Bridges or Traps? Casualisation and Labour Market Transitions in Australia

Abstract

In this article, I re-examine the familiar debate on whether casual jobs represent a ‘bridge’ into permanent employment or a ‘trap’ that keeps workers locked into ongoing casualised work or joblessness. My analysis looks at the labour market destinations of casual workers over time, making use of the Household, Income and Labour Dynamics in Australia (HILDA) survey data for the period 2001 to 2009. The novelty of my approach is twofold. First, I examine an extensive range of individual, locality and job characteristics to assess which of these are most strongly associated with various labour market destinations. Second, I conduct the analysis using longitudinal panel data, in which I make use of random intercepts multinomial logit panel models to estimate various conditional predicted probabilities for these destinations. The findings show that as far as individual characteristics are concerned, age and years in paid employment matter a great deal, while education matters much less. Increasing age leads to worse outcomes, more years in paid employment lead to better outcomes, and increased levels of educational qualification have only a modest link to better outcomes. In regard to locality, the more disadvantaged the area, the more likely that casual jobs will persist, transitions to permanent jobs will diminish and transitions to joblessness will increase. In regard to the jobs themselves, casualisation persists in those industries where casual density is high, where organisations are small, where the work is part-time and where skills development is limited. These findings suggest that systemic influences count for a great deal, while human capital elements count for much less. I conclude that the very nature of casual jobs is itself responsible for perpetuating casualised employment.

Keywords

Casual employment fixed-term employment job characteristics labour market transitions unemployment

Introduction¹

The vast majority of new jobs created in Australia during the 1990s were casual jobs (Borland et al., 2001), a phenomenon that left labour market researchers divided in their assessments of the labour market. Some argued that the growth of casual employment showed that the Australian labour market had become more ‘flexible’, something they regarded as desirable (Wooden, 2001; Wooden and Warren, 2004). Others argued that it represented a growing polarisation in the labour market between ‘good’ jobs – those with permanency – and ‘bad’ jobs. From this perspective, casual jobs were seen as poor-quality jobs, insecure, poorly paid and with few long-term prospects for career advancement (Burgess and Campbell, 1998a, 1998b; Watson et al., 2003). As Chalmers and Waddoups (2007: 2) observed, the growth of casual employment raised the prospect of creating a large pool of ‘second-class industrial citizens’.

Within this debate, an interesting set of metaphors have arisen. While the defenders of labour market casualisation sometimes concede that the jobs are of poor quality, they suggest that they play an important bridging role, providing stepping stones for the unemployed to re-enter the labour market. On the other hand, the critics of casualisation suggest that such bridges are illusory and that most casuals stayed trapped in a cycle of job-churning.

The Australian literature is somewhat ambiguous in arbitrating between these two positions. The early study by Burgess and Campbell (1998a) concluded that for job-seekers, casual jobs did not serve as a bridge. Looking at the mid-1990s’ Australian Bureau of Statisics (ABS) Survey of Employment and Unemployment Patterns (SEUP) data,² Burgess and Campbell (1998a) found that casual jobs did not lead to permanent jobs and they argued that ‘casual employment is just another form of exclusion and precariousness that encompasses unemployment and income deprivation’ (Burgess and Campbell, 1998a: 48). Also using the SEUP data, Chalmers and Kalb (2001) concluded on a more positive note. They examined the length of time it took to transition from unemployment to permanent employment, and whether taking a casual job shortened that time. They concluded that it did for some job-seekers, and that casual jobs might be a ‘promising path’ to permanent jobs for some job-seekers. However, they also noted that there was a large amount of variability in the outcomes, and considerable proportions of job-seekers remained stuck in either unemployment or casual employment.

With access to more recent data – in the form of the Household, Income and Labour Dynamics in Australia (HILDA) survey – a number of researchers have returned to the question. Chalmers and Waddoups (2007) used four waves of HILDA data to apply survival analysis to casual employment. They found that people’s duration in casual jobs was associated with the length of their job tenure and with whether the job was part-time. Their overall judgement on the bridge–trap question was, however, inconclusive.

Also using the HILDA data, as well as survival analysis, Mitchell and Welters concluded in a more negative vein. They showed that structural factors, such as industry location, firm size and locality, played an important role in whether workers found themselves trapped in casual jobs (Mitchell and Welters, 2008). In a later study, which examined duration dependence in casual jobs, the authors concluded that ‘casual employment does lock in workers, which is in line with findings from studies that cannot find conclusive evidence that casual employment functions as a stepping stone towards non-casual employment’ (Welters and Mitchell, 2009: 11).

A different econometric approach, which modelled employment transitions between different labour market states, was undertaken by Buddelmeyer and Wooden (2011), also using the HILDA data. They found more positive results for casual jobs, although this depended on gender. In the case of men, they concluded that workers were ‘better off accepting casual work rather than remaining unemployed’. For women, however, ‘we find that unemployment has the edge over casual employment when it comes to enhancing the probability of permanent employment 1 year onwards’ (Buddelmeyer and Wooden, 2011: 128).

Comparing the different approaches taken by Buddelmeyer and Wooden (2011) vis-a-vis Mitchell and Welters (2008) is particularly illuminating. Buddelmeyer and Wooden (2011) used a series of dynamic, multinomial logit panel models with random intercepts to estimate transition probabilities between various labour market states over adjacent years.³ These states were a set of comprehensive destinations – which included self-employment, unemployment and not in the labour force (NILF), as well as the casual, fixed-term and permanent categories. By comparing all labour market transitions, the authors were able to construct the counterfactual: ‘what would have happened to persons working in non-standard jobs had they been in a different labor market state instead’ (Buddelmeyer and Wooden, 2011: 116). The random intercepts specification allowed them to control for unobserved heterogeneity. As is well known, heterogeneity effects are common in labour market processes. These might be educational, motivational or skill characteristics of the worker or contextual aspects of their location. Some of these can be controlled for explicitly – such as educational attainment – but others are not measurable. Incorporating random intercepts into the modelling allows researchers to control for these unobserved effects.

There is a serious downside to the approach taken by Buddelmeyer and Wooden (2011), one that the studies by Mitchell and Welters explicitly target. While there are some measures of locality included, the majority of the regressors in these models of labour market transitions are individual characteristics: things like educational background, age, years in paid employment, marital status, presence of children and so on. The inclusion of the lagged employment state (and the original employment state) are the only regressors that capture systemic aspects of the labour market situation that are not reducible to these individual characteristics, but they are not explicitly identified, as would be the case were they included as specific regressors. The authors’ preference for this approach is partly philosophical and partly statistical. The perspective behind the Buddelmeyer and Wooden (2011) approach is overwhelmingly supply-side neo-classical economics, a framework that is based on methodological individualism. When it comes to their statistical approach, the authors are restricted in their options because their regressors must be chosen from those common to all labour market states. Important job characteristics are available in the HILDA data, but only for those respondents who were employees at the time of the interview.⁴

By way of contrast, Mitchell and Welters (2008: 5) argue for an analysis that incorporates both individual and systemic influences, an approach that takes account of local labour market conditions and the level of macroeconomic activity. They are able to do this because their philosophical perspective alerts them to the wider structural settings in which labour market outcomes occur, and because their method is based on survival analysis for those currently employed in casual jobs. They thus have access to a wide range of job characteristics from which to fashion their regressors. The downside to their approach, inherent in using survival analysis, is that they can only model non-casual outcomes as a single category, that is, as an exit from casual employment.

In the analysis that follows, I pursue the emphasis on systemic influences but I also consider all possible labour market outcomes. In this respect, my approach ‘bridges’ these two divergent methodologies. Like Buddelmeyer and Wooden (2011), I estimate transition probabilities using multinomial logit panel models with random intercepts. While I examine all possible labour market outcomes, the subjects for this analysis are those individuals currently working in casual jobs. In this way, like Mitchell and Welters, I am able to draw upon a wider range of systemic influences in choosing my regressors, particularly the characteristics of the casual jobs. Unlike Buddelmeyer and Wooden (2011), I do not model all labour market transitions since I do not examine how individuals who are unemployed, permanent employees or self-employed fare. In this respect, I am not considering the counterfactual: how the same person might have fared had they been a permanent worker, for example, instead of a casual.

The questions this analysis asks is thus: In what labour market situation does a male/female casual/fixed-term worker find themselves in the following year? How does this relate to their demographic characteristics (age, education, years in paid employment, health), to the locality where they live (the unemployment rate, the socio-economic characteristics) and to the casual or fixed-term job itself (hours, pay, industry, organisational size)? Many of the regressors used for this analysis are common in most labour market studies, but the richness of the HILDA data also allow for some quite unique variables to be included. These include the effects of social support networks and the skills opportunities that jobs offer. Most importantly, the HILDA data allow the researcher to distinguish between casual and fixed-term employees, and this proves to be a fundamental distinction in this subject area.

Data and Analysis

The HILDA survey is a household-based longitudinal survey covering a broad range of social and economic questions, which has been conducted annually since 2001 (for more details, see http://www.melbourneinstitute.com/hilda/). Respondents aged 15 or over living in the sampled households are surveyed each year (called a ‘wave’), generally in the latter half of the year, and respond to both interviewer-administered questionnaires and a self-completion questionnaire. There are a set of core questions that remain the same every year, thereby allowing for a valuable accumulation of consistent data on the same individual over time. New subjects are recruited into the survey from two sources: existing members of a household may turn 15, or new members may enter a household (e.g. through marriage).

The data for this analysis come from nine waves of the HILDA survey, spanning the period 2001 to 2009. I work with four subsets of the data: male and female casual employees; and male and female employees on fixed-term contracts. While the categories of casual and fixed-term employee are often merged in labour market studies – due to a reliance on the ABS definition of a casual, which is based on leave entitlements – it is possible with the HILDA data to separate the two categories because a question is included that explicitly asks interviewees how they are employed. Research over the last decade using this distinction has emphasised its importance, with the situation of fixed-term employees being quite different from that of casuals (see e.g. Wooden and Warren, 2003). An obvious, and very important, difference is that fixed-term employees are dominated by management and professional occupations, while casual jobs are dominated by sales and labouring occupations.

A further restriction on the population studied here is that the age range of the subjects spans 15 to 64 and excludes full-time students in the current year and in the subsequent year. The exclusion of students is crucial, since a considerable proportion of casual jobs are held by students whose working situation usually changes abruptly once they graduate. A casual job in hospitality, for example, is usually very transitory for a full-time student studying accountancy or teaching.

A person’s current labour market state – either casual or fixed-term – is the basis for defining each population, and the regressors are ones that are available for that current situation. The full list of regressors is shown in the tables in the Appendix. The outcome variable is the labour market state in the following year. This is composed of six categories: permanent, casual, fixed-term, self-employment, unemployment and NILF.

The use of a lead-variable (i.e. the situation the following year) reduces the sample to eight waves of data, and the other restrictions mentioned earlier further reduce the sample size: 2731 observations for male casuals; 4725 for female casuals; 1849 for male fixed-term employees; and 2008 for female fixed-term employees. Transition outcomes are not normally distributed but follow an extreme value type 1 (EV1) distribution, which makes fitting a multinomial logit model (MNL) the appropriate estimation strategy.⁵

With longitudinal data, such as the HILDA survey panel data, the modelling needs to accommodate repeated observations on the same individual. The appropriate model for this is a random intercepts MNL model in which the probability of observing an outcome j is conditional on observed characteristics X _it and unobserved individual effects a_i. The former vary over time and between individuals; the latter vary between individuals, but are time invariant. The notation for this model (Haan and Uhlendorff, 2006: 230) is as follows:

\Pr (j | X_{it,} α_{i}) = \frac{\exp (X_{it} β_{j} + α_{ij})}{\sum_{k = 1}^{J} \exp (X_{it} β_{j} + α_{ij})}

Here, j represents one of the possible outcomes, i is the individual and t represents the time period, that is, the wave in which the individual is observed. In the analysis for this article, j is actually j_i + 1 and reflects the fact that the outcome is for the following year. The unit of analysis is an ‘occasion’, which is nested within an individual person. The unobserved individual effects, α _i , can be modelled as random intercepts and while they do not (by definition) have parameters, their variability can be estimated (this is shown as the standard deviation of the random intercept in the modelling results in the Appendix).

Models such as these are referred to as mixed MNL models or multi-level MNL models, depending on the discipline (Gelman and Hill, 2007; Pinheiro and Bates, 2004; Skrondal and Rabe-Hesketh, 2004), and they require particular estimation procedures. For this analysis, maximum simulated likelihood (MSL) estimation is used.⁶ When it comes to interpretation, the MNL coefficients for each of the observed characteristics, that is, the covariates x_it for each of the J - 1 outcomes, can be presented as raw estimates or as relative risk ratios (RRRs). The tables in the Appendix show the raw estimates, but for ease of interpretation predicted probabilities are much more intuitive. A common presentation device is to set all the values of the regressors, apart from the variable of interest, to their mean value, and to allow the variable of interest to alternate between set values. There are two common methods in using this approach: predictions at the mean and mean predictions (sometimes called ‘the method of recycled predictions’).⁷ The latter approach is taken in this article.

Results

The unconditional probabilities for each of the labour market destinations for the four populations are shown in Table 1. The destinations are for the following year, and are shown in the vertical rows. The percentages shown here suggest that duration dependence – that is, being stuck in the same situation – is very high for casuals but weaker for fixed-term employees. Among casuals, nearly half of males, and more than half of females, remain casuals the following year. In the case of fixed-term employees, the fraction who remain fixed-term is closer to two-fifths. Fixed-term employees have a much better chance of becoming permanents: 48% for male fixed-term employees and 44% for females. By contrast, among casuals, the proportions who become permanents are just 28% and 21%. These are, nevertheless, higher proportions than those who become jobless: 13% of male casuals end up either unemployed or outside the labour force; the equivalent figure for females is 15%, with most of these leaving the labour force.

Table 1

Unconditional transition probabilities: Destinations in following year for each population

	Casuals		Fixed-term
	Male	Female	Male	Female
Permanent	28	21	48	44
Casual	48	54	5	8
Fixed-term	5	5	39	38
Self-employed	6	4	4	3
Unemployed	6	3	2	2
NILF	7	12	3	5
Total	100	100	100	100
Sample size	2792	4815	2006	2118

Notes: Unweighted data. All waves of data. Includes repeated observations. Note that the sample sizes for estimation are slightly smaller than these numbers because of missing observations for some of the covariates.

One can see why researchers regard the bridge–trap debate as inconclusive. On the one hand, permanent destinations outweigh jobless destinations, particularly for male casuals. On the other hand, poor labour market outcomes – in the form of remaining casual or becoming jobless – considerably outweigh good labour market outcomes. However, if the purpose of the research exercise is more than just drawing up a crude balance sheet, then these unconditional probabilities are not very informative in themselves. If the research goal is to actually understand the dynamics, and the generative mechanisms, within casual labour markets, then conditional probabilities are what really matter. We need to know not only which individuals – in terms of personal characteristics – stay locked in casual employment, but what kinds of jobs and what kinds of localities consistently reproduce casualised work.

In this respect, the most important findings about individuals from this analysis are that age and years in paid employment matter a great deal, while education matters much less. Increasing age leads to worse outcomes, more years in paid employment lead to better outcomes, and increased levels of educational qualification have only a modest link to better outcomes. In regard to locality, the more disadvantaged the area, the more likely that casual jobs will persist, transitions to permanent jobs will diminish and transitions to joblessness will increase. In regard to the jobs themselves, casualisation persists in those industries where casual density is high, where organisations are small, where the work is part-time and where skills development is limited. In summary, systemic influences count for a great deal, while human capital elements count for much less.

In the discussion that follows, I often refer to ‘joblessness’ as an outcome, a categorisation where unemployment and NILF are lumped together. While for women the NILF category can be a unique destination given the gendered nature of unpaid domestic labour and caring work, for men in the working-age population used in this study (keeping in mind the exclusion of full-time students) the NILF category often masks hidden unemployment or forced early retirement. In this respect, this category of ‘jobless’ is quite a reasonable measure of the lack of employment opportunities for this population.

Age, Years in Paid Employment and Education

The effect of age is shown as a series of line plots in Figure 1. All present the same sobering story that movement into permanent jobs falls with age, particularly once workers reach their mid-40 s. For male casuals, the fall (as a trend line) is modest until the mid-40s, but then drops sharply. For female casuals, it is a steady downhill slide from their 20s.

Figure 1

Predicted probabilities of labour market status for age groups (%)

Male fixed-term workers fare somewhat better, with the fall (again, as a trend line) quite slight until their mid-40s, but then a sharp drop sets in. Female fixed-term workers resemble their casual counterparts in that the downhill slide (as a trend line) is steadily downward from their 20s onwards.

The other destinations show considerable variation. For male casuals, casual destinations continue to rise with age, right through into their 50s. Unemployment rises during their late 40s, but it is movement outside the labour force that takes off dramatically when male casuals enter their 50s. For female casuals, casual destinations stop rising after they reach their 40s, and the same pattern as for men is evident with the NILF outcomes.

Male fixed-term workers are inclined to stay in that labour market state over the life course, with no (trend) decline evident. This is not the case for women, whose fixed-term job destinations begin to decline once they reach their 50s. While male fixed-term workers have virtually no movement into casual jobs, for female fixed-term workers this destination actually increases towards the end of their working lives.

The results for years of paid employment are also shown as a series of line plots in Figure 2.⁸ With the exception of male fixed-term employees, these plots show a steady increase in permanent destinations for those workers with longer years of paid employment behind them. They also show some other interesting variations. For male casuals, casual employment falls as a likely destination and self-employment becomes much more likely. Jobless outcomes also decline for workers with a longer history of paid employment, though these effects are confined to casuals.

Figure 2

Predicted probabilities of labour market status for years in paid employment (%)

Among male casuals, job tenure has no impact on their employment destinations, or on their prospects of avoiding joblessness (Table 2). For women, however, the impact is considerable. Longer job tenure in their current casual job considerably increases their prospects of employment and reduces their prospects of being unemployed. But when it comes to staying employed, job tenure actually increases the prospects of remaining casual rather than moving into permanent employment. Whereas a female casual with one year’s job tenure has odds of about 1.3 (23% to 17%) of staying casual rather than becoming permanent, once that job tenure stretches out to four years, the odds have more than doubled (51% to 24%).

Table 2

Predicted probabilities of labour market status by job tenure for casuals (%)

	Male					Female
	Under one year	One yr	Two yrs	Three yrs	Four yrs	Under one year	One yr	Two yrs	Three yrs	Four yrs
Permanent	30	30	30	30	30	11	17	22	24	24
Casual	44	44	44	44	44	13	23	35	44	51
Fixed-term	6	6	6	6	6	4	5	6	6	6
Self-employed	7	7	7	7	7	2	3	4	4	5
Unemployed	6	6	6	5	5	63	40	21	8	3

Notes: Because of the nature of their contracts, job tenure is not included as a regressor for fixed-term employees.

The results for educational qualifications are shown in Table 3 with the top panel showing the labour market destinations for those currently employed as casuals, and the bottom panel showing destinations for those who are currently fixed-term. For male casuals, a degree does indeed confer an advantage in attaining permanency, particularly vis-a-vis early school-leavers. But the advantage is slight if the comparison is with those holding Certificates III/IV. Moreover, holders of a diploma (or advanced diploma) are no better off than early school-leavers. Compared with others kinds of qualifications, degree-holding does make it more likely that male casuals will move on to fixed-term jobs. Finally, degree-holding does make it less likely that male casuals end up jobless.

Table 3

Predicted probabilities of labour market status by highest educational level (%)

	Male					Female
	Degree^a	Dip/Adv Dip	Yr12	Cert III/IV	Yr11^b	Degree^a	Dip/Adv Dip	Yr12	Cert III/IV	Yr11^b
Casuals
Permanent	36	26	30	32	27	22	20	26	24	24
Casual	37	41	46	41	48	45	50	50	50	52
Fixed-term	12	6	7	5	4	10	9	4	6	4
Self-employed	6	11	5	10	4	6	6	5	5	3
Unemployed	4	8	4	5	7	3	2	4	3	4
NILF	6	9	9	8	8	14	12	11	12	13
Fixed-term
Permanent	43	50	60	59	55	47	50	54	52	50
Casual	4	5	2	4	8	7	10	8	9	9
Fixed-term	46	31	29	28	25	36	28	29	31	30
Self-employed	5	5	3	4	4	3	4	3	3	2
Unemployed	1	2	2	2	2	2	3	2	2	1
NILF	1	6	3	3	7	5	5	5	3	7

Notes: ^aIncludes those with post-graduate degrees. ^bIncludes Certificate I/II and those with less than Year 11.

In the case of female casuals, the results are much weaker. Degrees do not confer any advantage in attaining permanency, although they do make it slightly more likely that incumbents will move on to fixed-term employment. There is no association between degrees and destinations outside the labour force.

For workers on fixed-term contracts, the results are similar. The best prospects for permanency are found among Year 12 graduates and Certificate III/IV-holders, rather than those with higher qualifications. Among males, degree-holders are just as likely to stay fixed-term as to gain permanency, although for females, permanency is more likely than continuing as fixed-term. It is important to keep in mind that many fixed-term employees are working as professionals, so the association between continuity and degree-holding is not particularly informative.

Job Characteristics

Industry has been defined according to its casual or fixed-term density. The notes to Table 4 show which industry divisions have been allocated to which category, with the general rule being that low density refers to below-average levels of casualisation/fixed-term employment, moderate refers to about average and high refers to considerably above average. Table 4 shows that those industries with a high density have the worst outcomes for permanency, particularly for fixed-term employees. The likelihood of staying a casual increases steadily with density for males and jumps suddenly for females in high-density industries. Among male fixed-term employees, remaining in that category increases as one moves from low to moderate density, while for female fixed-term employees, the jump is again from moderate density to high density.

Table 4

Predicted probabilities of labour market status by industry density (%)

	Male			Female
	Low density^a	Moderate density^b	High density^c	Low density^d	Moderate density^e	High density^f
Casuals
Permanent	32	29	27	24	25	22
Casual	40	45	49	49	46	54
Fixed-term	6	6	6	7	6	4
Self-employed	7	6	6	4	6	4
Unemployed	6	6	6	3	4	3
NILF	9	8	7	12	13	13
Fixed-term
Permanent	56	48	50	52	54	43
Casual	4	5	6	8	7	9
Fixed-term	30	37	37	29	28	39
Self-employed	5	4	2	4	3	2
Unemployed	2	2	2	3	2	1
NILF	3	4	3	4	5	5

Notes: ^aDefined as: Mining; Manufacturing; Electricity, Gas, Water and Waste Services; Construction; Wholesale Trade; Information Media and Telecommunications; Financial and Insurance Services; Rental, Hiring and Real Estate Services; Professional, Scientific and Technical Services; Public Administration and Safety; Other Services.

Defined as: Agriculture, Forestry and Fishing; Transport, Postal and Warehousing; Education and Training; Health Care and Social Assistance.

Defined as: Retail Trade; Accommodation and Food Services; Administrative and Support Services; Arts and Recreation Services.

Defined as: Agriculture, Forestry and Fishing; Mining; Electricity, Gas, Water and Waste Services; Construction; Financial and Insurance Services; Professional, Scientific and Technical Services; Public Administration and Safety; Education and Training; Health Care and Social Assistance.

Defined as: Manufacturing; Wholesale Trade; Transport, Postal and Warehousing; Information Media and Telecommunications; Rental, Hiring and Real Estate Services; Administrative and Support Services; Other Services.

Defined as: Retail Trade; Accommodation and Food Services; Arts and Recreation Services.

The results for earnings quintiles suggest little variation in outcome. When it comes to organisational size, the results suggest that male casuals have better prospects for permanency if they work for large organisations. However, among female casuals, their destination patterns do not differ according to organisational size. On the other hand, among female fixed-term employees, being employed in a large organisation does favour permanency.

The results for hours of work are dramatic. As Table 5 shows, the prospects of a full-time casual gaining permanent employment in the following year are nearly 10 percentage points higher than for a part-time casual. This applies to both men and women. Among male fixed-term employees, the advantage conferred on full-timers is even higher: 17 percentage points. Vulnerability to subsequent joblessness is also higher among part-time casuals, particularly men. Some 17% of male part-time casuals face this prospect compared to 11% of male full-time casuals.

Table 5

Predicted probabilities of labour market status by hours of work (%)

	Casuals				Fixed-term
	Male		Female		Male		Female
	Full-time	Part-time	Full-time	Part-time	Full-time	Part-time	Full-time	Part-time
Permanent	34	25	31	22	53	36	52	43
Casual	42	46	45	51	4	14	6	13
Fixed-term	7	5	7	6	35	32	34	32
Self-employed	6	8	5	5	4	10	3	3
Unemployed	6	6	3	4	2	5	2	2
NILF	5	11	9	13	3	4	4	6

One of the more common criticisms levelled at casual jobs is that they are often ‘dead-end’ jobs. While not necessarily boring or repetitive, ‘dead-end’ jobs lead nowhere because they offer no prospects for a worker to enlarge their capacities. One useful measure of this is the question in the HILDA self-completion questionnaire that asks respondents about their opportunity to learn new skills in a job. For male casuals, the skills content of the job does have implications for its incumbent. As one moves along the scale measuring this potential, the probability of attaining permanency in the following year rises steadily. At the same time, the probability of staying in a casual job, or ending up outside the labour force, also declines. The results are similar for women, but weaker in strength. By way of contrast, the potential skills of fixed-term jobs are largely irrelevant because fixed-term jobs are already relatively high in skills content.

Aspects of Locality

In addition to the characteristics of the job, an individual’s locality also makes a difference. Areas with higher unemployment rates provide fewer employment opportunities for local residents. Such areas are also characterised by greater levels of social disadvantage in a broader sense. The higher an area’s unemployment rate (on a standardised scale), the worse are the prospects of gaining permanent employment. This applies to both males and females, and to both casuals and fixed-term employees. Instead, staying a casual or staying fixed-term is much more likely in these areas.

A more direct measure of social disadvantage can be found in the ‘socio-economic indicators for areas’ (SEIFA) indices, which measure the economic resources of households at an area level (things like income, expenditure, assets, dwelling size).⁹ These indices are also associated with labour market outcomes, although these effects are almost exclusively confined to casuals. As one moves to higher levels of the SEIFA index (again on a standardised scale), the probability of staying in a casual job drops and the probability of moving into a permanent job increases. This association is stronger for women than for men in terms of moving to permanency, but stronger for men than for women in terms of escaping casual jobs.

The social support networks in which people live also shape their labour market prospects.¹⁰ This can happen at a personal level, in the sense that support and encouragement can assist with confidence. It can also happen in practical ways, in that job openings are mediated through personal networks. In a more general sense, such networks are also an indicator of the depth of social capital in neighbourhoods. The modelling showed that male casuals, in particular, benefit from social support networks, with their probability of moving to permanent jobs being higher with greater degrees of social support. Their likelihood of remaining casual, or becoming jobless, also declines with more social support. For female casuals, the effect is much weaker, as it is with female fixed-term employees.

Cumulative Effects

Many of the factors considered in the preceding exposition do not operate in isolation. While regression analysis is useful for gauging the net effect of a particular factor of interest, in practice, the situation is more likely to be cumulative. This can be illustrated with cameos, where a number of factors are combined in their more likely combinations and a combined probability calculated. This is done in Table 6, which illustrates the impact of locality and job characteristics. It does this by combining the three measures of locality just discussed and highlighting the difference by contrasting ‘unfavourable’ and ‘favourable’ combinations of factors; in other words, comparing a locality with a high unemployment rate, low SEIFA score and low social support networks with a locality with the opposite characteristics. In the real world, the contrast will not be this stark, but the contrast illustrated here shows what the ‘outer boundaries’ are likely to be. The same contrast is done with job characteristics: a job with part-time hours, in a small organisation, at the lowest level of pay and with low opportunities for skill is contrasted with its opposite.

Table 6

Predicted probabilities of labour market status for contrasting cameos (%)

	Casuals				Fixed-term
	Male		Female		Male		Female
	Unfav	Fav	Unfav	Fav	Unfav	Fav	Unfav	Fav
Locality
Permanent	21	36	13	34	44	57	35	58
Casual	55	32	58	41	3	5	19	4
Fixed-term	3	10	4	6	33	33	36	27
Self-employed	3	13	4	6	8	2	6	2
Unemployed	8	3	7	2	3	1	1	2
NILF	11	6	14	11	9	1	3	6
Job characteristics
Permanent	16	37	14	34	30	53	34	48
Casual	48	42	54	43	28	2	20	7
Fixed-term	4	8	1	11	22	35	29	33
Self-employed	11	4	8	4	15	4	11	4
Unemployed	7	4	4	2	1	1	1	3
NILF	14	4	18	7	4	4	5	4

Notes: Unfav – unfavourable combination of factors; fav – favourable combination of factors. For locality, unfavourable means high unemployment rate, low SEIFA index and low social support score; favourable means the opposite. For job characteristics, unfavourable means part-time hours, working in a small organisation, being in the bottom earnings quintile and having the lowest opportunity to learn new skills; favourable means the opposite.

As the top panel in Table 6 shows, the prospects for permanency among male casuals jumps from 21% to 36% as one moves from an ‘unfavourable’ to ‘favourable’ locality, and among female casuals the increase is even greater, from 13% to 34%. Not only are prospects for casualisation greater in the ‘unfavourable’ localities, but joblessness is also much more likely: 19% for male casuals and 21% for female casuals. The equivalent figures are about half this in the ‘favourable’ localities. Fixed-term employment departs from this pattern. While there is a similar contrast in terms of permanency (but weaker in strength), there is no change in the fixed-term outcome among males. Only among women does the fixed-term destination fall as one moves from ‘unfavourable’ to ‘favourable’ localities.

The results of the cameo for job characteristics also illustrate a sharp difference between ‘unfavourable’ and ‘favourable’ combinations. Those male casuals in ‘unfavourable’ jobs have only a 16% probability in the following year of gaining permanency in employment and a 21% probability of ending up jobless. Self-employment – possibly a form of hidden unemployment – is also more likely for this group. By contrast, male casuals in ‘favourable’ jobs have a 37% probability of getting permanent jobs and only an 8% probability of joblessness. The pattern for female casuals in ‘unfavourable’ jobs closely follows that of the male pattern, although with self-employment less likely and remaining in casual jobs somewhat higher.

As with the locality cameo, fixed-term employees also depart from the pattern found with casuals. Certainly, their prospects for permanency increase as one moves from the ‘unfavourable’ to the ‘favourable’ category, but their likelihood of remaining fixed-term in the following year actually increases, whereas among casuals, continuation in that category falls. What seems to be happening is that the other destination categories – ending up in casual jobs or in self-employment – fall away as one moves from the ‘unfavourable’ to the ‘favourable’ combination of job factors. These findings are consistent with the fact that fixed-term employment is dominated by professional and managerial jobs.

Conclusion

These various results defy an easy human capital explanation and suggest a difficult conundrum for conventional analysis. On the one hand, increasing age reduces the prospects of good outcomes, such as permanency, and makes it more likely that casuals will either stay casual or enter joblessness. At the same time, years in paid employment have the opposite effect. Ordinarily, the latter is conceptualised as ‘experience’ in a human capital framework, and is often operationalised by the recourse to age (when no other direct measure is available). Here, they have opposite effects and are not correlated at all. At the same time, education, the other key human capital variable, has an impact on improving good outcomes only for male casuals, and only at certain levels.

Overall, the education results fly in the face of current policy wisdom, which entreats young people to stay in the education system as long as possible. While further education may reduce the prospects of initially entering casual employment – something not analysed in this current research – it only has limited value in helping people escape casual employment.

One explanation for these intriguing results lies in reconceptualising casualised labour markets and recognising that they exist as a secondary labour market in their own right, with their own dynamics and their own internal system of regulation. One unspoken convention in the labour market is that by a certain age, ‘good workers’ will have settled into a career path and their increasing maturity will see them consolidating the advantages of incumbency, such as higher earnings and promotions. But for casual workers, this axiom does not apply: to be in a casual job in one’s mature years signals ‘failure’. As many retrenched workers have found – and the workforce laid off by the clothing maker Bonds exemplifies this – such a judgement may apply even if the current casual job was preceded by decades of permanent employment. In other words, the casual job itself turns age into a liability.

On the other hand, years in paid employment (‘experience’) are definitely an asset. The reason this does not correspond to increasing age is because it actually represents continuity of employment. Extended periods of casual employment usually mean an intermittent labour market history, with periods in and out of joblessness. Such a history makes gaining a permanent job much harder, because the work-based networks that assist such a transition are continually disrupted by such intermittency. Even if the prior employment was in casual work, the continuity makes a difference. Earlier modelling work (not shown in this article) suggested that the lagged-employment state also made a difference to the employment state in the following year. In other words, those casuals who had been employed in the prior year, whether casual or permanent, had better prospects in the following year than those who had been jobless. It is patterns like these that lie behind the adage ‘any job is better than unemployment’, a sentiment that lies behind the welfare-to-work policies of many neoliberal governments. The point that it illustrates, in this analysis, is that continuing attachment to employment is a major asset, but that, by their very nature, casual jobs constantly undermine this attachment. For many workers, particularly women who have left the labour market to undertake parenting, the only prospects for re-entry into jobs is via casual employment, particularly if they seek part-time hours. The intermittency here is based on transitions between such casual jobs and the NILF category, rather than cycling through unemployment. Clearly, greater opportunities for permanent part-time employment would help break this intermittency.

As noted earlier, job tenure had no appreciable influence on the results. In human capital terms, years in employment represent ‘general experience and skills’, whereas job tenure represents ‘firm-specific experience and skills’. Clearly, in casualised labour markets, a worker’s job tenure record may have no value to employers, if value is measured in terms of granting permanency. That such transitions are meant to happen lies behind the notion that casual jobs can provide probationary periods for employees. Yet here we see casuals kept on indefinitely, but with no progression to permanency. They have presumably passed their ‘probation’, but their career prospects have not improved. The most likely explanation lies in the nature of the job: these are the casual jobs that are not intended to ever become permanent jobs. Keeping a reservoir of casual jobs is clearly part of the employment strategies of many firms.

The results for hours of work exemplify the commodification of labour power that is implicit in the casual labour market. Labour market researchers often despair at the layperson’s loose use of language when ‘casual’ and ‘part-time’ are used interchangeably. For researchers, these represent two separate dimensions: mode of engagement and working hours. Yet the layperson’s view is probably closer to the reality that the two are really interchangeable. With the easy availability of permanent part-time work a rarity, anyone seeking part-time hours must usually make do with accepting a casual job. From the employer side, seeking part-time workers generally means seeking casual employees. Not only is this ‘flexible’ employment strategy focused on buying smallish chunks of labour power, but it also aims to buy the ability to turn such labour power on and off with ease.

In terms of the bridge–trap debate, the unconditional probabilities outlined at the beginning of this section suggest that the conclusion drawn depends on how one evaluates the labour market outcomes. The bridge metaphor weighs up permanent outcomes against the avoidance of joblessness: more casuals end up in permanent jobs than jobless. By contrast, the trap metaphor emphasises the patterns of continuing casualisation and intermittent joblessness experienced by most casual workers. In looking at the conditional probabilities, on the other hand, it is clear that the characteristics of casual jobs, in themselves, are a major factor in perpetuating this kind of work. It seems reasonable to conclude that casual jobs do indeed operate as labour market traps, and they are actually crafted to do so.

Footnotes

Acknowledgements

I would like to thank Hielke Buddelmeyer for generously making available his unpublished modelling results and computer code. This article benefited from the feedback of a number of people and I would like to thank: Caroline Alcorso, Grant Belchamber, Murray Goot, Humphrey McQueen, Frank Stilwell and two anonymous referees. An earlier version of this article was presented at the 2012 Centre of Full Employment and Equity (CofFEE) Conference, University of Newcastle, Australia.

Funding

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

Notes

References

Borland

Gregory

Sheehan

(2001) Inequality and economic change. In: Borland

Gregory

Sheehan

(eds) Work Rich: Work Poor: Inequality and Economic Change in Australia, Melbourne: Centre for Strategic Economic Studies, Victoria University.

Buddelmeyer

Wooden

(2011) Transitions out of casual employment: The Australian experience. Industrial Relations 50(1): 109–130.

Burgess

Campbell

(1998a) Casual employment in Australia: Growth characteristics, a bridge or a trap? The Economic and Labour Relations Review 9(1): 31–54.

Burgess

Campbell

(1998b) The nature and dimensions of precarious employment in Australia. Labour and Industry 8(3): 5–22.

Chalmers

Kalb

(2001) Moving from unemployment to permanent employment: Could a casual job accelerate the transition? The Australian Economic Review 34(4): 415–436.

Chalmers J and Waddoups CJ (2007) Is casual employment a ‘bridge’ or a ‘trap’? Conference paper to the HILDA Survey Research Conference, University of Melbourne, 19–20 July.

Gelman

Hill

(2007) Data Analysis Using Regression and Multilevel/Hierarchical Models, Cambridge: Cambridge University Press.

Greene

(2007) NLOGIT Version 4.0, Plainview, NY: Econometrics Software, Inc.

Haan

Uhlendorff

(2006) Estimation of multinomial logit models with unobserved heterogeneity using maximum simulated likelihood. Stata Journal 6(1): 229–245.

10.

Hensher

Rose

Greene

(2005) Applied Choice Analysis: A Primer, New York, NY: Cambridge University Press.

11.

Long

(1997) Regression Models for Categorical and Limited Dependent Variables, Thousand Oaks, CA: SAGE Publications.

12.

Mitchell W and Welters R (2008) Does casual employment provide a ‘stepping stone’ to better work prospects? Working Paper 08-11, University of Newcastle, NSW, Australia, Centre of Full Employment and Equity.

13.

Pinheiro

Bates

(2004) Mixed-effects Models in S and S-Plus, New York, NY: Springer.

14.

Productivity

Commission

(2006) The Role of Non-traditional Work in the Australian Labour Market. Commission Research Paper, Melbourne: Productivity Commission.

15.

R Development Core Team (2011) R: A Language and Environment for Statistical Computing. Austria: R Foundation for Statistical Computing Vienna. Available at: http://www.R-project.org/.

16.

Skrondal

Rabe-Hesketh

(2004) Generalized Latent Variable Modeling: Multilevel, Longitudinal, and Structural Equation Models, Boca Raton: Chapman & Hall/CRC.

17.

Watson

Buchanan

Campbell

(2003) Fragmented Futures: New Challenges in Working Life, Sydney: Federation Press.

18.

Welters R and Mitchell W (2009) Locked-in casual employment. Working Paper No. 09-03, University of Newcastle, Australia, Centre of Full Employment and Equity.

19.

Wickham

(2009) ggplot2: Elegant Graphics for Data Analysis, New York, NY: Springer.

20.

Wooden

(2001) Are non-standard jobs sub-standard jobs? Australian Social Monitor 3: 65–70.

21.

Wooden M and Warren D (2003) The characteristics of casual and fixed-term employment: Evidence from the HILDA survey. Working Paper No. 15/03, Melbourne Institute of Applied Economics and Social Research, University of Melbourne.

22.

Wooden

Warren

(2004) Non-standard employment and job satisfaction: Evidence from the HILDA survey. Journal of Industrial Relations 46(3): 275–297.

Bridges or Traps? Casualisation and Labour Market Transitions in Australia

Abstract

Keywords

Introduction 1

Data and Analysis

Results

Age, Years in Paid Employment and Education

Job Characteristics

Aspects of Locality

Cumulative Effects

Conclusion

Footnotes

Acknowledgements

Funding

Notes

References

Introduction¹