Although the difference between male and female hourly earnings has narrowed since the 1970s [1], studies on wages continue to report a persistent and sizable wage gap. Indeed, since 2005 evidence has emerged for a steady widening in gender wage differentials [2, 3]. Several factors have been offered as explanations for the gap. Some factors relate to individual characteristics and work choices, others to institutional factors. Wages have been found to vary with the gender composition within occupations [4, 5]. Rewards for human capital endowments, rather than differences in these endowments, offer another explanation for these gaps [6–8]. Other key determinants include industrial segregation, labour force history, under-representation of females with vocational qualifications and under-representation of females in large firms [3].

Compared with other OECD countries, Australian females possess a relatively small earnings disadvantage. This is mainly attributed to the comparatively high degree of labour market regulation [9]. Investigations reveal that, within Australia, gender wage differentials tend to be entrenched in the system of pay determination [10–13]. Sectors and occupations dominated by individual pay setting arrangements show a wider gender wage gap than those under a collective agreement. A larger gender wage gap exists among high-paid workers than low-paid workers [11], among the self-employed than wage/salary earners [9], and among the private sector than public-sector employees [10].

Typically, individuals achieve higher incomes by working more hours than their counterparts [14]. Advancement and promotion in the workplace is partly based on market signals such as positive unpaid overtime. Based on investigations of wage inequality between the United States and Germany, Bell and Freeman argue that labour supply decisions are forward looking and incentive driven [15]. Longer hours worked in one period improve the probability of promotion and improve the wage of workers in the future. Working longer hours is the most significant way to signal that an individual is a good (that is, quality) worker [15, 16]. Higher-quality workers tend to be paid higher wages [17]. Various economic models offer explanations for wage differentials based on the quality of the worker [17]: firms may pay high wages to attract high-skill workers [18]; firms offer efficiency wages as an incentive for effort [19]; or workers bargain for a share of the gains from investments in their firm-specific skills [20].

Recognizing that wage differentials are partly the result of different hours of work, studies relating to wage gaps typically incorporate hourly wage into the analysis (for example, [7, 10, 14]). However, given the limited availability of information, little empirical evidence exists that accounts for unpaid overtime. Unpaid overtime is defined as any work undertaken in addition to ordinary or standard working hours in which an individual receives no compensation; for example, a health professional working 45 hours in a week but is paid for only 40 hours.

The Blinder–Oaxaca decomposition technique [21, 22] addresses the ways in which different factors contribute to gender wage differentials. Previous studies that have examined the gender wage gap using the Blinder–Oaxaca decomposition method have reported a proportion of the gap explained by gender differences in human capital characteristics, job characteristics and family responsibilities. Yet an unexplained gap remains [9, 23, 24].

The aim of this study is to identify the magnitude of gender wage differentials among Australian employees within the health sector. This study takes advantage of new data that account for unpaid hours of market labour in the analysis of wage differentials. The Blinder–Oaxaca decomposition method is used to identify and quantify the contribution of individual endowments, and in particular unpaid overtime, to wage differentials. The hypothesis is that the sexual division of labour in the household [25] has led to females performing less unpaid overtime than males within the formal labour market. This hypothesis may partly explain the higher incomes of males relative to their female counterparts. Watson’s finding of a difference in the number of hours worked between male and female managers working full-time reinforces the argument that the domestic division of labour contributes to wage differentials [24].

In addition to fairness and equity there are strong economic imperatives for addressing the gender wage gap [3]. Using rigorous macroeconomic modelling techniques, researchers in Australia at the National Centre for Social and Economic Modelling concluded that the negative impact of the gender wage gap on performance stemmed primarily from the disincentives to work more hours [3]. Other researchers have found that as the gender wage gap decreases, females work more due to the added wage incentive [26]. This implies that cost of the gender wage gap on the health sector may be substantial. The demand for health services, partly driven by an ageing population, is expected to escalate. Workforce ageing and the greater feminization of some health professions have led to reductions in average hours worked, contributing to persistent labour shortages within the Australian health sector [27]. An understanding of the factors that contribute to wage differentials will assist agencies in implementing incentives to improve labour-force participation rates and increase hours worked.

An econometric model for hourly wage is developed, and explanatory variables customary to studies of this nature are included in the log-lin model.

where lnWi is the log of hourly wage, Gender is a dummy variable with the value 1 for male, UO _i becomes unpaid overtime and X _ij becomes a vector of j variables known to impact on wage differentials that includes the level of education, occupation, experience, experience squared, marital status, number of children, number supervised, and private/public sector.

Initially, ordinary least-squares regression is used to identify the gender wage gap when unpaid overtime is included and then excluded from the model. As displayed in Equation (1), log of hourly wage is regressed against several variables characterizing individual and work characteristics. The gender dummy variable measures the gender differential. Later we stratify the sample by gender to allow a comparison between male and female incomes on the impact of unpaid overtime. Additional analysis estimates the hourly wage separately for each occupational group.

Lastly, the Blinder–Oaxaca decomposition [21, 22] is employed to decompose the gap in outcomes between males and females. The gender pay gap in this log-lin method arises from two functions. The first is the actual difference in variables as shown by the difference in mean values between males and females – a part that is explained by group differences in productivity characteristics such as education, work experience and unpaid overtime. The second function is the discrimination function or unexplained residual – a part that cannot be accounted for by differences in characteristics [28]. This unexplained component is traditionally interpreted as a measure of discrimination. The expectation is that similar endowments should translate into similar wage levels among females and males. Any differences in male and female earnings from this function arise from the difference in male and female coefficients [29].

where ln(W) is the log of hourly wages, x _ij ^M is the vector of means from the male equation, x _ij ^F is the vector of means for the female equation, β _ij ^M is the vector of coefficients from the male equation, and β _ij ^F is the vector of coefficients from the female equation.

A cross-sectional approach is used to investigate gender–wage differentials. Although cross-sectional analysis ignores the effects of institutional and technological change and changes in the labour market over time [30], it does permit the inclusion of certain variables, such as human capital, and allows an examination of the wage distribution across individuals [31]. Furthermore, the persistent nature of the gender wage gap shows that cross-sectional analysis is not as biased as it would be in other applications, since the wage gap shows little variation overtime [32].

This study identified that there are significant differences between males and females regarding their human capital endowments, job characteristics and family responsibilities that partly explained the gender wage gap. The results of the Blinder–Oaxaca decomposition revealed that, after adjusting for endowments, a gap of 16.7% remained unexplained.

This is in line with other Australian studies across all sectors that reported estimates of between 10 and 20% [9, 11, 39]. Stratifying the sample showed evidence of gender wage differentials that varied across occupation. Even so, males consistently earned disproportionately more than females across all occupational categories. The adjusted gender wage gap tended to be larger in higher-income occupational categories (for example, 19.4% in the professional/technical group) and smallest for the lower income occupation (for example, 9.3% in the clerical/services group). These findings are consistent with previous research [11].

Similar to previous studies [9, 11, 14], the returns on education were positive and increased with the level of education. Males received higher returns to schooling than females. Marital status was also an important determinant of earnings. Married males possessed a higher income than their single counterparts. These results are consistent with studies by Van Der Meer [14], Eastough and Miller [9] and Langford [32]. Married males’ greater attachment to paid employment may explain their relatively higher wages. In his paper on the sexual division of labour, Becker argued that the persistence of gender wage differences may arise from women’s greater responsibility towards informal unpaid work (for example, caring for children, the older person or those with physical disabilities, household chores, and so forth) [25]. Males under this arrangement are allocated the main financial responsibility of supporting the household. Married males are therefore likely to possess a greater commitment to paid employment compared with single males.

Similar to Miller [11] and Eastough and Miller [9], the analysis showed that married females earned less than unmarried females. The result that a greater number of children was associated with highe incomes may be due to the lack of information regarding the dependency of children; that is, older children may be more likely to have two working parents. Alternatively, a greater number of children in the household drive demand for higher income. Eastough and Miller found that only males with dependent children earned higher wages [9]. A limitation of the study is that no information on the age of the respondents’ children was collected. This would have facilitated stratifying the sample to assess any differences between new parents and parents of less dependent adolescents. Similar to others [11, 40], we found that public-sector employees were better remunerated than their private-sector counterparts.

In the determination of hourly wages, years of experience became important for males but not for females. Females’ contribution towards household labour may perhaps limit their advancement within an organization and thus dampen the impact of experience on hourly wages. Another possibility is that the experience variable may be a better reflection of work experience for the male rather than the female sample. The data did not include information on actual labour market experience. Females’ intermittent workforce participation can result in potential experience being a poor measure of their actual experience. When the analysis was stratified by occupation, the work experience coefficients were not significant at all. Males possibly tend to progress as they gain experience and move out of lower paid occupations into manager roles. Across occupational groups, educational qualifications were important for higher ranks of occupational status, but the effect progressively disappeared in the lower ranks. This finding is consistent with previous research [6].

The descriptive statistics showed higher amounts of unpaid overtime performed by males. To ascertain the relationship between unpaid overtime and hourly wages, a log-lin regression was performed that controlled for confounding variables. When unpaid overtime was accounted for, the wage gender gap decreased slightly. Omitting unpaid overtime therefore tended to overestimate the wage gender gap. In addition, this study showed that females were less likely to work unpaid overtime than males, and that the wage effect for unpaid overtime for females was lower than that for males. That is, males were more likely to work unpaid overtime and receive higher incomes than females. This is in line with research from the Netherlands where it was found that the effect of unpaid overtime on wages was less for females than for males [14].

Miller [11] and others [9] have argued that the varying gender wage gap among the lower and higher income groups tends to reflect the methods by which pay is set. Consistent with their research, this study found a wider gender wage gap among higher-paid occupations. Among higher-paid occupations in the health sector (managerial and professional/technical) there is scope for bargaining and managerial discretion (for example, individual pay setting) to reward employees that signals their greater productivity. This is in contrast to the occupations with lower incomes (clerical/service), where enterprise bargaining and higher union coverage rates create fixed pay structures through collective agreements for the occupational group. This difference in pay determination may therefore go some way to explaining the wider gender wage gap within higher-earning occupations.

Workers with more complex job tasks and/or leadership roles may have a greater mismatch between paid work hours and actual work hours. Bell and Hart [36] showed that workers in occupations such as managers and professionals were more likely to report extra work for no pay. These workers tended to have a greater vested interest in the company, to have more responsibility and satisfaction in their work, may be less mobile and may be more difficult to replace. Nevertheless, male and females in these higher-paid occupations are likely to have the same level of vested interest, and therefore should have similar incomes. Workers in less skilled jobs, where work scheduling and job tasks are more precisely defined and demarcated, tend to be paid for all hours worked in excess of their standard work week. Meng [13] found that narrower gender wage gaps existed in firms that were easily able to identify labour productivity at the individual level.

A substantial gender wage gap existed within the professional/technical group. Female-dominated subcategories (such as the nursing profession) are likely to set their pay under a collective agreement. For instance, in Queensland 94% of nurses in 2005 were female [41] and most worked under collective agreements. Male-dominated subcategories, such as medical specialists, tend to work under an individual arrangement. Various studies confirm that pay determination significantly impacts on gender wage differentials [10–13]. The wage setting method of male-dominated and female-dominated subcategories and the subcategories themselves could not be identified. This is a limitation of the study. Another limitation is the use of broad occupational groupings. The estimate of the role of the occupational group would most probably increase if the model included more disaggregated information [42].

Studies that model gender differences in wages often extend the model to include job characteristics such as hours worked, occupation and industry [23]. Regarding the potential endogeneity of hours worked, our study deals with unpaid overtime in the Blinder–Oaxaca decomposition in a similar manner to previous studies of this nature that include hours worked [23, 24]. Although a person may make more money because they work more unpaid overtime, or may work more unpaid overtime because they make more money, these effects are controlled for to some extent by the inclusion of job characteristics (occupation, supervision and sector) as control variables in the model. Furthermore, unpaid overtime was calculated by the actual minus expected hours worked per week. If an employee was expected by their employer to perform a certain amount of overtime, then actual and expected hours would equal each other. This further mitigated the endogenous nature of unpaid overtime. Although various studies use cross-sectional data to explain gender wage gaps, it is acknowledged that data from different periods would have been ideal since it would have produced unpaid overtime as a similar variable to the classic ones. A limitation of the study is the unavailability of data from different periods.

The inclusion of the unpaid overtime variable is an important factor that appeared to be sensitive to the pay determination arrangements of the occupational category. Unpaid overtime should thus be included in any analysis of wages and incomes. However, the fact remains that, in the health sector, simply being female is the major contributing factor to the gender wage gap. Previous research shows that the primary cost of females earning less than males is the disincentive to work more hours [3, 26]. The Australian health sector is expected to continue to face a shortage in skilled labour. Addressing wage differentials in this sector is one of several strategies that could be implemented to improve labour supply and meet the healthcare needs of an aging population.

^aFor example, in contrast to the Work Outcomes Research Cost-benefit survey, the Australian Bureau of Statistics census reported certain groups of workers belonging to the Community and Personal Service sector rather than the Health and Community Services sector. The Community and Personal Service category includes workers that assist health professionals in the provision of patient care, provide information and support on a range of social welfare matters, and provide other services in the areas of aged care and childcare, education support, hospitality, defence, policing and emergency services, security, travel and tourism, fitness, sports and personal services [43].

^bThis concept of actual and expected is not unusual in the economics discipline. For example, economists often refer to actual and planned investment.

^cIn Queensland, children generally start school at 6 years of age.

^dVarious studies include additional variables in wage differential models such as duration of current employment [23, 24], primary caregiver of dependent children, employment status of partner [23], union membership [24], and size of the organization [10, 24]. Owing to data limitations, these variables were not included in our model.