Casual Employment in Australia: A Further Analysis
by John Mangan (The University of Queensland) and Christine Williams (Queensland Treasury)*
March 1999
* The views and information contained in this paper are not necessarily the opinions or views of Queensland Treasury or the Queensland Government.
Casual employment is steadily increasing its share of total employment in Australia. This paper analyses some of the contributing factors that have led to this situation by extending the work of Simpson, Dawkins and Madden (1997). The results, while confirming some of their research and clarifying the role of union membership in limiting the spread of casual employment, also show that the determinants of casual employment in Australia are sensitive to the period of estimation and the form of model used.
 |
The past 30 years have seen major changes in working arrangements away from the former dominant model of permanent full-time employment to non-standard forms of employment such as part-time, casual, home-working and contract-based employment. These forms of employment have increased so rapidly in importance that, for some groups such as females, they are as common as full-time work.
Australia, with its tradition of strong labour unions and the presence of a union-supported Labor government for the period 1983-1996, seems an unlikely setting for such large scale changes to employment patterns. However, along with, New Zealand, Australia has been effected over the last decade by government policies designed to improve international competitiveness, including the dismantling of tariffs, the floating of exchange rates and the deregulation of the financial sector. It is thought that these changes, by inducing employers to seek greater labour market flexibility, have promoted the spread of non-standard employment from the demand-side. On the supply-side, there are also groups of workers in the economy particularly responsive to non-standard forms of employment. Irrespective of their determinants, the changes in employment patterns have implications for labour market legislation concerning workplace health and safety, industrial relations and training, as well as for social and family life.
This paper concentrates upon casual employment because this has been identified as the most unstable and precarious form of non-standard employment (Mishel and Bernstein, 1995). To this end, the paper takes as its starting point the work of Simpson, Dawkins and Madden (1997) into the determinants of casual employment and extends their model both temporally and spatially by examining the different experiences of the Australian States. In addition, the paper tests for endogeniety between casual employment and previously identified key variables, such as union membership and non-wage costs, within an instrumental variable approach. This latter innovation proves to be important in correctly identifying the impact of union membership in inhibiting the spread of casual employment.
|
II. Casual Employment in Australia
Simpson, Dawkins and Madden (1997), (hereafter referred to as Simpson et al.) showed that casual employment as a proportion of the total employees grew rapidly over the period 1984-93. The data in Table 1 confirm that this growth has continued into 1997. Currently 26 per cent of employees may be classified as being employed on a casual basis. This is particularly true for females, where almost one third are employed as casuals. In recent years male casual employment has shown a greater percentage increase, albeit from a smaller base.
Simpson et al. showed also that casual employment in Australia displays a distinct industrial distribution. Casual employment is concentrated in Agriculture, Recreation & Personal Services (both with over 40 per cent casuals), Wholesale & Retail Trade (36.2 per cent in 1996) and Construction (34.4 per cent in 1996). The data in Table 2 confirm that, over the period, casual employment as a percentage of employees grew in all industries (except Recreation and Personal Services). Highest growth occurred in Electricity, Gas & Water, and Mining. Noticeable growth occurred also in Manufacturing and, towards the end of the period, Public Administration.
Table I Casual Employment in Australia: per cent of Labour Force
| Year (August) | Sex | ||
| Male | Female | Total | |
 |
|||
| 1988 | 12.0 | 28.4 | 18.9 |
| 1989 | 13.1 | 29.3 | 20.0 |
| 1990 | 12.8 | 28.2 | 19.4 |
| 1991 | 13.0 | 29.0 | 20.3 |
| 1992 | 15.6 | 30.9 | 22.3 |
| 1993 | 16.4 | 30.6 | 22.7 |
| 1994 | 18.1 | 30.8 | 23.7 |
| 1995 | 18.5 | 30.8 | 24.0 |
| 1996 | 21.2 | 32.0 | 26.1 |
| 1997 | 20.9 | 31.7 | 25.8 |
|
|||
| Period % point change | 8.9 | 3.3 | 6.9 |
|
|||
| Source: 1988-1996, Employment Benefits Australia (Cat no. 6334.0), 1997 data from Weekly Earnings of Employees (Cat no. 6310.0, August 1997) | |||
These latter two industries are of particular interest because they have not previously been thought of as major areas for the growth of casual employment. However, casual employment as a percentage of employment in Manufacturing has almost doubled over the period. While Manufacturing is still a less intensive employer of casuals than other industries, the pattern seems to be following that of the United States where non-standard workers are becoming a significant proportion of blue collar workers (Segal and Sullivan 1997). The sudden increase in the percentage of casual workers in public administration after 1995 is also of considerable interest. Simpson et al. noted the higher density of casual employment in the private rather than the public sector for the period up to 1992. Further, it had been argued that the public sector with its relatively higher degree of unionisation and more ingrained equal opportunity and affirmative action program would remain more resistant to the spread of casual employment than the private sector (Mangan 1998). Yet, while the percentage of public administration employees who are employed on a casual basis remains relatively low, the rate of increase observed in recent years may suggest future convergence of rates between the private and public sectors. The drop in percentage employment in Recreation and Personal Services since 1992 reinforces the evidence from the United States that casual employment is highly cyclical (Segal and Sullivan, 1997).
Table II Percentage Casual Employment in Australia, by Industry: 1988-1996
| Industry | Year | ||||||||
| 1988 | 1989 | 1990 | 1991 | 1992 | 1993 | 1994 | 1995 | 1996 | |
|
|||||||||
| Agriculture | 40.0 | 44.3 | 44.6 | 41.7 | 49.1 | 50.0 | 47.6 | 48.0 | 53.6 |
| Mining | 4.8 | 3.2 | 5.2 | 4.1 | 5.5 | 8.8 | 9.3 | 8.7 | 11.7 |
| Manufacturing | 8.7 | 9.3 | 9.1 | 11.1 | 11.9 | 12.9 | 13.3 | 14.6 | 15.0 |
| Electricity | 0.9 | 0.9 | 1.7 | 1.1 | 2.6 | 2.2 | 2.7 | 5.0 | 6.4 |
| Construction | 18.9 | 20.9 | 20.5 | 23.0 | 26.6 | 28.9 | 30.2 | 28.7 | 34.4 |
| Wholesale & Retail | 30.0 | 31.2 | 30.6 | 31.1 | 33.5 | 35.1 | 34.8 | 35.2 | 36.2 |
| Transport & Storage | 11.1 | 12.8 | 13.0 | 14.2 | 16.4 | 15.6 | 16.9 | 19.7 | 19.4 |
| Communication | 2.1 | 4.5 | 2.6 | 6.2 | 4.0 | 6.1 | 7.5 | 9.6 | 8.4 |
| Finance* | 15.0 | 16.2 | 14.4 | 15.5 | 17.4 | 17.9 | 19.8 | 19.8 | 22.7 |
| Public Administration. | 6.2 | 7.2 | 6.9 | 6.6 | 6. 8 | 8.8 | 8.4 | 8.4 | 13.8 |
| Community Services | 17.6 | 17.5 | 16.7 | 16.1 | 18.0 | 16.7 | 18.8 | 18.3 | 19.4 |
| Recreation & Personal Services | 46.5 | 48.0 | 46.9 | 48.1 | 50.6 | 49.3 | 41.6 | 41.2 | 43.9 |
|
|||||||||
| All Industries | 18.9 | 20.0 | 19.4 | 20.3 | 22.3 | 22.7 | 23.7 | 24.0 | 26.1 |
|
|||||||||
| Source: Australian Bureau of Statistics, Weekly Earnings of Employees, Catalogue 6310.0 (Unpublished data) *Includes Property and Business Services | |||||||||
One important aspect of casual employment in Australia not covered by Simpson et al. is its spatial distribution. The data in Table 3 show that casual employment, as a proportion of total employees, is concentrated in Queensland, South Australia and Tasmania. This may be due to the industrial mix factors in these States. For example, Queensland and Tasmania place heavy reliance upon agriculture and tourism. The reasons for the high levels of casual employment in South Australia are less immediately obvious, but industrial relations legislation and practice may also influence the results for all three States.
Table III Casual Employment by State as a Proportion of the Total: 1988 and 1996
| State | Year | ||
| 1988 | 1996 | % point change | |
|
|||
| New South Wales | 17.8 | 24.8 | 7.0 |
| Victoria | 15.6 | 23.6 | 8.0 |
| Queensland | 24.8 | 30.0 | 5.2 |
| Western Australia | 18.2 | 26.3 | 8.1 |
| South Australia | 22.7 | 30.4 | 7.7 |
| Tasmania | 21.9 | 29.3 | 7.4 |
|
|||
| Source: ABS Weekly Earnings of Employees (Distribution) Cat No.6310.0 (unpublished data) ABS Trade Union Members Cat No. 6325.0 (unpublished data) | |||
|
III. The Determinants of Casual Employment
Many view changes in the proportion of the labour force working under casual and other non-standard arrangements as evidence of a fundamental restructuring of the employee-employer relationship. They argue that further growth in casual employment is inevitable as firms replace full-time workers in an effort to build a more flexible work force (Tilly 1991; Thurman and Trah, 1990; Brosnan and Thornthwaite, 1994, Brault, 1997). Implicit in this belief is that demand-side pressures are the driving forces in labour market trends. As a result, the trend has been condemned by many as creating precarious forms of employment, promoting unequal labour force treatment and of being particularly unfair to women (Buchtemann and Quack, 1990; Thurman and Trah, 1990; Tilly, 1992; Brosnan and Thornthwaite, 1994). By contrast, the same trends have been defended as "sound means of reconciling the needs and preferences of workers with the operational requirements of enterprises and benefiting workers with family responsibilities, workers approaching retirement, and other special groups" (Sapsford and Tzannatos, 1994).
Therefore, it is likely that both demand-side and supply-side factors have contributed to the spread of casual employment. As a result, the determination of the predominant causes behind the spread of casual employment becomes an empirical question of major relevance to current social and economic issues, both in Australia and overseas. This is particularly true if the changes in working arrangements are being imposed upon a reluctant workforce by demand factors and institutional change. The concept of forced change seems to be at the heart of most concerns expressed about increased casualisation in the workforce.
These concerns relate to the potential for the destruction of worker protection and union influence (Freeman, 1995; Brault, 1997; Dagg, 1997) and the erosion of gender equity gains as females are pushed into the secondary labour market (Sundstrom, 1993; Rubery, 1994). Other concerns relate to job stability and income distribution (Freeman, 1995; Gregory, 1996). Less concern appears to be expressed where the expansion of casual work results from a supply-side push because this is often seen as an indication of increased labour market democracy and recognition of different work preferences and family responsibilities in the economy (Rubery 1994).
However, such a view may be superficial. Even when the move to casual employment is voluntary, there are a number of issues involved of both individual and national interest. These relate to training, both in terms of opportunity and finance, (Sloan, Carson and Doube, 1992; Mangan, 1998), access to promotion (Romeyn, 1992) and workplace health and safety issues (Dagg, 1997). Finally, to the institutions of Government and those concerned with the implications of labour law and employment policy, such as unions and employer associations, the growth in casualisation raises considerable legal and economic problems (Underhill and Kelly, 1995). For all these reasons, identification of the main determinants of casual employment is important.
Restrictions on the availability of data have meant that most studies into the causes of the spread of casual employment have been at the industry level, with the notable exception of Hawke and Wooden (1998). The most recent of these studies is that by Simpson et al. As part of their analysis they used a pooled regression model to examine the determinants of casual employment in Australia over the period 1984-1992.
They found industry-specific effects, a positive bias towards small firms as employers of casual workers and a negative relationship between union membership and the relative importance of casual employment in the workforce. The union effect is clearly important, as it appears to represent the major current institutional barrier to the spread of casual employment. However, as Hawke and Wooden (1998) conclude when discussing the results of single equation models, "Unfortunately, it is not at all clear that inverse associations between union membership levels and the share of casual employment necessarily reflect causation running from unions to casual employment" (p.19).
The modelling set out below adds to the Simpson et al. model in a number of ways. Firstly the period of investigation is shifted to cover more recent data. Secondly, spatial aspects, in the form of State effects, are specifically included. Thirdly, the relationship between key variables identified in the Simpson et al. model, such as union membership and non-wage costs, are further investigated within an instrumental variable approach.
|
IV. Model Specification and Description of Variables
The nature of the data available for analysis of casual employment in Australia, particularly the relatively short time-series available, favours the use of data pooling techniques that allow the analysis of several years of cross-section data. A number of different models have been proposed for pooled data regression. All these derive from the general regression model but vary according to assumptions made concerning the disturbance term. In most cases the technique used is a one-way error component model of either the fixed effects or random effects type. In this paper a fixed effects model is used, for a number of reasons. The current data set was tested using a Chow test for poolability and no evidence was found that the parameters varied over time. Pooling over regions was, however, strongly rejected by the data.
a) Description of Variables
The chosen variables for the analysis coincide with those used in Simpson et al. (with some minor differences) and also relate back to the discussion in section III of the paper. For example, there are variables reflecting the relative importance of non-wage costs to total labour costs (Non-Wage Costs); the proportion of firms in the industry employing less than 20 employees (Firm Size); the relative importance of private sector workers in the industry workforce (Private Sector); the proportion of employees in the industry that are union members (Union); the proportion of workers aged 20-25 years (Young Workers); the proportion of industry employees aged 55 years and over (Older Workers); and the ratio of employed females to employed males (Sex).
All of these variables have counterparts in the work by Simpson et al. although our Young Workers variable covers the age group 20-25 years while theirs refers to the age group 15-19 years (Simpson et al.; Simpson, 1994). The reasoning for using this particular form of Young Workers variable is that many workers aged 20-25 would have finished post-school training and would be less likely to seek casual work than those aged 15-19 (many of who would be students). The sign and significance for this variable would give some indication as to whether these slightly older workers have the same propensity for casual work as the 15-19 years age group. It may also indicate if casual work is being imposed upon this category of young workers. Finally, the annual rate of industry unemployment was included (Unemployment) as an indication of demand conditions in the industry.
As with most studies in this area, data availability is a problem. Data on casual employment in Australia were available for eight years (from 1988 to 1995) across 12 industry groups and six States, although some variables were not available for the whole period. For example, data on the non-wage cost variable are available only for 1990, 1991, 1992 and 1994, and union membership is only available biannually. Data on the full set of variables were only available for one year, 1991. Included in the explanatory variable matrix is a set of State dummy variables to assess whether there are differences between the States with regard to the proportion of casually employed in the workforce. Industry dummies are also included to test for the presence of unobserved factors that may be industry specific.
It is recognised that there may be some endogeneity problems with two of the above listed variables: Non-Wage Costs and Union. Both variables were tested for endogeneity using a Hausman specification test (Hausman 1978). No simultaneity was found between Non-Wage Costs, however, there was clear indication of a contemporaneous correlation between the Union variable and the error term in the PCE equation. To avoid the potential problem of inconsistency in the parameter estimate for the Union variable, an instrumental variable approach was used to remove this simultaneity. For this reason, two separate sets of results are discussed below, the first referring to the OLS estimation of the casual employment equation, the second relating to the IV-estimated equations.
Table IV Model Results *
| Key Variables | |||||
| Equation 1 | Equation 2 | Equation 3 | Equation 4 | Equation 5 | |
|
|||||
| Non-Wage Costs | 0.386 (1.88) | 0.175 (0.72) | 0.323 (2.51) | Excluded | Excluded |
| Firm Size | -0.034 (-0.37) | 0.025 (0.27) | 0.100 (2.17) | 0.138 (3.12) | 0.141 (3.14) |
| Private Sector Union | 0.212 (2.31) -0.254(-3.28) | 0.254 (2.54) -0.631(-3.33) | 0.205 (4.18) excluded | 0.193 (4.02) -0.081 (-2.32) | 0.203 (4.23) -0.330(-4.40) |
| Young Workers | -0.058 (-0.22) | -0.306 (-1.10) | -0.079 (-0.68) | -0.234 (1.80) | -0.252 (-1.92) |
| Older Workers | 0.634 (2.99) | 1.017 (3.41) | 0.134 (1.59) | 0.065 (0.69) | 0.083 (0.85) |
| Sex | 0.351 (2.46) | Excluded, used in IV | 0.418 (5.20) | 0.306 (3.25) | Excluded, used in IV |
| Unemployment | 0.092 (0.48) | 0.286 (1.43) | -0.087 (-1.14) | 0.079 (1.02) | 0.094 (1.21) |
| State Dummies | |||||
| Queensland | 0.050 (3.40) | 0.057 (3.59) | 0.045 (5.36) | 0.022 (2.88) | 0.027 (3.64) |
| Victoria | 0.002 (0.16) | -0.004 (-0.24) | -0.006 (-0.71) | -0.018 (-2.34) | -0.016 (-2.06) |
| Western Australia | 0.003 (0.22) | 0.013 (0.81) | 0.005 (0.60) | -0.016 (-2.01) | -0.009 (-1.18) |
| South Australia | 0.065 (4.77) | 0.058 (3.92) | 0.049 (6.14) | 0.038 (5.05) | 0.039 (5.13) |
| Tasmania | 0.071 (3.98) | 0.037 (2.33) | 0.036 (4.22) | 0.013 (1.96) | 0.012 (1.75) |
| Industry Dummies | |||||
| Agriculture | 0.185 (2.06) | 0.251 (3.41) | 0.287 (7.72) | 0.355 (7.72) | 0.473 (15.19) |
| Mining | -0.125 (-1.04) | 0.107 (3.25) | -0.242 (-4.41) | -0.112 (-1.87) | 0.049 (0.80) |
| Manufacturing | -0.199 (-1.46) | 0.009 (0.08) | -0.262 (-4.00) | -0.136 (-1.92) | 0.040 (0.64) |
| Electricity Gas & Water | -0.124 (-1.28) | 0.181 (1.31) | -0.235 (-5.48) | -0.075 (-1.44) | 0.141 (2.24) |
| Construction | 0.024 (0.20) | 0.176 (1.35) | -0.062 (-1.10) | 0.022 (0.39) | 0.161 (2.79) |
| Trade | -0.093 (-0.59) | 0.119 (0.88) | -0.148 (-1.81) | -0.012 (-0.14) | 0.171 (2.54) |
| Transport and Storage | -0.044 (-0.41) | 0.191 (1.55) | -0.158 (-3.06) | -0.031 (-0.53) | 0.166 (2.86) |
| Communication | 0.008 (0.10) | 0.406 (2.86) | -0.137 (-4.07) | 0.011 (0.24) | 0.284 (4.89) |
| Finance Property & Business | -0.220 (-1.58) | 0.016 (0.14) | -0.293 (-3.99) | -0.162 (-2.06) | 0.043 (0.77) |
| Public Administration | -0.036 (-0.41) | -0.158 (-4.09) | -0.158 (-4.09) | -0.008 (-0.15) | 0.263 (5.44) |
| Community Services | -0.126 (-0.91) | -0.283 (-4.01) | -0.283 (-4.01) | -0.112 (-1.32) | 0.220 (4.81) |
| Recreation & Personal | 0.052 (0.31) | -0.025 (-0.3) | -0.025 (-0.30) | 0.102 (1.16) | 0.322 (5.08) |
| R2 | 0.972 | 0.965 | 0.939 | 0.942 | 0.945 |
| F | 68.2 (24, 47) | 58.7 (23,48) | 171.2 (24,263) | 186.4 (23,264) | 187.4 (22,242) |
|
|||||
| All Industries | 18.9 | 20.0 | 19.4 | 20.3 | 22.3 |
|
|||||
| * t-statistics in parentheses | |||||
| Equation 1; OLS using full data set for 1991 | |||||
| Equation 2; Instrumental variable estimation of Union variable | |||||
| Equation 3; OLS with Union variable excluded | |||||
| Equation 4; OLS panel estimation with Non-Wage Cost variable excluded | |||||
| Equation 5; Instrumental variable estimation with Non-Wage Cost variable excluded | |||||
| Source: Australian Bureau of Statistics, Weekly Earnings of Employees, Catalogue 6310.0 (Unpublished data) *Includes Property and Business Services | |||||
|
Using OLS estimation on the 72 observations available for the full variable set (only available for 1991), five of the eight key variables tested were found to be significant determinants of the incidence of casual employment and all of the coefficients had the anticipated signs (Equation 1). The results from this equation indicate that the incidence of casual employment within an industry is positively related to increases in the relative importance of non-wage costs, the percentage of private sector employees, the proportion of older employees and the relative proportion of women in the workforce but is inversely related to the proportion of employees who are members of a union. In some cases, however, the performance of some variables is equation specific. For example, the Non-Wage Costs variable loses significance in equation 2, while the Young Worker variable gains significance in equations 4 and 5.
There are significant differences between the incidence of casual employment between the States, with Queensland, South Australia and Tasmania having a higher proportion of casually employed: approximately 5-7 per cent more casually employed in these States than in NSW. Limited differences occur between industries, with only Agriculture having a higher incidence of casual employment after allowing for the differences in all the other variables (18 percentage points above the average level for all industries). Finance Property and Business has a much lower incidence of casual employment than other industries (22 percentage points below the average). The results are shown to be sensitive to the model specification for the key Union variable. With instrumental variable estimation (Equation 2), the coefficient on the Union variable increases to -0.631 (more than doubling from -0.254). This larger coefficient is the consistent estimate of the effect of union membership on casual employment. Most other variables are relatively unaffected by the IV estimation.
To attempt to gain the potential benefit from pooling over time (only available if some variables are excluded), a two-step approach was followed. The variables for which incomplete data are available for all the time periods are Union and Non-Wage Costs, where there is only one year of overlap (1991). However, by excluding alternately the other variable from the analysis, a period of four years is possible for pooling in each instance. This procedure does run the risk of inducing omitted-variable bias, as both variables are significant in the 1991 equation. However, this doesn't seem to occur. A comparison of Equations 1, 3 and 4 shows only minor instability in the estimated coefficients for these variables. Overall the coefficients of most variables are quite robust under both specifications, justifying the two-stage approach followed.
First, the Union variable was excluded from the analysis. In the second stage, the Union variable replaced the Non-Wage Cost variable and the analysis was repeated. As before, the model was estimated twice, the second time using instrumental variable estimation for the Union variable. The coefficients of most variables are quite robust under both specifications, justifying the two-stage approach followed. Allowing for the factors that are industry specific and represented by the industry effects columns in the table 4, the following conclusions can be drawn:
A number of factors appear to promote an increase in the proportion of employees that are hired as casuals. For example, an increase of 10 per cent in the proportion of employees in the private sector will lead to a 2 per cent increase the percentage of casuals. Similarly, a 10 per cent increase in the number of firms in an industry with less than 20 employees will increase the proportion of casual employment by between 1-3 per cent. Finally, both a 10 per cent increase in the proportion of female employees and in the ratio of non-wage costs to total labour costs will lead to an increase in the proportion of casual employees of between 3-4 per cent. Conversely, an increase of 10 per cent in the proportion of employees who are members of a union would lead to a reduction of around 3-6 per cent in the proportion of casual workers.
The industry effects are quite stable between Equations 3 and 4 and are of greater intensity than those showed in Equation 1. This reflects the gain in efficiency from the larger data set obtained through pooling. From Equations 3 and 4, casual employment is seen to be disproportionately high in Agriculture and disproportionately low in Finance, Property and Business. The State effects are robust over the alternative specifications, with three States experiencing a higher than average incidence of casual employment. Allowing for differences in the industrial structure between the States, the proportion of employees that is casually employed is 3 to 7 per cent higher in Queensland, South Australia and Tasmania.
a) Comparisons with Simpson et al.
Not surprisingly, there is some overlap with the results obtained by Simpson et al. but there are also some important differences that suggest that the relationship between casual employment and commonly used explanatory variables may be sensitive to both the period of estimation and the form of the model used. In terms of variable performance, our models were able to obtain significance at the 5 per cent level or above for 7 of the 8 key variables. These are Non-Wage Costs (Equation 1 and 3), Private Sector (Equations 1 to 5), Firm Size (Equations 3 to 5) Older Workers (Equations 1 to 3), Sex (all Equations in which it was used) and Union (all equations in which it was used) and Young Workers (Equations 4 and 5). Simpson et al. report significance only for their Firm Size and Union variables. We also identify strong spatial effects and, through the use of the instrumental variable estimation, obtain a consistent estimate of the Union effect.
Both studies identify similar sized impacts of firm size (a ten per cent increase in small firms inducing up to a 3.0 percentage point change in the proportion of casual employment) to those by Simpson et al. However, the use of the State dummies in the current model may have affected the coefficient size on the Firm Size variable because they picked up some of the differences in firm size that exist among Australian States. Queensland in particular has a disproportionate number of its workforce employed in small firms. The importance of correctly specifying the Union effect is demonstrated by the fact that our consistent estimate of the impact of this variable is significantly higher than that obtained by Simpson et al. The behaviour of the Young Persons variable is also of interest. The decision to use the older 20-25 years age group to represent this variable was based on the belief that this group, many of whom would have completed post-school training, would be much less likely to seek or accept casual employment than those aged 15-19 years. This seems to be supported by the negative sign on the significant coefficients in equations 4 and 5.
There are a number of other reasons why some of the results might differ between the two studies. Firstly, Simpson et al used, as their dependent variable, the ratio of casual to permanent workers in an industry while the current analysis has taken the ratio of casual employment to total employment. However, this would only affect the magnitude of the estimated coefficients, not their sign or significance. More importantly, Simpson et al did not specifically allow for endogeneity among his variables and this is particularly important for variables such as union where clear endogeniety exists. Given our knowledge about the gender distribution of casual employment and the findings here and by Hawke and Wooden (1998), is of interest that Simpson et al do not achieve significance or even the anticipated sign for their gender variable. Furthermore it might have been expected that the private sector variable and the older worker variable would have performed better in the Simpson et al model.
Clearly the time period used in estimation is important because it is likely that the characteristics of casual employment have changed over time. For example, witness the recent upsurge of casual employment in the public service and manufacturing. However, the disparity in the performance of some variables and their apparent sensitivity to minor variations in the period of estimation, the form of the model and the exclusion of some variables, casts some doubt on the efficacy of industry level cross section and pooled models in general to identify the casual variables (and their magnitude) in the spread of casual employment. At this stage a more promising approach may be that by Hawke and Wooden (1998) who have moved away from this type of analysis in favour of micro data provided by the 1990 and 1995 Australian Workplace Industrial Relations Surveys (AWIRS). Firm level survey data and a further analysis of the AWIRS data may be the best means of casting further light onto the causes of casual employment until the results of the ABS Supplementary Survey (August, 1998) and ABS survey into Non-Standard Employment and Superannuation (May, 2000) become available.
|
The data for this project was supplied, from original Australian Bureau of Statistics (ABS) sources, by the Office of the Government Statistician (OGS) of the Queensland Treasury. Thanks are due to James Hinchcliffe and Susan Koch of the OGS and Michael Jones and Ken Clarke of the Australian Bureau of Statistics. Most of the data were taken from supplementary surveys run in conjunction with the Labour Force Australia, survey (ABS cat. 6203.0). The data were required by variable by state by industry and as a result were, in the main, obtained by special request and from unpublished sources. Specifically the data sources are for the relevant years:
Table A2. Data source by variable
|
|||
| Non-Wage Costs | ‘Labour Costs, Australia’ (ABS cat. no. 6348.0) unpublished data | ||
| Firm Size | ‘Small Business in Australia’ (ABS cat. no. 1331.0) | ||
| Private Sector | ‘Employed Wage and Salary Earners’ (ABS cat. no. 6248.0) | ||
| Union | ‘Trade Union Members, Australia’ (ABS cat. no. 6325.0) | ||
| Young Workers | ‘The Labour Force Australia’ unpublished data, microfiche | ||
| Older Workers | ‘The Labour Force Australia’ unpublished data, microfiche | ||
| Sex | ‘DXData-ABS Labour Force Australia’ | ||
| Unemployment | ‘Labour Force Australia’ unpublished data, microfiche | ||
| PCE | ‘Weekly Earnings of Employee (Distribution)’ (ABS cat. no. 6310.0, unpublished data, 1988-1995), ‘Trade Union Members, Australia’, ABS cat. no. 6325.0 (unpublished data) 1996 data | ||
|
|||
|
Brault, S. (1997), The Implications of Non-Standard Employment" in Collective Reflections on the Changing Workplace, Chapter 4, Department of Labour, Canada.
Buchtemann, C., and J. Quack, (1990), "How Precarious is Non-Standard Employment? Evidence from West Germany", Cambridge Journal of Economics, September, pp. 315-329.
Brosnan, P. and L. Thornthwaite, (1994), "Atypical work in Australia: Preliminary Results from a Queensland Study", in R. Callus and M. Schumacher (eds.), Current Research in Industrial Relations, Proceedings of the 8th AIRAANZ Conference.
Dagg, A. (1997), "Worker Representation and Protection in the New Economy", Chapter 5 in Collective Reflections on the Changing Workplace, Department of Labour, Canada.
Dawkins, P. and K. Norris (1990), "Casual Employment in Australia" Australian Bulletin of Labour Vol 16, pp. 156-i73.
Freeman, R. (1995), "Doing it Right: The US Labor Market Responses in the 1980s/1990s" Discussion Paper No. 231, Centre for economic Performance, London School of Economics.
Gregory, R. (1966), "Wage Deregulation, Low Paid Workers and Full Employment" in Sheehan, P.J., Kummick, M. and B.S. Grewal (eds) Dialogues on Australia's Future, Centre for Strategic Economic Studies, Victoria University Melbourne, pp. 81-101.
Gujerati, D.N., (1995), Basic Econometrics, 3rd ed., McGraw-Hill, New York.
Hausman, J.A. (1978), "Specification Tests in Econometrics" Econometrica, 46, pp. 1251-71.
Hawke, A. and M. Wooden (1997), "The Growth in Casual Employment: Evidence from Australian Workplaces", Paper presented to the Australian Labour Market Workshop, Victoria University of Technology, Melbourne 19-20 February 1997 (Version dated January 1998).
Mangan, J. (1988), Non-Standard Employment in Queensland: An Empirical Analysis, Department of Training and Industrial Relations, Brisbane.
Mishel, L. and J. Bernstein (1995), The State of Working America, 1994-95, Economic Policy Institute, New York.
Romeyn, J. (1992), Flexible Working Time: Part-time and Casual Employment, Industrial Relations Research Monograph, No. 1, Sydney.
Rubery, J. (ed.), (1994), Women and the Recession, Routledge and Kegan Paul, London.
Sapsford, D. and Z. Tzannatos (1994), The Economics of the Labour Market, Macmillan, London.
Segal, L.M. and D.G. Sullivan (1997), "The Growth of Temporary Services Work", Journal of Economic Perspectives, Vol. 4, No.3, pp. 121-134.
Simpson, M. (1994), An Analysis of the Characteristics and Growth of Casual Employment in Australia 1984-92, Western Australian Labour Market Research Centre, Curtin University of Technology.
Simpson, M. Dawkins, P. and G. Madden (1997), "Casual Employment in Australia: Incidence and Determinants" Australian Economic Papers, 36(2): December, pp. 194-204.
Sloan, J., Carson, E. and L. Doube (1992), Disadvantaged Job Seekers: Casual, Part-Time and Temporary Work, National Board of Employment, Education and Training, Commissioned Report No. 18, AGPS, Canberra.
Sundstrom, M. (1993), "Part-Time Employment in Sweden: Trends and Equality Effects", Journal of Economic Issues, Vol. 25, No. 1, pp.167-178.
Thurman, J. and J. Trah (1990), "Part-time Work in the International Perspective" International Labour Review, pp. 23-40.
Tilly, C. (1991), "Reasons for the Continuing Growth of Part-time Employment" Monthly Labour Review, March, pp. 167-78.
Tilly, C. (1992), "Dualism in Part-time Employment" Industrial Relations, Vol. 31, No. 2, pp. 330-347.
Underhill, E. and D. Kelly (1995), "Eliminating Traditional Employment:
TroubleShooters Available in the Building and Meat Industries", Journal
of Industrial Relations, Vol. 3 No. 3, pp.398-422.
|
1. They used data from the 1990 and 1995 Australian Workplace Industrial
Relations Survey to conduct a micro-based study.
2. These include a desire to maximise comparison with the results of Simpson
et al., the fact that there are a limited number of industries and the
belief that casual employment patterns are expected to differ between
industries.
In all equations the fixed effects model was tested against both the classical
regression model and the random effects model and found to be the appropriate
choice.
3. In the case of Non-Wage Costs the t-value on its error term was 1.251, which was not statistically different from zero. For the Union variable, the t- value on the included error term was 2.031, which is significantly different from zero at the 5 per cent level.
4. The instruments used were the proportion of women in the work force and the 12 industry dummy variables. The use of these variables as instruments reduces the effective measurement of the (separate) impact of these variables, particularly Sex, on PCE, introducing some collinearity with the instrumented Union variable. However, the parameter estimates of these variables are consistent under OLS estimation.
5. It should be noted that these differences exist after adjustment for the industry effect.
6. No instrumental variable estimation was thus required for this section of the analysis.
|
Last reviewed: Dec 20, 2006, Last modified: Sep 15, 2006
Other languages