. The compensation rationale for these jobs, used by both employers and governments, was that these workers were not primary earners and that women and younger workers were supplementing a partner s or father s living wage - even though significant proportions of women and immigrant workers were sole support heads of households. This justified paying lower wages and benefits, limiting the rights codified into legally enforceable contracts, and limiting rights under the emerging system of labour market regulations. Compared to workers in standard employment relationships in this period, the largely female workforce in precarious employment had less control over future employment or the terms and conditions under which they worked. Precarious employment began to grow steadily and significantly through the 1980s. Figure 1 indicates changes in the forms of employment that are usually associated with precarious employment. Overall, these forms of employment have increased from 24% to 32% of the total Canadian workforce. The share of Canadians working in non-standard employment forms (i.e. not full-time permanent employees) stabilized in the mid-1990s. However, this masks an important shift in the composition of non-standard employment away from permanent part-time work to more precarious forms including full-time temporary employment and part-time temporary employment. Work in a Global Society, McMaster University page 6

10 Figure 1 Non-Standard Forms of Employment in Canada: (1) % total employment Part-time permanent Self-employed Full-time temporary Part-time temporary Source: Cranford, Vosko & Zukewich, 2003, p. 12. Many new workers in precarious employment relationships find themselves negotiating terms and conditions of employment within the context of existing norms regarding non-standard work (Vosko 2000). It is assumed that workers in precarious employment relationships are supplementary earners justifying low wage packages, limited benefits, lack of rights to future employment, and marginal legislative and social program coverage. As well as the direct costs in terms of lower earnings and benefits, we argue that there are potential health costs associated with precarious employment. Section Two: Precarious Employment and Health Serious investigation of the health and safety consequences of precarious employment is quite recent. The evidence suggests a complex association between the employment relationship and health. A number of studies suggest that workers in precarious employment relationships are more likely to be exposed to physical situations that increase the risk of work related injury or Work in a Global Society, McMaster University page 7

11 illness (Platt 1999; Quinlan 1999, Benavides & Benach 1999; Storrie 2002; Benach et.al. 2002b; Pedersen et.al. 2003; Daubas-Letourneux & Thébaud-Mony 2003). Quinlan, Mayhew, and Bohle (2001) reviewed 93 studies dealing with precarious employment and health outcomes. Over 80 percent reported precarious employment was associated with an increased risk of injury, disease and hazard exposure or reduced knowledge of occupational health and safety issues and regulatory responsibilities. However, other studies have suggested a more mixed relationship between precarious employment and health outcomes. Precarious employment was associated with poorer health outcomes in a study of German workers (Rodriguez 2002) but no effect was found in a study of British workers (Bardasi & Francesconi 2004) or a sample of Finnish workers (Virtanen et.al. 2002). Studies based on a sample of European Union workers found a complex relationship between the employment relationship and health outcomes (Benach et.al. 2002b; Pedersen et.al. 2003). The European Foundation for the Improvement of Living and Working Conditions (EFILWC) studies on non-standard employment arrangements are an important source of research on the health impacts of precarious employment. Researchers associated with EFILWC have conducted three waves of surveys (in 1991, 1996 and 2000) with approximately 1,500 workers in each of the 15 member states of the EU (Goudswaard & Andries 2002; Benach et.al. 2002b; Daubas- Letourneux & Thébaud-Mony 2003; Pederson et.al. 2003). On the question of whether precarious employment is associated with poorer health outcomes, the EU studies provide evidence that certain workers with non-permanent contracts tend to report poorer levels of quality of life and health, on some, but not all, measures. For example, non-permanent workers, in general, report lower levels of job satisfaction (Benach et.al. 2002b). Self-employed workers in particular report more back pain, muscular pain and fatigue than those in other work arrangements. Full-time fixed term workers report high levels of back pain and fatigue, and small employers (those employing less that 15 workers) report high levels of fatigue and stress. Interestingly, non-permanent workers, in general, report lower levels of stress, and less absenteeism, than permanent workers. Work in a Global Society, McMaster University page 8

12 The European studies suggest that some workers in precarious employment relationships face certain kinds of high-risk physical working conditions. For example, non-permanent or flexible workers appear more likely to work in ergonomically unsafe positions (Goudswaard & Andries 2002; Daubas-Letourneux & Thébaud-Mony 2003,) compared to permanent workers. However, some forms of non-standard employment can be protective: part-time workers report being exposed to fewer physical hazards (Goudswaard & Andries 2002), less intensity of work (Benach et.al. 2002b). Here, less time at work appears to reduce the level of hazard resulting from exposure to physical and environmental risks. Some findings are mixed. Non-permanent workers appear to face lower levels of overall job demands (Goudswaard & Andries 2002) perhaps because of their position in the organization. Other analysis suggests that nonpermanent workers are not more exposed to physical hazards such as noise, vibration and temperature (Goudswaard & Andries 2002). Research on Finnish workers also challenges the simple notion that precarious employment is always associated with poorer psychosocial working conditions (Saloniemi et.al. 2004). The studies are nearly unanimous that precarious employment increases levels of insecurity and that insecurity can lead to poorer health outcomes (Ferrie 2001; Goudswaard & Andries 2002; Oeij et.al. 2002; Ferrie et.al. 2002). This point has been made from a different perspective in a recent paper on worker power under flexible labour market regimes (Scott 2004). Research also indicates that workers in precarious employment relationships may experience less social support in the workplace. For example, non-permanent workers have been found to experience more interruptions (Benach et.al. 2002b) and higher levels of harassment and discrimination (Goudswaard & Andries 2002) compared to permanent workers. Workers with atypical schedules also experience more harassment and discrimination (Daubas-Letourneux & Thébaud-Mony 2003). They have been found to have fewer training opportunities, and fewer discussions about work problems with colleagues, bosses and personnel representatives Work in a Global Society, McMaster University page 9

13 (Goudswaard & Matthieu 2000). There are also examples of tension between permanent and non-permanent employees (Byoung-Hoon & Frenkel 2004). An important point in understanding relative health risks is that it is difficult to separate work related risks narrowly defined as exposures such as noise, toxins and ergonomic conditions, with other factors affecting health such as living conditions, housing and social protection which can be directly influenced by the work one does (Daubas-Letourneux & Thébaud-Mony 2003). In this study, factors such as income, household income, household employment benefits and social support are included as work related variables affecting health outcomes In summary, these studies suggest various reasons why workers in precarious employment relationships may experience work related health problems: workers in such arrangements may face marginally more difficult working conditions, they experience higher levels of job insecurity, they have lower levels of control over their working conditions and arrangements, they experience poorer social interactions, and they may be exposed to particular demands associated with their employment arrangements. However, at the same time these studies suggest that on some measures of health outcomes, workers in standard employment relationships report the worst outcomes. These studies do point the way to a deeper understanding of the relationship between the employment relationship and health outcomes. They suggest that insecurity, continuous evaluation, frequent changes in worksites and work colleagues, frequent bouts of unemployment, unpredictability of work schedules and earnings, poor living conditions and differences in coverage of social regulations related to precarious employment may play an important role in understanding the health outcomes of these workers. The final section of this paper uses these insights as a starting point and offers a new way of modeling the health risks of precarious employment. Work in a Global Society, McMaster University page 10

14 Section Three: Modeling the Health Effects of Precarious Employment Our starting point is that differences in the work tasks actually performed by workers in precarious and standard employment relationships are not the only, and quite possibly not the main, factor explaining differences in their work related health outcomes. Hence it is not surprising that studies that focus on standard work-related health risks (exposure to dangerous substances, biomechanical risks and job strain) sometimes find ambiguous results. To advance our understanding of how precarious employment affects health outcomes we explore the impact of a relatively understudied set of work organization characteristics related to the employment relationship itself. We divide the total work related health effect of employment into a work process component and an employment relationship component. The work process component measures the health effects related to exposures to traditional work related health risks including exposure to dangerous substances, biomechanical risks, and job strain. These hazards are the product of the production process and how work is organized. The level of exposure to these risks may be influenced by the bargaining power associated with different forms of the employment relationship and the ability to accept or to decline particularly hazardous work. They are not however characteristics of the employment relationship itself. The employment relationship component measures exposure to employment strain, a hazard directly determined by the employment relationship. Employment strain will be high where there is uncertainty regarding access to work, where there is uncertainty regarding the terms and conditions of employment, where search is costly, where wages and benefits are low, and where relations with employers and co-workers are temporary. We expect that precarious employment will lead to high levels of employment strain and high levels of employment strain are predicted to lead to poorer health outcomes. Work in a Global Society, McMaster University page 11

15 Work Process Health Risks and Precarious Employment The effect of precarious employment relationships on work process related health risks is predicted to vary. Those involved in precarious employment may have less invested in a particular employment relationship and may find it easier to abandon a work assignment rather than face exposure to this class of risks. Others with highly sought after skills may find themselves in a strong bargaining position and be able to demand healthier work environments. However, for the less skilled, the insecurity associated with precarious employment and the lack of control over the terms and conditions of employment may lead workers to accept more hazardous assignments. As argued above, research supports the conclusion that there is no clear relationship between precarious employment and levels of exposure to physical and work related psychosocial health risks (Goudswaard & Andries 2002; Parker 2002; Daubas-Letourneux & Thébaud-Mony 2003; Saloniemi et.al. 2004). Employment Relationship Health Risks and Precarious Employment The core of our hypothesis is that focusing on the work-related health risks of precarious employment ignores a second source of work-related health risks associated with the employment relationship. The growth of precarious employment re-problematises the employment relationship as a source of work-related health risks. While there are health risks present in all employment relationships, studies that assume the standard employment relationship have not made them visible. The differences in the rights enjoyed by those in precarious employment relationships and standard employment relationships leads us to argue there is a second pathway between work and health outcomes, fundamentally different from the production process based pathway examined by Karasek. This alternative set of risks can be described as employment strain. Work in a Global Society, McMaster University page 12

16 While employment strain is a hazard all workers face related to the nature of their employment relationship, these risks are likely to be higher for those in precarious employment relationships where the unique social organization of their employment relationship may lead to: C C C C C Increased uncertainty over access to future employment and the terms and conditions of employment. Added demands associated with the constant search for new employment and the need to balance multiple employers and worksites. Added demands related to ensuring a positive employer assessment of work performance needed to increase the probability of being offered more work. Increased uncertainty over the ability to satisfy minimum household economic demands as a result of low pay, limited benefits and high levels of variance in earnings. Reduced levels of social support and increased risk of harassment from employers and co-workers as a result of the temporary nature of social relations. Employment strain is not a substitute for Karasek s job strain construct. It is constructed from different dimensions of the work experience than job strain. Measuring Employment Strain To measure employment strain we designed a fixed response self-administered survey. Between 2002 and 2004 we received surveys from 800 Canadian workers. 1 Those who had not worked in the last month were dropped from the sample leaving 786 surveys with useable data. The majority of the respondents lived in large urban centres, most in or near Toronto. The sample is composed of workers who responded to ads placed in newspapers; employees of a temporary agency; homecare workers; university workers; community workers and a diverse group found through employment agencies and worker-based groups in Toronto. 2 A number of unions helped us in the research which has likely had the effect of increasing the levels of union Work in a Global Society, McMaster University page 13

17 representation in our sample relevant to the underlying population of workers in precarious employment relationships. We used the survey questions to construct new measures of control, workload and support related to the employment relationship. Together these constructs measure employment strain. Details of these new constructs and how they are measured are provided below. 3 Employment Relationship Uncertainty We define three types of uncertainty associated with the employment relationship; work uncertainty, earnings uncertainty and scheduling uncertainty. High levels of uncertainty are synonymous with low levels of control. 4 Employment relationship uncertainty is calculated by summing the values of the questions representing its three components described below. Work uncertainty measures the level of control over future employment and the frequency with which employment terms are re-negotiated. All employment in a competitive labour market is uncertain, but for those involved in precarious employment, the degree of uncertainty is qualitatively different. In the absence of an ongoing relationship with an employer or contractually defined rights to further employment, workers in precarious employment relationships face a high degree of uncertainty over getting more work. Work uncertainty includes two questions on respondents perceived uncertainty about whether current employers will offer more work and average contract length. Earnings uncertainty measures the level of control workers have over future earnings. It includes seven questions on whether the worker can predict future earnings, the existence of written pay records, whether unemployment insurance and government pensions are deducted from earnings, whether workers are paid when they are sick, whether they are paid on time, and whether they have disability insurance and pension entitlements. Work in a Global Society, McMaster University page 14

18 Scheduling uncertainty measures the control workers have over when and where they will work. It is constructed from three questions about the length of advance notice of work schedules, hours to be worked and work location. Employment Relationship Workload We define four types of workload associated with the employment relationship; effort finding work; multiple employers/worksites effort; constant evaluation effort, harassment and discrimination effort. Employment relationship workload is calculated by summing the values of the questions representing its four components described below. Effort finding work measures time spent looking for work. For workers in standard employment relationships this is unlikely to be significant, however for non-permanent workers this may require significant effort. Effort finding work is calculated from a single question asking how much time individuals spend looking for work. Multiple employers/worksites effort measures the effort expended as a result of having multiple employers and working at multiple worksites. It combines eight questions about the number of employers, supervisors and work locations, unpaid time spent traveling between jobs, frequency of working with new sets of co-workers in unfamiliar locations, and conflicts arising from having multiple employers or work locations. Again, these are effort expenditures most likely to be incurred by workers in temporary positions. Constant evaluation effort measures the effort expended by workers trying to increase the probability that their current employer will offer them more work. Where contracts are shortterm in nature, workers may feel they are constantly being evaluated and have to perform at levels beyond that expected of workers in standard employment relationships. A Finnish study revealed that this was one of the negative aspects associated with precarious employment (Pedersen et.al. 2003). Constant evaluation effort includes three questions about the extent to which attitude and performance evaluations affect the length and nature of future work offers. Work in a Global Society, McMaster University page 15

19 Harassment and discrimination effort measures the effort expended as a result to exposure to harassment and discrimination at work. Where employment is temporary and change of employers and locations frequent, the probability of facing harassment and discrimination may increase. Harassment and discrimination effort includes five questions about the frequency of harassment at work, the frequency discrimination is a barrier to getting work or how one is treated at work, the role of favoritism in getting work, and the frequency of being asked to do things unrelated to work. Employment Relationship Support We define three types of employment relationship support; work support, household insecurity and social support. Work support measures the support workers receive at work. It combines four questions about the availability of help with a job, assistance at work if a worker is stressed, the presence of a union and its effectiveness. Household insecurity measures the capacity of an individual to satisfy household economic needs. Low levels of household insecurity may make it easier for a worker to handle low levels of control and high levels of effort associated with their particular employment relationship. It combines 3 questions regarding individual and household earnings, and household benefit coverage (drug, medical, dental, eye, life). Social support measures the support an individual has in the community at large. It combines four questions about whether an individual has access to someone who provides emotional, practical or financial support in a crisis situation, and questions asking if they can draw on the support of friends & family, people in their neighbourhood, or in their community to deal with problems they might face. Employment Strain and Health Outcomes Below we present findings on employment strain and health outcomes based on a first wave of data from workers in standard and precarious employment relationships. Work in a Global Society, McMaster University page 16

20 In the tables that follow, we examine four different clusters of employment relationships; three representing different forms of precarious employment and one representing the standard employment relationship. The temporary agency and short-term contract cluster is made up of workers employed through temporary employment agencies, or who work on short-term contracts, are self-employed or work seasonally. This group is the most representative of the segment of the labour market that has increased dramatically since the early 1980s. The parttime cluster is made up of workers who reported having a permanent part-time job of less than 30 hours per week. The on-call cluster is a class of workers who have an ongoing relationship with an employer, either full-time or part-time, but whose hours vary from week to week based on the employer s needs. The full-time cluster includes workers in permanent positions who work 30 or more hours per week. 5 The characteristics of the sample are reported in Table 1. Interestingly, the two groups with the most dramatic difference in their employment relationships, the temporary agency cluster and the full-time cluster, were demographically very similar in this sample, except that the temporary agency workers were more highly educated, but earned less. The average age of both groups was in the mid-30s with about one-quarter being 25 years old or less. About half were female, twothirds white and just under 90 percent had lived in Canada at least 5 years. They worked in similar sectors of the economy. The temporary cluster had more students than the full-time cluster, but neither included a large percentage of students. They differed significantly in their educational backgrounds. Nearly half of the temporary cluster had some university education compared to 29 percent of the full-time cluster. The temporary cluster was less likely to work in unionized positions than the full-time cluster. There were also significant differences in economic status. Those in the temporary cluster were more likely to earn less than $25,000 a year, live in households with a total income of less than $35,000 and lack any employment benefits compared to the full-time cluster. The part-time cluster had more of the characteristics often referred to in the literature for nonpermanent workers. While the average age of the group was only marginally lower than the Work in a Global Society, McMaster University page 17

21 temporary and full-time clusters, nearly half were 25 or younger. Nearly three-quarters were female and almost one-third were students. They worked the fewest hours on average in the two weeks prior to the survey, were the most likely to earn less than $25,000 a year, but were less likely than the temporary cluster to live in households with a total income of less than $35,000. They were the most racially diverse with almost half being non-white, and were the most likely to be recent immigrants. The on-call cluster differed again. These workers were concentrated in the health sector and were predominantly female. Of the three precarious clusters, they were the oldest, 40 percent were older than 50, the least racially diverse, the least likely to include recent immigrants, the least likely to live in a household with total income of less than $35,000, the least likely to include students, and the most likely to work in a unionized environment. Work in a Global Society, McMaster University page 18

22 Table 1. Socio-Economic Characteristics of Sample by Employment Relationship Types of Precarious Employment Temp agency Part-time On-call Full-time & short-term contracts Average age (years) Female (%) White (%) ********Lived in Canada < years (%)********** Some university (%) Own income <$25,000 (%) Household income <$35,000(%) No employment benefits (%) Full-time student (%) Union member in all workplaces (%) Hours last two weeks Occupation/sector: Factory work (%) Clerical (%) Education (%) Health (%) number of observations Work in a Global Society, McMaster University page 19

23 Table 2 reports our findings about the extent to which employment characteristics of uncertainty, effort and support are experienced differently by workers in different employment relationships. In general, the three precarious clusters reported higher levels of employment relationship uncertainty and employment relationship workload, higher levels of household insecurity and lower levels of work and social support than the full-time cluster. This was particularly true of temporary agency and short-term contract workers. This cluster reported the highest score on the work uncertainty index, reflecting their lack of an ongoing relationship with an employer. The uncertainty surrounding whether they would have more work is reflected in the high degree of uncertainty concerning earnings. Compared to the full-time and part-time clusters, they also reported higher scheduling uncertainty, but not as high as the on-call cluster which reported significantly more scheduling uncertainty than the other three clusters. The on-call cluster s low score on the work uncertainty index reflects the existence of an ongoing relationship with an employer, but a high degree of uncertainty regarding how much work that employer will provide. The three precarious clusters all reported employment relationship workload about 50 percent greater than full-time employees. However, the composition of this workload varied across the precarious clusters. The temporary agency and the part-time clusters expended significant effort getting work and responding to discrimination and harassment at work. The temporary agency and the on-call cluster reported more effort balancing multiple employers/work sites. All three precarious clusters were required to expend more effort as a result of constant evaluation relative to the full-time cluster. All three reported higher levels of household insecurity relative to the full-time cluster. Patterns of work and social support were more diverse. The temporary agency cluster reported less work and social support than the full-time cluster, the part-time cluster the same work support as the full-time cluster and the on-call cluster higher work support and the same social support as the full-time cluster. Work in a Global Society, McMaster University page 20

24 Table 2. Components of Employment Strain by Employment Relationship Index scores range from 0 to 100 where higher scores indicate higher uncertainty and insecurity, more effort and more support. Types of Precarious Employment Temp agency Part-time On-call Full-time & short-term contracts Employment relationship uncertainty Work uncertainty Earnings uncertainty Scheduling uncertainty Employment relationship workload Effort getting work Multiple employers/worksite effort Constant evaluation effort Harassment/discrimination effort Household Insecurity Work Support Social Support Table 3 reports results of our analysis of the relationship between employment relationship characteristics and health outcomes. Each cell represents the change in the odds ratio (the relative probability) of a ten point increase in the relevant employment relationship index controlling for differences in age, sex, physical work environment and prior health status. 6 Numbers greater than one represent increased odds of reporting the relevant health outcome. Work in a Global Society, McMaster University page 21

25 We found that greater employment relationship uncertainty, and in particular greater uncertainties related to scheduling and earnings, were correlated with poorer self-reported health, more frequent tension at work, and more frequent reporting that everything was an effort. A ten percentage point increase in employment relationship uncertainty increased the likelihood by about the same amount that workers reported less than excellent health, tension at work, or that everything was an effort. Greater employment relationship uncertainty was not, however, significantly correlated with reports of working in pain or exhaustion. This supports the argument made above that the effect of the employment relationship on the physical characteristics of work varies and that precarious employment may not systematically be correlated with poorer physical working conditions. In our sample the four employment clusters were not statistically correlated with the frequency of exposure to poor air, noise and temperature extremes and the full-time cluster was more likely than the temporary cluster to report their physical workload was too heavy. It was surprising that the work uncertainty component of employment relationship uncertainty was not correlated with our measures of health status. Increases employment relationship effort were associated with poorer health status. This was particularly true for the effort it takes to handle multiple employers/sites, constant evaluation, and harassment/discrimination. The association was particular strong between employment relationship effort and workers tension at work. A ten percentage point increase in employment relationship effort increased the likelihood that workers would report tension at work by forty percentage points. We did not find a statistically significant relationship between the effort individuals expend finding work and health outcomes. We found that workers with greater household insecurity were more likely to report poorer health status and working in pain. Increases in employment related support were significantly correlated with better health outcomes on all the measures of health status, except the indicator of everything an effort most of the time. Higher social support was correlated with better selfreported health, less exhaustion and lower levels of tension. Work in a Global Society, McMaster University page 22

26 These results suggest that even after correcting for differences in age, sex, race, physical work environment and prior health problems, those characteristics of the employment relationship associated with temporary employment and in particular with temporary agency workers are correlated with poorer health outcomes. When workers have higher levels of employment relationship uncertainty and higher levels of employment relationship workload they are more likely to have poorer health status, particularly tension at work. Workers who have stronger work and social support and less household insecurity are more likely to have better health status. Table 3: Employment Relationship Characteristics and Health Outcomes (Figures represent the change in the odds-ratios associated with a ten point increase in the relevant employment relationship index.) Health less than very good Pain half the time or more Exhausted after work most days Tense half the time or more Everything an effort most of the time Employment relationship uncertainty 1.10 (**) (**) 1.13 (**) Work uncertainty Earnings Uncertainty 1.11 (**) (**) Scheduling Uncertainty 1.05 (**) 1.06 (**) 1.07 (**) 1.06 (**) 1.07 (**) Employment Relationship Effort 1.13 (**) 1.20 (**) 1.28 (***) 1.39 (***) 1.14 (**) Effort Getting Work (**) Effort Multiple Employers/Sites (*) 1.19 (***) 1.21 (***) 1.09 (**) Constant Evaluation Effort 1.04 (*) 1.05 (**) 1.06 (**) 1.08 (**) 1.05 (**) Harassment/Discrimination Effort 1.13 (***) 1.16 (***) 1.15 (***) 1.24 (***) 1.02 Household Insecurity 1.09 (***) 1.10 (**) Work Support 0.82 (***) 0.86 (***) 0.92 (**) 0.80 (***) 0.94 Social Support 0.91 (***) (**) 0.92 (***) 0.95 (*) *** p<.001, **, p<.05, * p<.10; n ranges from model: health index = f(age, sex, white, prior health status, work environment, work index) See endnotes for details. Work in a Global Society, McMaster University page 23

27 Employment Strain and Health Outcomes In this final section of the paper we examine how the interaction between employment relationship uncertainty and employment relationship workload affect health outcomes. Employment strain was defined as having high scores on both the employment relationship uncertainty index and the employment relationship workload index. 7 In our sample, 45 percent of the temporary cluster, 34 percent of the part-time cluster, 30 percent of the on-call cluster and 21 percent of the full-time cluster were exposed to *****employment ****** strain. This suggests that working full-time significantly reduces the risk of being exposed to employment strain relative to temporary agency workers, but does not eliminate it. Table 4 reports the relationship between employment strain and health outcomes correcting for the effects of age, sex, ****race*****, physical work environment and prior health status. In each case, employment strain was associated with poorer health outcomes. These associations were statistically significant in the case of exhausted after work most days and tense half the time or more. In both cases those exposed to employment strain were more than twice as likely to report these two conditions relative to those not exposed to employment strain. Table 4: Employment Strain and Health Outcomes (Figures represent the change in the odds-ratios associated with being exposed to employment strain.) Odds-ratio Health less than very good 1.25 Pain half the time or more 1.27 Exhausted after work most days 2.12 (***) Tense half the time or more 2.30 (***) Everything an effort most of the time 1.50(*) *** p<.001, **, p<.05, * p<.10; n model: health index = f(age, sex, white, prior health status, work environment, ******employment strain*******). Work in a Global Society, McMaster University page 24