Is there a part-time/ full-time pay differential in Australia?

Abstract

The incidence of part-time employment in Australia is high by international standards and the trend towards part-time work is on the rise. In this paper, we show that, on average, part-timers earn significantly less than full-timers. After controlling for human capital characteristics, the adjusted hourly earnings gap between men employed full-time and women employed part-time is equal to 22.5%. When comparing women only, the human capital adjusted part-time/full-time gap is equal to 8.9%. The latter falls to 1.1% when industry and occupation are controlled for, reflecting important segregation effects. These findings are consistent with international evidence. Consistent with other Australian studies, we also find that the part-time pay gap varies with casual status. In the female labour market, our human capital adjusted models show a part-time casual premium equal to 3.6% relative to the base group (women employed full-time on a non-casual basis).

Keywords

Casual gender pay gap industry occupation part-time earnings segmentation working hours

Introduction

In January 2013, part-time employment in Australia (defined as less than 35 hours per week across all jobs) accounted for 29.9% of total employment (Australian Bureau of Statistics (ABS), 2013a and Table 1), which is high when benchmarked internationally. Across the Organisation for Economic Co-operation and Development (OECD) the average share is around 16.5% (OECD, 2012: 239), with countries such as the Netherlands (37.2%), Ireland (25.7%) and the UK (24.6%) at the high end and France (13.6%) and the US (12.6%) at the low end. Notwithstanding the already high incidence of part-time employment in Australia, its share is on the rise as employers increasingly demand more flexible labour and employees combine work with study and parenting (Abhayaratna et al., 2008).

Table 1.

Part-time and full-time employment, growth and shares in Australia, 1980 to 2013.

	Employed – full-time; males	Employed – full-time; females	Employed – full-time; persons	Employed – part-time; males	Employed – part-time; females	Employed – part-time; persons	Employed – total; males	Employed – total; females	Employed – total; persons
Unit	000	000	000	000	000	000	000	000	000
January 1980	3752.0	1467.3	5219.3	214.3	775.8	990.1	3966.3	2243.1	6209.4
January 1990	4249.3	1952.8	6202.1	370.2	1265.9	1636.1	4619.5	3218.7	7838.2
January 2000	4330.6	2198.0	6528.6	637.1	1687.4	2324.5	4967.7	3885.5	8853.1
January 2013	5257.6	2854.8	8112.4	1018.3	2441.9	3460.2	6275.9	5296.6	11572.6
January 1980	60.4	23.6	84.1	3.5	12.5	15.9	63.9	36.1	100.0
January 1990	54.2	24.9	79.1	4.7	16.2	20.9	58.9	41.1	100.0
January 2000	48.9	24.8	73.7	7.2	19.1	26.3	56.1	43.9	100.0
January 2013	45.4	24.7	70.1	8.8	21.1	29.9	54.2	45.8	100.0
Rate of change (%)
1980–1990	13.3	33.1	18.8	72.7	63.2	65.2	16.5	43.5	26.2
1990–2000	1.9	12.6	5.3	72.1	33.3	42.1	7.5	20.7	12.9
2000–2013	21.4	29.9	24.3	59.8	44.7	48.9	26.3	36.3	30.7
Share of new jobs (change as %)
1980–1990	30.5	29.8	60.3	9.6	30.1	39.7	40.1	59.9	100.0
1990–2000	8.0	24.2	32.2	26.3	41.5	67.8	34.3	65.7	100.0
2000–2013	34.1	24.2	58.2	14.0	27.7	41.8	48.1	51.9	100.0

Source: ABS (2013a), Cat. No. 6202002, Table 2.

In recent years, the Australian government has also promoted increased labour market participation through initiatives such as individual flexibility arrangements and the right to request provisions. There is, however, a concern that these sorts of provisions may see a further deterioration of the quality of part-time jobs. Early research by Charlesworth (2012), for example, shows that the new arrangements have seen the weakening of certain employee protections, such as reducing minimum engagements for casual workers and removing the right of part-time workers to predictable and regular rosters of weekly hours. Such findings are consistent with a large body of international research demonstrating the extent of poor quality part-time jobs (e.g. Fagan and Burchell, 2002).

Research on the quality of part-time jobs in Australia has generated mixed findings. There are numerous studies showing instances of poor quality part-time work (e.g. Harley and Whitehouse, 2000; Jefferson and Preston, 2010; Pocock et al., 2004, 2008; Whittard, 2003). The Australian-based research on the part-time/full-time wage gap does, however, suggest a different story. In stark contrast to much international research on the part-time/full-time wage gap, studies for Australia have generally found either a premium (in favour of part-timers) or no significant difference in the hourly earnings of part-timers relative to full-timers (e.g. Booth and Wood, 2008; Rodgers, 2004).

The general finding of a part-time premium and/or no part-time wage penalty in Australia is of particular interest and worthy of further inquiry since it runs counter to earlier Australian research (e.g. Preston, 2003) and much of the published research on this question (e.g. Bardasi and Gornick, 2008; Fernandez-Kranz and Rodriguez-Planas, 2011; Manning and Petrongolo, 2008; Mumford and Smith, 2009; O’Dorchai et al., 2007).

In the remainder of this paper we revisit the question of part-time pay in Australia using data from the 2010 Australia at Work Survey. The particular advantage of this data set is that it allows us to control for a range of human capital variables as well as workplace specific characteristics. Mumford and Smith (2009) have shown the latter to be of particular importance in studies of the wage gap. The paper is structured as follows. In the following section we review the theory and literature on part-time pay relativities; thereafter we present some descriptive statistics on part-time employment in Australia before moving to a discussion of the data and the proposed estimation model and methods. We then present the results and conclude with a summary and discussion.

Theory and empirical findings

From an orthodox economic or supply side perspective a part-time/full-time wage gap may arise where part-time employees have less human capital as reflected in lower levels of experience and/or lower qualifications. Orthodox economics also posits a gap (or compensatory wage differential) if part-timers accept lower wages as a trade-off for other job characteristics such as preferred non-standard hours of work.

On the demand side, a part-time/full-time wage gap may be present where employers are in a monopsony power situation, for example where part-time job opportunities with child-friendly/school-friendly hours are limited. If there are additional fixed costs associated with employing part-timers which are not proportional to the hours that they will work (e.g. training costs, recruitment costs), then this may also give rise to a part-time pay penalty. Alternatively, it may be that part-time employees are comparatively more ‘productive’ than their full-time counterparts if they are only employed for shorter hours in the day, for instance during peak demand periods. In the case of the latter, they may receive a part-time pay premium. In other situations, part-time employees may be able to command an additional loading to compensate for less favourable contract arrangements (e.g. job insecurity). In Australia, the latter is reflected in the loading paid to casual workers, and multivariate research typically shows a significant pay premium attached to casual work (Austen et al., 2008; Booth and Wood, 2008).

From a normative or non-economic perspective there are a number of additional theories to explain observed wage gaps. The sociology literature, for example, emphasizes the historical undervaluation of women’s work. In feminist literature, the focus is also on the undervaluation of women’s work as well as on institutional and normative factors causing women to be segmented and crowded into particular segments (often low paying jobs where the capacity to share in rents is limited) (Bergmann, 1986). A related literature is that on core and periphery jobs. Women are more likely to be found in periphery jobs where the work is undervalued relative to core jobs (Grimshaw and Rubery, 1998).

Feminist writers have also emphasized the constrained choices that women often face. Women’s choices around work and labour market participation are often made in a joint or household setting and are affected by societal norms and expectations around women’s roles. They may also be influenced by the household’s monetary considerations. If, for example, particular work cultures (such as routinely working long hours) cause a gender segmentation of occupations and jobs such that women are, on average, less able to secure the high paying jobs of their male counterparts, then, in the case of a heterosexual couple deciding on work–family schedules, this may see the male partner take on the primary breadwinner role whilst the female becomes the primary care giver. Whilst the women would have ‘chosen’ to reduce their time in paid work, feminists would argue that this is a ‘constrained’ choice (as opposed to a completely free choice) affected by other factors such as societal norms as well as discriminatory factors affecting job segmentation and the remuneration of female jobs.

Furthermore, if there is a shortage of part-time jobs in higher skilled occupations (e.g. perhaps because of discrimination on account of negative employer attitudes to part-timers in certain occupations and industries or because of cost considerations), then persons ‘choosing’ part-time employment may have little option but to downgrade (relative to their skills and qualifications) should they wish to remain in work. Connolly and Gregory’s (2008: F67) research shows that such demand side characteristics have significantly shaped occupational outcomes in the UK and play a much more significant role as determinants of occupational outcomes than labour supply side factors such as personal characteristics and presence of pre-school children. Women managers have, for example, a 60% chance of downgrading should they decide to go part-time (Connolly and Gregory, 2008).

When estimating wages and controlling for numerous factors known to affect earnings, it is therefore particularly important to consider the causal relationships suggested by specific independent variables such as industry and occupation controls. According to Grimshaw and Rubery (2007), women face risks of undervaluation that can be attributed to two main causes. First, they may be paid less than men for the same productive characteristics and second, they may be employed in jobs which are, themselves, undervalued. In a similar vein, Grimshaw and Rubery (2007: 34) note that ‘… adjusting for occupation may not fully reveal the influence of part-time work on pay as some jobs may become effectively associated with part-time working’. In other words, jobs and sectors with a high share of part-time work (and often high share of women) may be marked down or undervalued because of historical undervaluation and systemic discrimination. In such situations it may be that there is no within sector or within occupational part-time/full-time wage gap because full-timers and part-timers are treated equally poorly relative to jobs or sectors which are less feminized. A similar argument is advanced by Austen et al. (2013) in their analysis of contrasting economic approaches to investigating the gender pay gap.

Empirical findings

There is a growing body of literature examining wage gaps within and between different groups of labour market participants. Much of it is based on British data, motivated by a desire to understand the causes of observed high unadjusted part-time pay penalties. In these studies, occupation also shows up as a significant determinant of the gap. Harkness (1996), using data for 1992/1993, for example, found a raw gap part-time/full-time of 20.4% in the female labour market falling to 14% after controlling for differences in human capital of women employed part-time and women employed full-time. In an expanded model controlling for occupation, industry and union membership, the gap fell to as low as 1.5%. Bardasi and Gornick (2008), using the Luxembourg Income Study for 1994/1995 and comparing part-time/full-time gaps of women across six countries (Canada, US, Italy, Germany, Sweden and the UK), like Harkness, observed a raw wage gap of 15.1% in the UK, falling to 10.0% after controlling for differences in human capital characteristics. Once industry and occupation controls were added to the model, the adjusted wage gap was reduced to 1.1%. It is noteworthy that this effect did not hold in the other countries. In other words, whilst their model (which included industry and occupation) could explain 92.8% of the observed female part-time/full-time gap in the UK, it could only explain 20.5% of the gap in Canada, 21.3% of the gap in the US and 8.9% of the gap in Germany.

Occupation and job segregation also showed up as significant determinants of observed part-time/full-time pay gaps in Connolly and Gregory’s (2009) longitudinal study (based on the UK New Earnings Survey Panel Dataset (NESPD) and a sample of women aged 16 in 1975 and 43 in 2001) as well as in the cross-sectional studies of Manning and Petrongolo (2008) (based on the 2001–2003 Labour Force Survey of women) and Mumford and Smith (2009) (using the 2004 British Workplace Employee Relations Survey). In Connolly and Gregory’s (2009) study, the adjusted part-time penalty was 10.6% (associated with a basic wage equation). When more detailed controls were included, the adjusted gap fell to around 2.5%. Connolly and Gregory were able to use the longitudinal nature of their data to study the effect of different types of labour market experience and concluded that the penalty to part-time status ‘… is driven by negligible earnings return to part-time work experience in lower level occupations’ (i87). In Manning and Petrongolo (2008), the raw part-time/full-time gap (amongst women) was 25%, falling to 11.4% after adjusting for basic determinants (such as education and experience). When occupation was controlled the adjusted gap fell to 3.2% (and to 2.4% when using narrow occupational dummies). They found that occupation accounted for about 70% of the explained wage gap.

Mumford and Smith (2009) explicitly allow for the impact of gender segregation at the workplace and occupational levels and find that, relative to part-timers, full-timers are more likely to work in higher paying occupations. They also show that industry of employment is an important determinant of observed gaps, as is the share of females in a workplace. The latter has a significant negative effect on the earnings of all employees (except men employed part-time). Occupational segregation is a significant determinant of the gender wage gap in the part-time labour market, but not in the full-time labour market. Occupational segregation similarly plays a significant role in explaining the part-time/full-time gap between males employed full-time and women employed part-time.

Like Mumford and Smith, Fernandez-Kranz and Rodriguez-Planas (2011) also find job-related characteristics of particular importance when explaining part-time/full-time pay gaps. In their study they employ a longitudinal data set (1996–2006) which is based on Spanish Social Security Records. They restrict their analysis to women and separate the sample into those on permanent contracts and those on fixed term contracts. Amongst permanent employees the part-time/full-time penalty is 8.9%, rising to 14.1% for those on fixed term contracts. Consistent with Connolly and Gregory (2008), the Spanish study shows that the smaller part-time/full-time pay gap amongst permanent workers is related to the fact that part-time permanent employees are more likely to have stayed with high paying firms. In contrast, those on fixed term contracts who switch to part-time work are more likely to be concentrated in low paying industries.

Surprisingly, notwithstanding the high and rising incidence of part-time employment in Australia, only a few studies have studied the part-time wage structure and part-time/full-time wage differentials. Moreover, unlike the British studies where the findings are fairly consistent, the Australian studies generate mixed findings, making it hard to conclude whether there is or is not a part-time wage penalty (or indeed premium).

Australian studies using data from the Income Distribution Surveys for 1990 and 1998 (Preston, 2003) find evidence of a large and significant part-time penalty (relative to full-timers). In 1990, the adjusted part-time penalty (all persons) was equal to 8.9% and by 1998 it had fallen, marginally, to 8.1%. In Rodgers (2004), using data from the 2001 Household Income Labour Dynamics Australia (HILDA) (1st wave) database, there was no evidence of significant part-time/full-time pay differential after controlling for a detailed set of controls. Booth and Wood (2008) also used the HILDA data set, although they undertook a longitudinal analysis using the first four waves of panel data (2001 to 2004) and allow for sample selection effects through controlling for transitions. Like Rodgers, they found no evidence of a part-time penalty and, indeed, observe a significant part-time pay premium. In the absence of controls for industry and occupation, the adjusted part-time premium (relative to full-times) was equal to 10.1% in the female labour market and 15.6% in the male labour market. Once controls for industry and occupation were included, contrary to international studies, they observed a slight increase in the adjusted gaps (to 10.7% and 15.9%, respectively). Their results also showed a part-time casual premium, equal to 5.4% for women (and pay disadvantage of 5.2% if they were men).

Finally, as noted earlier, Austen et al. (2008), also using HILDA (wave 6 for 2006, with a pooled sample of men and women) and, like Rodgers, employing a cross-sectional approach, found a part-time pay premium although it was not significantly different from the full-time rate. The coefficient on the casual dummy variable was, however, statistically significant and, consistent and Booth and Wood (2008), showed a casual premium of 5.5%. Within industries there was no significant difference in the earnings of part-timers and full-timers, with the exception of the retail trade where they observe a statistically significant part-time pay penalty of 6.4%.

As previously indicated, these HILDA-based findings are relatively surprising given earlier studies showing significant part-time penalties (Preston, 2003) and scarring effects of part-time work (Chalmers and Hill, 2007). The aggregate HILDA findings also run counter to the studies of Harley and Whitehouse (2000) and Whittard (2003), showing that, in Australia, part-timers are more likely than full-timers to be in low paid jobs and less likely to have access to training and promotion.

The difficulty associated with studying part-time earnings in Australia is that there are increasingly limited data sets with which to work as very few capture data on hourly earnings (required for any analysis of part-time earnings). The HILDA data set has, therefore, become the primary data set for much labour market analysis in Australia and, whilst clearly a robust and reliable data set, it is always useful to be able to compare studies across different data sources. Accordingly, the analysis which follows revisits the question of a part-time/full-time pay differential using data from the 2010 wave of the Australia At Work survey. Prior to describing the empirical analysis we first present some background data on part-time employment in Australia.

Part-time employment in Australia

As noted earlier, around 30% of the workforce are employed part-time, many of those on a casual basis (see Table 2). Historically, casual employees have been paid a loading to compensate for the absence of sick and holiday pay entitlements. Casual loadings have also been awarded as a form of deterrence against using casual contracts by employers (Campbell and Brosnan, 2005: 36). Notwithstanding these payments, when benchmarked against permanent full-time employees, a sizeable raw or unadjusted casual/full-time wage gap may be observed (equal to 21.6%) (Table 3). This is larger than the observed part-time/full-time gap of 15% observed in the permanent labour market.

Table 2.

Non-managerial employees by form of employment contract, May 2010 (%).

	Full-time permanent or fixed term contract (%)	Full-time casual (%)	Part-time permanent or fixed term contract (%)	Part-time casual (%)	%
Males	34.6	2.2	4.0	7.8	48.5
Females	23.4	1.1	15.3	11.7	51.5
Persons	58.0	3.3	19.3	19.5	100.0

Source: ABS (2011: 3), Cat. No. 6306.

Table 3.

Average hours worked and average hourly earnings of non-managerial employees by type of employee, 2010.

	Permanent full-time		Permanent part-time		Casual employment		PT/FT wage ratio	Casual/FT wage ratio
	Hours worked	Hourly rate ($)	Hours worked	Hourly rate ($)	Hours worked	Hourly rate ($)	%	%
Males	40.1	33.3	21.0	26.6	21.8	26.1	79.9	78.4
Females	37.9	29.7	22.8	27.2	16.3	24	91.6	80.8
Persons	39.2	31.9	22.4	27.1	18.7	25	85.0	78.4

Source: ABS (2011: 1).

Part-time and casual work are not evenly spread across industry sectors within Australia. They are particularly concentrated in sectors such as the retail trade (accounting for 17.4% of all part-timers at January 2013), health care and social assistance (17.4%), accommodation and food services (13.6%), and education and training (9.2%) (Table 4). Together these four industry sectors account for 57.7% of all part-time employees. It is noteworthy that these sectors also have an over-representation of women.

Table 4.

Industry of employment, 2000–2013.

	Share of total employment, persons January 2013			% point change in employment 2000–2013				Share PT employment January 2013			% point change in PT employment 2000–2013			Share of women January 2013
	FT	PT	Total	FT	FT	PT	PT	Men	Women	Per	Men	Women	Per	FT	PT	Per
	%	%	%	Men	Women	Men	Women	%	%	%	%p	%p	%p	%	%	%
Agriculture, forestry and fishing	2.9	2.6	2.8	−32.4	−32.5	−5.2	−20.4	15.7	51.9	27.0	4.0	4.1	4.3	20.6	60.0	31.2
Mining	3.0	0.2	2.2	230.4	208.2	117.7	41.3	1.2	8.4	2.2	−0.6	−8.3	−1.9	12.5	51.5	13.4
Manufacturing	10.1	3.9	8.3	−12.0	−21.3	48.0	2.2	6.8	33.3	13.6	2.7	5.5	3.1	19.7	62.5	25.5
Electricity, gas, water and waste services	1.5	0.3	1.2	77.1	97.4	427.8	213.8	5.2	17.3	7.9	3.4	5.7	4.1	20.1	48.9	22.4
Construction	10.8	4.5	9.0	44.2	96.2	144.0	15.5	9.4	49.6	14.5	3.6	−13.0	1.4	7.4	43.1	12.6
Wholesale trade	4.7	2.5	4.1	9.8	18.2	82.9	19.0	10.0	33.5	17.5	3.7	0.1	2.8	25.8	61.2	32.0
Retail trade	7.6	17.4	10.5	3.4	14.0	38.9	25.2	32.8	60.3	48.2	6.1	2.3	4.5	42.9	70.1	56.0
Accommodation and food services	4.1	13.6	6.9	1.2	14.2	68.5	46.1	47.2	65.9	57.5	12.3	5.7	9.1	44.2	63.2	55.1
Transport, postal and warehousing	5.8	3.5	5.1	27.8	26.7	124.8	33.8	14.9	35.5	19.5	5.8	1.2	4.6	18.1	41.0	22.6
Information, media and telecommunications	2.2	1.2	1.9	−2.4	−7.9	30.7	2.9	10.0	31.6	18.7	2.3	2.4	2.1	33.9	68.0	40.2
Finance and insurance services	4.2	2.0	3.6	25.5	14.0	−12.4	28.8	4.1	26.9	16.1	−1.7	2.3	0.2	45.8	88.0	52.6
Rental, hiring and real estate	2.0	1.3	1.8	41.3	87.3	16.1	18.4	10.7	30.3	20.3	−2.0	−10.4	−5.2	42.8	73.2	49.0
Professional, scientific and technical services	8.8	6.0	8.0	50.1	49.6	82.9	105.1	10.8	36.4	21.9	1.8	6.9	4.4	35.2	72.0	43.3
Administrative and support services	2.9	4.8	3.4	17.3	9.6	82.4	28.9	32.1	48.2	40.3	8.8	4.0	5.8	44.0	60.7	50.7
Public administration and safety	7.3	3.3	6.1	41.2	62.2	28.7	79.5	6.8	25.7	15.4	−0.6	1.9	1.2	40.0	76.0	45.6
Education and training	7.1	9.2	7.7	33.8	41.8	102.9	63.8	19.7	41.8	34.8	5.8	3.5	4.5	60.7	81.9	68.1
Health care and social assistance	9.6	17.4	11.8	60.1	63.1	196.6	67.5	25.5	47.4	42.6	9.9	0.7	2.2	71.5	86.9	78.0
Arts and recreation services	1.4	2.5	1.7	41.6	26.0	52.6	36.8	30.6	56.1	41.6	1.6	2.0	1.2	32.3	58.0	43.0
Other services	4.0	3.9	4.0	3.8	17.8	−13.8	34.9	12.9	49.2	28.3	−2.2	3.4	1.7	30.2	73.9	42.6
Total (ANZSIC06)	100	100	100	20.5	29.7	63.7	41.4	15.8	44.9	29.0	3.7	2.1	3.5	35.0	70.1	45.2

Source: ABS (2013b).

Drawing on a comprehensive descriptive analysis of part-time work in Australia by Abhayaratna et al. (2008), there are other features of part-time work in Australia worthy of comment. Firstly, almost half of Australia’s part-time workforce (in 2006) worked less than 20 hours per week and fewer than 25% of part-timers worked 30–35 hours per week. This high share of short hours of work is similar to countries such as the UK and New Zealand, but different from Japan and the US where almost half work 20–29 hours per week. Abhayaratna et al. (2008: 23) also show that in Australia the majority of part-timers are aged 25–54 years and that the share of part-timers in this category is roughly similar to the UK. When compared with the latter, Australia has a slightly higher share of younger part-timers (15–24 years) and a slightly lower share of older part-timers (55+ years). Within the OECD the US has the highest share of younger part-time workers and a relatively high share of older part-timers too. When age is cross-tabulated with hours Abhayaratna et al. (2008: 25) further show that, relative to the OECD average, the incidence of short hours work is much higher amongst young workers (aged 15–24) and older workers (aged 55–64 and 65+) in Australia. This shift towards part-time work amongst younger workers and older workers is also observed and detailed in Jefferson and Preston (2011).

Data and methodology

In contemporary economic literature a dominant framework for the study of wage determination is the human capital model. The model may be stated algebraically as follows:

ln {\overset{\land}{Y}}_{i} = {\overset{\land}{β}}_{0} + V_{i} {\overset{\land}{β}}_{i}

(1)

where lnY denotes the natural logarithm of earnings, V is a vector of human capital characteristics known to affect wages (e.g. qualifications, labour market experience and demographic characteristics) and

\overset{\land}{β}

is a vector of estimated slope coefficients associated with these characteristics. The latter provides a measure of the rates of return to the characteristics controlled for in the wage equation.

In this study, equation (1) is estimated using data from the fourth wave (conducted in 2010) of the Australia at Work Survey. This is a longitudinal telephone study of the changes in working lives of individual Australians. The first wave was conducted in 2007 with interviews of 8341 people who were in the workforce at March 2006. Information is collected on current employment situation, occupation and industry, forms of employment, labour contract, employment income, working conditions, attitudes towards work and demographic characteristics.

The sample is restricted to men and women who are employed and wage and salary earners. We exclude the self-employed as well as those in the defence and agricultural sectors. We restrict the sample to those individuals with full information on variables used in the models, and use unweighted data in our model estimation. These restrictions reduce the sample to 4252. Just over half (50.3% or n = 2159) were women and around 28% (n = 1217) were employed on a part-time basis.

The dependent variable is the basic hourly wage rate in the respondent’s main job. To calculate this, respondents were asked ‘What is your basic hourly rate in your job?’. Multiple job holders were asked for the basic hourly rate in their main job. For respondents who did not know their hourly rate of pay but who provided either their daily, weekly, fortnightly, monthly or yearly pay, their hourly rate of pay was calculated by dividing this by their paid hours of work.

We are unable to control for the type of hours worked (e.g. weekend or evenings) and, related to this, components in the total wage package (e.g. base wage, bonuses, overtime pay, performance pay, salary sacrificing, etc.). This is a weakness shared with many data sets. The derived hourly wage rate may, therefore, be inflated (e.g. include overtime work or shift premium) or may understate actual remuneration (e.g. if packages include more non-wage benefits such as a car).

The set of independent variables (i.e. the vector V) may be categorized into different groups of human capital controls; contract type and method of pay setting; workplace characteristics; industry and occupation. The group of human capital controls include controls for potential labour market experience (which, in this study, is controlled for through a series of age-related dummy variables (aged <21 (control group); 21–24; 25–34; 35–44; 45–54; 55+)), tenure and its square, and qualifications (year 11 or below (control group); year 12; certificate or diploma; Bachelor's degree and higher). The tenure variable is based on the number of years the respondent has been working for the same employer at the time of being interviewed. Demographic characteristics correlated with human capital investments and productivity are also controlled. They include a dummy variable for the presence of a dependent child under the age of 16; a variable controlling for ethnicity (equal to 1 if a language other than English is spoken at home); and a variable controlling for household type (capturing whether the individual is in a single person household, a dual earner household or a ‘breadwinner’ household (where the respondent is in paid work but their partner is not)). Current student status is included to control for individuals likely to undertake part-time work during their studies.

The set of contract-related variables capture whether or not the individual is on a permanent contract, fixed term contract or casual contract. The method of pay setting controls captures whether or not the individual sees their pay set at a national or industry level through the industrial award wage setting process (control group) or, alternatively, is covered by either a collective agreement or an individual agreement. We also include a control for trade union membership.

Workplace characteristics include controls for firms size (three dummies capturing small (1–19 employees; control group), medium (20–99 employees) and large (100+ employees). A series of geographic controls are also included to capture variations in remuneration across states but also between rural and urban labour markets. We include a control for regional location (equal to 1 if the person resides outside a major city) and a control for state of enumeration. The industry and occupational controls are at the 1 digit level.¹ A control for sector (private, public or not-for profit) is also included.

Method

The estimating approach uses ordinary least squares (OLS) on unweighted data. Standard errors robust to heteroskedasticity are presented using White’s (1980) standard technique. Following previous studies such as Booth and Wood (2008), Harkness (1996) and Mumford and Smith (2009) we begin with a basic model (controlling for the set of human capital controls) and then extend the model through the gradual inclusion of additional controls. In the first instance, we pool the male and female data and employ a simplistic approach to capture the difference in earnings of full-timers and part-timers through the use of a dummy variable (equal to 1 if the respondent is employed part-time, working less than 35 h per week in their main job). The advantage of the simplistic approach is that it offers an expedient way of summarizing the data, which is particularly useful since more sophisticated approaches generate results which are fairly similar in magnitude (e.g. Manning and Petrongolo, 2008 found a part-time/full-time gap (amongst women) of 11.6% using the part-time dummy approach and an adjusted wage gap of 11.4% using the decomposition approach).

Thereafter we use a more sophisticated approach which involves estimating and decomposing separate wage equations using the Blinder (1973)/Oaxaca (1973) approach. At this point we estimate separate wage equations for four groups of employees: males employed full-time (MFT); women employed full-time (WFT); men employed part-time (MPT); and women employed part-time (WPT). This approach allows the parameter estimates to vary by group and allows us to explicitly measure the contribution of individual or groups of variables towards explaining the raw wage gaps. It is through such approaches that studies have been able to demonstrate that occupation, for example, accounts for 70% of the explained wage gap in the UK (Manning and Petrongolo, 2008).

The Blinder/Oaxaca method used in the decomposition of the part-time/full-time earnings gap may be described as follows:

\bar{ln Y_{ft}} = {\overset{\land}{β}}_{0 ft} + {\bar{V}}_{ft} {\overset{\land}{β}}_{ft}

(2)

\bar{ln Y_{pt}} = {\overset{\land}{β}}_{0 pt} + {\bar{V}}_{pt} {\overset{\land}{β}}_{pt}

(3)

where, as before, V denotes a vector of wage determinants that includes educational attainment, years of labour market experience and other determinants known to affect earnings,

\overset{\land}{β}

is a vector of estimated coefficients associated with the wage determinants,

{\overset{\land}{β}}_{0}

is the estimated constant term in the regression model and ‘ft’ and ‘pt’ denote full-timers and part-timers, respectively. A bar over the variable denotes a sample mean.

The difference in the mean values of the two dependent variables ( $\bar{ln Y_{ft}} - \bar{ln Y_{pt}}$ ) provides a measure of the raw wage gap (RAW). That is, it measures the part-time/full-time wage gap. To measure how much of the gap may be explained or attributed to differences in characteristics (e.g. qualifications, occupational tenure, etc.) between the two groups, Blinder and Oaxaca propose the following decomposition procedure:

\bar{ln Y_{ft}} - \bar{ln Y_{pt}} = ({\bar{V}}_{ft} - {\bar{V}}_{pt}) {\overset{\land}{β}}_{ft} + {\bar{V}}_{pt} ({\overset{\land}{β}}_{ft} - {\overset{\land}{β}}_{pt}) + ({\overset{\land}{β}}_{0 ft} - {\overset{\land}{β}}_{0 pt}) (RAW) (EXPL) (COEF) (UNEXP)

(4)

where the first term on the right-hand side (EXPL) measures the portion of the wage gap attributable to differences in individual characteristics and is often labelled the explained portion (or skill differential). The second term on the right-hand side (COEF) is a measure of the component of the wage gap due to differences in returns to these characteristics. The third term on the right-hand side (UNEXP) measures the difference in the constants (

{\overset{\land}{β}}_{0 ft} - {\overset{\land}{β}}_{0 pt}

) and is a further unexplained portion. Generally COEF + UNEXP is referred to as the unexplained portion of the part-time/full-time wage gap. If there are systematic differences in the traits or characteristics of full-timers and part-timers (e.g. different preferences for full-time work) or in the treatment (e.g. discrimination) this will also be included in this unexplained component. In equation (4) above, the decomposition analysis uses the full-time wage structure as the non-discriminatory norm. In the literature there is some debate as to which wage structure should serve as the non-discriminatory norm when undertaking decompositions. One alternative (as proposed by Cotton, 1988) is to use a weighted average approach. In our results later, we similarly take an average of the wage structures of each group (using equally weighted reference coefficients) as our non-discriminatory wage structure.

In some studies, researchers also seek to minimize the effect of unobservables (omitted variable bias) using a two stage estimation procedure proposed by Heckman (1979). The results using this technique have, however, been mixed and there is a view that the problems introduced to the wage equation using the selectivity bias correction term may be greater than the bias associated with analysis of a non-random sample in the first place (Miller and Rummery, 1991). Fernandez-Kranz and Rodriguez-Planas (2011: 599) and Manning and Petrongolo (2008: F33) make a similar point, noting that, in studies of the part-time wage differential, those employing selectivity techniques first estimate a model on the propensity to work part-time (with children and marital status (but not earnings)) employed as key determinants. They argue that this is a fairly strong assumption and may be no better than the exogeneity assumption that it is supposedly replacing. Moreover, Connolly and Gregory’s (2008) study shows that part-time employment is not a simple selectivity issue with longitudinal data showing that women employed part-time have held high paying full-time jobs (no doubt with the same human capital investments) and are segregated into a narrow set of occupations because of institutional barriers resulting in occupational downgrading. We, therefore, choose not to account for selection using the Heckman (1979) instrumental variable approach and instead control for family composition variables (marital status, children, household type) and employer characteristics (e.g. firm size) directly in our models.

Results and discussion

The results are summarized in Table 5.² They show that the effect of part-time employment status is statistically significant across all five estimated models. In the most simple model, controlling only for demographic and human capital characteristics, the adjusted part-time/full-time wage gap (or penalty) is equal to 13%. When additional controls are included, such as form of contract and workplace characteristics, the part-time pay penalty reduces but remains significant at 11%. Fixed term contractors earn a premium of around 6–8% (across all models estimated), whilst casual workers also earn a premium of around 6% (once industry, sector and occupation are controlled for). This result is consistent with other Australian studies (e.g. Austen et al., 2008; Booth and Wood, 2008) and reflects the contingencies related to these forms of employment.

Table 5.

Pooled (men and women) OLS results, part-time/full-time wage differentials, n = 4252.^a

	Human capital^b	Model ‘a’ + labour contract controls^c	Model ‘b’ + workplace and geographic controls^d	Model ‘c’ + industry and sector controls	Model ‘e’ + occupation controls
Part-time	−0.13 (−10.01)***	−0.11 (−7.77)***	−0.09 (−6.81)***	−0.07 (−5.17)***	−0.04 (−3.21)***
Fixed term contract		0.07 (3.36)***	0.08 (3.69)***	0.06 (2.73)***	0.06 (3.15)***
Casual		0.00 (−0.30)	0.01 (0.88)	0.03 (1.83)*	0.06 (4.03)***
Adjusted R²	0.38	0.41	0.45	0.48	0.56

T-statistics are reported in parentheses and ***, ** and * indicate significance at the 1%, 5% and 10% levels, respectively.

The human capital controls include age, tenure, qualifications, student status and gender (with the coefficient on the female dummy variable equal to −0.10, −0.09, −0.08, −0.08 and −0.07 in models (a), (b), (c), (d) and (e), respectively; all significant at the 1% level).

The labour contract controls include fixed term or casual employment, wage setting mechanism and union membership.

The workplace and geographic controls include workplace size, regional location and state.

With respect to the part-time/full-time gap, it is clear that the inclusion of industry and occupational controls enhances the explanatory power of the model and, simultaneously, soaks up much of the observed pay gap. The results in the final column of Table 5 show that, relative to employees on full-time permanent contracts, their part-time counterparts earn 4% less (significant at the 1% level). These results provide strong evidence in support of the argument that a large component of the observed part-time/full-time pay gap relates to industry and occupation segregation effects. But is it legitimate or appropriate to explain the part-time/full-time pay gap using industry and occupation? If the distribution of part-time employment across industry and occupation was exogenously determined, then the answer would be yes. There is, however, reason to believe that part-time work opportunities are not exogenously determined (i.e. free of discrimination) and are, rather, concentrated in particular low paying industries and occupations. If this is the case the inclusion of industry and occupation will overstate the portion of the part-time/full-time gap that may be explained by the model. We explore this further in our decomposition analysis which is summarized in Table 6.

Table 6.

Part-time/full-time wage gaps decomposed.^a

	WFT/WPT^b	MFT/WPT^c	MFT/MPT^d	WFT/MPT^e
n	2140	2867	2112	1385
Raw wage gap (%)	0.120	0.228	0.279	0.171
Explained share of gap (%-point) using model:
(a) Human capital and demographic controls	0.031 (2.81)***	0.004 (0.33)	0.100 (3.60)***	0.109 (4.47)***
(b) Model ‘a’ + labour contract controls	0.059 (4.43)***	0.043 (3.30)***	0.105 (3.09)***	0.121 (3.94)***
(c) Model ‘b’ + workplace and geographic controls	0.074 (5.35)***	0.059 (4.46)***	0.126 (3.75)***	0.130 (4.30)***
(d) Model ‘c’ + industry	0.087 (5.96)***	0.086 (5.03)***	0.153 (4.33)***	0.155 (4.92)***
(e) Model ‘d’ + occupation	0.109 (7.18)***	0.101 (5.52)***	0.190 (5.32)***	0.185 (5.60)***
Adjusted PT/FT wage gaps using model (%-point):
(a)	0.089	0.225	0.179	0.061
(b)	0.061	0.186	0.174	0.050
(c)	0.046	0.169	0.153	0.040
(d)	0.033	0.143	0.126	0.016
(e)	0.011	0.142	0.089	−0.014

*** and ** indicate significance at the 1% and 5% levels, respectively.

WFT/WPT denotes decomposition for women employed full-time (WFT) vis-a-vis women employed part-time (WPT).

MFT/WPT denotes decomposition for men employed full-time (MFT) vis-a-vis women employed part-time (WPT).

MFT/MPT denotes decomposition for men employed full-time (MFT) vis-a-vis men employed part-time (MPT).

WFT/MPT denotes decomposition for women employed full-time (WFT) vis-a-vis men employed part-time (MPT).

Looking at the first column of results (for WFT vis-a-vis WPT) we see a raw or unadjusted part-time/full-time wage gap of 12 percentage points (Table 6). Of this gap three percentage points (or 26%, calculated as 0.031/0.120) may be explained by differences in the observed human capital endowments of women employed full-time and women employed part-time. From our descriptive statistics we note that almost half of all women employed full-time hold at least a Bachelor's degree, compared to 37% of women employed part-time. If we deduct this ‘explained’ portion the adjusted ‘unexplained’ part-time/full-time wage gap amongst women falls to 8.9%.

If we extend the model to account for different forms of contract (noting, for example, that 57% of women employed part-time are award reliant compared to 45% for women employed full-time) workplace characteristics (such as firm size and union membership) and geographic controls (i.e. we estimate model ‘c’), we are now able to ‘explain’ 61.5% (or 7.4 percentage points) of the 12 percentage point raw part-time/full-time gap. Taking these as ‘legitimate’ explanatory factors, if we deduct this explained portion from the observed raw wage gap we are left with an adjusted unexplained gap (or part-time penalty) of 4.6%.

As previously observed, part-time work is highly segregated by industry and occupation and inclusion of these variables soaks up much of the unexplained residual. In our extended model with industry and occupation included we find that, in the female labour market, 10.9 percentage points (or 90.6%) of the observed raw 12 percentage point part-time/full-time wage gap can be explained. The industry controls on their own account for 22.7% of this raw gap and occupation accounts for a further 34.1%. In other words, the segmentation effects are considerable. In this extended model the human capital controls are only capable of explaining 9.9% (or 1.2 percentage points) of the raw wage gap.

In column two we investigate the part-time/full-time wage gap between men employed full-time (MFT) and women employed part-time (WPT). Here we observe a raw gap of 23%. The extended model (model ‘e’ with occupation and industry) is able to ‘explain’ 44.1% (or 10.1 percentage points) of this raw gap. This explained share is considerably lower than the share of the gap explained by model ‘e’ (equal to 90.6%) when the sample is restricted to just the female labour market. The difference may be attributed, in large part, to gender discrimination. In the MFT/WPT case, and applying the extended model, the human capital variables are only able to explain 4% of the gap, whilst industry accounts for 22.7% of the gap. Occupation accounts for a further 5.7%. Once this explained portion is deducted, the adjusted MFT/WPT gap falls to 14.2 percentage points.

In column 3 (for men full-time vs. men part-time), the raw wage gap is largest, equal to 28%. When the extended or full model (with industry and occupation) is employed we estimate that 68% (or 19.0 percentage points) of the raw wage gap may be explained. Deducting this explained portion leaves an adjusted part-time/full-time wage gap in the male labour market of 8.9%. Our analysis of the components shows that occupation is the most important determinant of the gap, accounting for 26.1% of the raw gap. The industry was also important, accounting for a further 14.3%. Differences in human capital endowments between MFT and MPT accounted for a further 33.6% of the gap; this compares to the explained human capital share of 4% when comparing MFT and WPT. This is also consistent with earlier findings in Preston (2003) which showed that whilst higher educated and experienced women were moving into part-time work, this was not the case for males (where growth tended to be amongst those with less qualifications and less experience).

The final column of Table 6 shows the decomposition of the part-time/full-time gap amongst women employed full-time (WFT) and men employed part-time (MPT). The raw gap is equal to 17% (i.e. men employed part-time earn 17% less than women employed full-time). Nearly half this gap (48.2%) can be explained by differences in the human capital endowments of the two groups (with WFT significantly more qualified). A further 52% can be accounted for by the different occupations of the two groups. These results are entirely consistent with the earlier observation of a less qualified, less experienced group of male part-time employees.

Part-time/full-time gap and contract status

Finally, to test whether or not there is a difference in the gaps depending on contract status (as in the Fernandez-Kranz and Rodriguez-Planas, 2011 study), we restricted the sample to just women and estimated separate models for permanent and fixed term contract workers (reported in Table 7). We also follow Booth and Wood (2008) and explore the effect of a part-time casual interaction on wages (reported in Table 8). In all cases, we use the simplistic part-time/full-time dummy variable approach (rather than the decompositions) for expedience.

Table 7.

OLS simple model (part-time dummy coefficients) by contract type.^a

	Women permanent contract	Women fixed term contract
n	1623	168
(a) Basic controls (human capital)	−0.112 (−6.55)***	−0.037 (0.61)
(b) Expanded including industry and occupation	−0.036 (−2.39)**	0.034 (0.66)

In the interest of space the regression models associated with our results have not been reported but are available from the authors on request. *** and ** indicate significance at the 1% and 5% levels, respectively.

Table 8.

Coefficient estimates of part-time/full-time log-wage differentials; OLS; women.^a

	OLS basic model (human capital, experience, education)		OLS expanded model (including industry and occupation dummies)
	Booth and Wood (2008)	Preston and Yu (this paper)	Booth and Wood (2008)	Preston and Yu (this paper)
n = (sample women)		2140		2140
PT	−0.002 (0.18)	−0.104 (−6.46)***	+0.051 (4.55)	−0.034 (−2.4)**
Casual	−0.132 (4.52)	−0.142 (−3.66)***	−0.042 (1.52)	−0.036 (−1.01)
PT × Casual	+0.095 (2.94)	+0.140 (3.19)***	+0.060 (1.98)	+0.105 (2.7)***
Adjusted R²		0.387		0.572

Source: Booth and Wood (2008: 126, Table 2).

In the case of permanent employees we observed a part-time/full-time wage gap of around 11% in the basic model falling to 4% in the extended model with industry and occupation (Table 7). We found no evidence of a statistical difference in the earnings of part-time and full-time fixed term contract workers. These findings run counter to those reported in Fernandez-Kranz and Rodriguez-Planas (2011). In the latter they attributed the larger part-time pay penalty amongst fixed term contract workers to job transition effects. In other words, in the Spanish case, they found that women in permanent work were more likely to remain with high paying firms when transitioning to part-time work, whereas fixed term workers were more likely to transition to low paying firms when moving to part-time work. Clearly the Australian labour market is operating differently and it would seem that the limited part-time opportunities (in terms of industry and occupation) are having a detrimental impact on their jobs and, therefore, earning opportunities.

The models using a part-time × casual interaction control are revealing and suggest the presence of a compensatory premium for casual work done on a part-time basis. Consistent with Booth and Wood (2008) we find that, once casual employment is taken into account in the form of an interaction term, women in part-time work have a pay advantage relative to full-timers. In Table 8, we report the Booth and Wood coefficients and t-statistics from their pooled OLS results for women (estimated using panel data from the HILDA survey). Alongside we also report the coefficient and t-statistics on the comparable variables from our basic and extended models. In both cases, the sample is restricted to just women. If we view the results from the basic model (human capital controls), Booth and Wood show that female full-time casuals earn 13.2% less than female full-time non-casuals. This contrasts with a 9.3% premium earned by female part-time casuals (−0.002 + 0.095 = 0.093). In our study, we estimate a female full-time casual penalty of −14.2% relative to the female full-time non-casual base. Once part-time status is taken into account, like Booth and Wood, we also find a pay premium. In our case, the latter is equal to 3.6% (−0.104 + 0.14 = 0.0.36) (i.e. women who are employed on a part-time casual basis earn 3.6% more than women who are employed on a full-time non-casual basis). When industry and occupation are included (the expanded model), the pay premium for part-time casuals rises to 11.1% in Booth and Wood (0.051 + 0.060) and to 7.1% in our paper (−0.034 + 0.105). We therefore conclude that, in the female labour market, there is a part-time premium but only for casuals. We believe this is representative of a compensatory wage payment.

Summary and conclusion

Within the OECD Australia has one of the highest shares of part-time employment. In January 2013, nearly a third (29.9%) of all employment was part-time (less than 35 h per week) and evidence suggests that the incidence is rising. Whilst a number of researchers have convincingly demonstrated the risks associated with part-time work (such as constrained employment and promotional opportunities), the Australian literature has generally found no evidence of a part-time pay penalty (relative to full-timers). Indeed, Booth and Wood (2008) found evidence of a part-time pay premium. This led the authors of a recent Productivity Commission report to conclude that ‘… engaging in part-time work does not necessarily mean lower pay on an hourly basis’ (Gilfillan and Andrews, 2010: 127). The Australian evidence on the part-time/full-time pay differential is at significant odds with international (particularly UK) evidence on this topic.

In this paper we revisit the part-time/full-time pay question using a new and unique database which enables us to control for additional work and employment characteristics (such as payment method) hitherto not controlled for. We estimate a number of different specifications across men and women, employ a range of decompositions, and throughout our decomposition analysis consistently report the presence of large, significant and negative part-time pay differentials. Using the simple dummy variable approach for expedience, we also offer an analysis of the part-time/full-time wage gap in the female labour market, disaggregated by contract status. Here, consistent with our earlier work, we observe a significant part-time pay penalty amongst women who are permanent employees and no significant part-time/full-time gaps amongst women on fixed term contracts. The interesting result is with respect to women employed on a part-time casual basis. When an interaction term is employed we find that, in the female labour market, full-time casuals earn significantly less than their non-casual counterparts. Female part-time casuals, on the other hand, earn significantly more than female full-time non-casuals – equal to 3.6% when the only other controls are human capital variables, rising to 7.1% in an extended model with industry and occupation controls. The part-time casual pay premium is most likely reflective of the casual loading paid in lieu of sick pay and holiday pay, although it is unclear why, in the female labour market, a pay premium is present for part-time casuals but not full-time casuals, ceteris paribus.

In conclusion, we note the significant explanatory power of industry and occupation in accounting for observed part-time/full-time gaps, although we question the ‘legitimacy’ of these controls given the highly segmented nature of part-time work in Australia. Overall our findings caution against a benign view of part-time employment. Our research shows that, for many, there are significant financial penalties associated with this form of employment. We further caution against an uncritical promotion of part-time work in Australia. It may be that the right-to-request provisions will see an expansion of part-time work into new industries and occupations, and with it the creation of better ‘quality’ part-time jobs. It is, however, too early to test the effects of this, but it will warrant attention in the future. From a public policy perspective, policies to further expand part-time work must be accompanied by arrangements that ensure fair and equitable treatment of all part-timers relative to full-timers, otherwise wage gaps will continue to grow and greater inequality will result.

Footnotes

Declaration of conflicting interests

The authors declare that there is no conflict of interest.

Funding

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

Acknowledgements

The authors would like to thank anonymous JIR referees, and also Sally Wright, Toby Evans and participants at the 27th AIRAANZ (Association of Industrial Relations Academics of Australia and New Zealand) conference, including Jill Rubery, for extremely helpful comments and suggestions on earlier versions of this paper. Alison, in particular, would like to dedicate this paper to Paul Miller and David Plowman who both supervised her PhD thesis on wage determination, both provided important research mentorship over several years and, in the case of Paul, also provided very helpful comments on this particular paper. Tragically both Paul and David died in late 2013 leaving an enormous gap in the IR and Labour Economics communities.

Notes

Biographical notes

Alison Preston is Deputy Dean of the Business School at The University of Western Australia. Her research interests include wage determination, wages policy, women's employment and women's pay.

Serena Yu is a Senior Research Analyst at the University of Sydney Business School. Her research interests include women's employment and pay, retirement economics, and the economics of wellbeing.

References

Abhayaratna

Andrews

Nuch

(2008) Part-time employment: The Australian experience, Staff Working Paper: Productivity Commission.

Austen

Jefferson

Preston

(2013) Contrasting economic analyses of equal remuneration: The Social and Community Services (SACS) case. Journal of Industrial Relations 55(1): 60–79.

Austen S, Jefferson T, Preston AC, et al. (2008) Gender pay differentials in low-paid employment. Australian Fair Pay Commission, Research Report No. 3/9. October.

Australian Bureau of Statistics (ABS) (2011) Employee Earnings and Hours. May 2010. Cat. No. 3606.0. Canberra: ABS.

Australian Bureau of Statistics (ABS) (2013a) Labour Force Survey, Australia. Cat. No. 6202. Canberra: ABS.

Australian Bureau of Statistics (ABS) (2013b) Labour Force Australia, Detailed. Quarterly. Cat. No. 6291.0.55.003. Table 11. Employed persons by actual hours worked, industry and sex.

Bardasi

Gornick

(2008) Working for less? Women’s part-time wage penalties across countries. Feminist Economics 14(1): 37–72.

Bergmann

(1986) The Economic Emergence of Women, New York, NY: Basic Books.

Blinder

(1973) Wage discrimination: Reduced form and structural estimates. Journal of Human Resources 7(4): 436–455.

10.

Booth

Wood

(2008) Back-to-front down under? Part-time/full-time wage differentials in Australia. Industrial Relations 47(1): 114–135.

11.

Campbell

Brosnan

(2005) Relative advantages: Casual employment and casualisation in Australia and New Zealand. New Zealand Journal of Employment Relations 30(3): 33–45.

12.

Chalmers

Hill

(2007) Marginalising women in the labour market: ‘wage scarring’ effects of part-time work. Australian Bulletin of Labour 33(2): 180–201.

13.

Charlesworth S (2012) Working time regulation. Briefing notes for Work+Family Policy Round Table meeting. University of South Australia, November.

14.

Connolly

Gregory

(2008) Moving down: Women’s part-time work and occupational change in Britain 1991–2001. The Economic Journal 118: F52–F76.

15.

Connolly

Gregory

(2009) Part-time pay penalty: Earning trajectories of British women. Oxford Economic Papers 61: i76–i97.

16.

Cotton J (1988) On the decomposition of wage differentials. Review of Economics and Statistics 70(2): 236–243.

17.

Fagan C and Burchell B (2002) Gender, Jobs and Working Conditions in the European Union. European Foundation for the Improvement of Living and Working Conditions, Dublin.

18.

Fernandez-Kranz

Rodriguez-Planas

(2011) The part-time pay penalty in a segmented labor market. Labour Economics 18: 591–606.

19.

Gilfillan G and Andrews L (2010) Labour force participation of women over 45. Staff working paper. Productivity Commission. Canberra.

20.

Grimshaw

Rubery

(1998) Integrating the internal and external labour markets. Cambridge Journal of Economics 22(2): 199–220.

21.

Grimshaw D and Rubery J (2007) Undervaluing Women’s Work. European Work and Employment Research Centre, University of Manchester. Working paper series no. 53.

22.

Harkness

(1996) The gender earnings gap: Evidence from the UK. Fiscal Studies 17(2): 1–36.

23.

Harley B and Whitehouse G (2000) Women in part-time work, a comparative study of Australia and the UK. Working paper in Human Resource Management, Employee Relations and Organisation Studies No. 13, University of Melbourne.

24.

Heckman

(1979) Sample selection bias as a specification error. Econometrica 47(1): 153–161.

25.

Jefferson

Preston

(2010) Negotiating fair pay and conditions: Low paid women’s experience and perceptions of labour market deregulation and individual wage bargaining. Industrial Relations Journal 41(4): 351–366.

26.

Jefferson

Preston

(2011) Labour markets and wages in Australia: 2010. Journal of Industrial Relations 53(3): 303–323.

27.

Manning

Petrongolo

(2008) The part-time pay penalty for women in Britain. The Economic Journal 118: F28–F51.

28.

Miller

Rummery

(1991) Male–female wage differentials in Australia: A reassessment. Australian Economic Papers 30(5): 50–69.

29.

Mumford

Smith

(2009) What determines the part-time and gender earnings gap in Britain: Evidence from the workplace. Oxford Economic Papers 61: i56–i75.

30.

Oaxaca

(1973) Male–female wage differentials in urban labor markets. International Economic Review 14(3): 693–709.

31.

O’Dorchai

Plasman

Rycx

(2007) The part-time wage penalty in European countries: How large is it for men? International Journal of Manpower 28(7): 571–603.

32.

Organisation for Economic Co-operation and Development (OECD) (2012) Employment Outlook 2012.

33.

Pocock

Buchanan

Campbell

(2004) Securing Quality Employment: Policy Options for Casual and Part-time Workers in Australia, Sydney: Chifley Research Centre.

34.

Pocock

Elton

Preston

(2008) The impact of ‘Work Choices’ on women in low paid employment in Australia: A qualitative analysis. Journal of Industrial Relations 50(3): 475–488.

35.

Preston

(2003) Gender earnings and part-time pay in Australia, 1990–1998. British Journal of Industrial Relations 43(3): 417–433.

36.

Rodgers

(2004) Hourly wages of full-time and part-time employees in Australia. Australian Journal of Labour Economics 7(2): 231–254.

37.

White

(1980) A heteroskedasticity-consistent covariance matrix estimator and a direct test of heteroskedasticity. Econometrica 48(4): 817–838.

38.

Whittard

(2003) Training and career experiences of women part-time workers in a finance sector organisation: Persistent remnant of the ‘reserve army’? Australian Journal of Labour Economics 6(4): 537–557.