The economics of gender discrimination
It has been 40 years since Australian women won the right to equal pay, yet women continue to earn about 87 cents on average for every dollar that men earn. What do we know about gender pay gaps in Australia?
(pages 17-21 of printed journal)
Forty years after the 1969 Equal Pay Case there continues to be a substantial gap in the wages of Australian men and women. Women’s relative wages are lower amongst the self-employed (Eastough and Miller, 2004), at the top of the wage distribution (Miller, 2005; Kee, 2006) and in the private sector (Barón and Cobb-Clark, 2010). Across the labour market as a whole, women earn approximately 87 cents for every dollar that men earn. This gap in
men’s and women’s pay stems largely from the fact that men and women earn a different return on their qualifications, experience and expertise – not because they have different skill endowments. It raises questions about what produces this gender pay gap and whether discrimination within the Australian labour market is at least partially to blame.
I will begin by outlining the way that economists typically think about labour market discrimination and then reviewing the empirical evidence on the source of the gender pay gap. I will also discuss some of the issues that we still do not understand well and will conclude by highlighting some of the challenges that I see facing us in achieving gender equity in the Australian labour market.
How do economists think about discrimination?
Economists generally define labour market discrimination to be the unequal treatment of equally productive groups. Specifically, discrimination exists when two equally qualified individuals are treated differently based solely on non-productive characteristics such as gender, religion, ethnicity, race, sexual orientation, and so on (see Blau et al, 2010). Discrimination can occur when (i) employers simply prefer to hire one group of workers to other groups, (ii) employees prefer that their co-workers have certain characteristics, or (iii) customers prefer to deal with specific types of workers. These so-called ‘taste for discrimination’ models, according to Becker, imply that preferred workers will be paid more – or be hired more frequently – because they are more valuable in some sense despite not being more productive. Statistical discrimination occurs, on the other hand, when employers deal with a lack of information about a particular worker’s productivity by applying stereotypes, prejudices or rules of thumb pertaining to the group to which a worker belongs.
Much of the empirical economics research has focused on measuring gender discrimination, though researchers have also studied discrimination that occurs because of workers’ race, ethnicity, religion or sexual orientation. Determining whether discrimination has occurred and measuring the extent of it can be extremely difficult. A myriad of econometric issues including feedback effects, omitted variables, sample selection and more all imply that our statistical measures of discrimination may be biased. In short, measuring discrimination is very hard because we very seldom know everything about workers’ productivity. We may think that we are comparing equally productive workers when that is not the case.
How does the gender pay gap differ for high- and low-pay workers?
Figure 1 shows how the ratio of Australian men’s to women’s wages varies across the wage distribution in the public and private sector. In public-sector employment, the gap in men’s and women’s pay is relatively constant. On average, women earn approximately 88 per cent of what men earn irrespective of whether they are in the bottom or the top quintile of the wage distribution. In contrast, the median gender wage gap in private sector employment is slightly larger and there is a sharp increase in the size of the gap as wages increase. Women in the bottom quintile of the wage distribution earn approximately 92 per cent of what low-paid men earn. Highly-paid women, on the other hand, earn closer to 72 per cent of what similarly paid men earn. In the top half of the wage distribution (above the median), the gender pay gap is significantly higher in private- than in public-sector employment.
Figure 1: Natural Logarithm of the Gender Wage Ratio by Sector (the shaded areas show 5 per cent confidence intervals)
Notes: The sample includes employees aged 22 to 60 receiving wages in each wave. We use data from HILDA waves 1 to 6. We use the Consumer Price Index (CPI) to transform wages to 2001 dollars. Wages per hour are calculated as the ratio between current weekly gross wage and salary to hours per week usually worked both in the main job. The bands represent bootstrapped-normal confidence intervals at the 95 per cent level with 5,000 replications and clustered at the individual level. The graph excludes the top and bottom 5 percentiles. The dashed and solid lines represent the coefficient on a gender dummy (1 for male, 0 for female) from quantile regressions, at each percentile, of the logarithm of the wage per hour (in 2001 dollars) on the gender dummy. There are 4,132 women and 4,181 men in the sample, with 25,982 person-year observations.
Source: Barón, J and Cobb-Clark, D, 2010 ‘Occupational segregation and the gender wage gap in private- and public-sector employment: distributional analysis’, Economic Record, Vol. 86 (273), June, pp. 227-246.
These differences imply that economic models of discrimination – and the public policies that are used to deal with it – need to be flexible enough to take into account the full range of men’s and women’s labour market experiences both at the top and the bottom of the wage distribution. Economists typically use the term ‘sticky floors’ to describe a situation in which women are disproportionately left behind in low-paid jobs, and ‘glass ceilings’ to describe situations in which women are unable to access highly-paid jobs.
Is the gender pay gap related to the type of work men and women do?
One of the most persistent and defining features of labour markets around the world is that men and women do very different kinds of work. This raises the obvious question: is the gender pay gap related to the type of work men and women do? Although much of the international literature finds that labour market segregation does contribute to the gender wage gap in many countries (see for example Blau and Kahn, 2000), it is less obvious that segregation can account for the gender wage gap in Australia. Historically, many authors have concluded that Australian women’s relative pay would decrease – not increase – if they were employed in occupations in the same proportions as men (Rimmer, 1991; Lee and Miller, 2004).
Figure 2 shows how the segregation in men’s and women’s occupations differs in public and private sector employment in Australia. Specifically, we calculate a standard index of dissimilarity to determine the proportion of women who would have to change occupations in order to have women represented in each occupation in the same proportion as men (see Barón and Cobb-Clark, 2010). In the public sector, 55 per cent of low-paid women (or men) would have to change occupations in order to equalise the distribution of occupations across gender. This degree of occupational segregation falls steadily as public sector workers become more highly paid. The occupational distributions of the highest paid men and women could be equalised with only 40 per cent of workers changing occupations. These large changes in occupational segregation occur despite the gender wage gap remaining relatively constant.
Figure 2: Occupational Segregation across the Distribution of Wages, by Sector
Notes: The figures represented in this graph are calculated as follows: we calculate the proportion of total women and total men working in each occupation for individuals at different points of the wage distribution (e.g. in the bottom 10 per cent of their corresponding gender wage distribution). At each of these points of the distribution, and for each sector, we calculate an index of occupational segregation as ID = 0:5£ P i jMi¡Fij where i indexes occupations, and Mi and Fi are the proportions of total men and total women in occupation i. The resulting statistic tells us the proportion of women or men workers who would have to change jobs to make the distribution of occupations for each gender identical. The occupations used are based on the Australian Standard Classification of Occupations (ASCO) Sub-Major Groups and the ANU4 (40 categories in the public sector and 64 in the private sector). There are 4,132 women and 4,181 men in the sample, with 25,982 person-year observations.
Source: Barón, J and Cobb-Clark, D, 2010 ‘Occupational segregation and the gender wage gap in private- and public-sector employment: a distributional analysis’, Economic Record, Vol. 86 (273), June, pp. 227-246.
In contrast, occupational segregation in private sector employment increases gradually as workers become more highly paid. At the bottom of the wage distribution, 42 per cent of workers would need to change occupations in order to remove gender segregation. At the top of the distribution, this increases slightly to 47 per cent. This gradual increase in occupational segregation is not immediately consistent with the much sharper increase in gender pay disparities for highly-paid, private sector workers. Thus, Figures 1 and 2 provide little evidence that occupational segregation in the Australian labour market is the source of the gender pay gap.
Cobb-Clark and Tan (2011) investigate the role of occupational segregation in gender pay imbalances in more depth. In particular, they explicitly consider whether any differences in men’s and women’s psychosocial traits – specifically personalities and sense of control over their lives – influence the occupations in which they are employed and, if so, whether this contributes to the disparity in men’s and women’s wages. They use data from the Housing, Income and Labour Dynamics in Australia (HILDA) Survey, which provides detailed information over a number of years for a large, nationally-representative sample of Australian workers.
The authors find that psychosocial skills have a substantial effect on entering many, though not all, occupations in ways that differ for men and women. Consequently, Australian men and women with the same psychosocial traits enter occupations at very different rates, which contributes to occupational segregation. Australian women, however, have lower wages on average not because they work in different occupations to Australian men, but because they are paid less when working in the same occupation. If anything, women’s psychosocial traits give them a slight pay advantage in the Australian labour market.
Is the gender pay gap due to discrimination?
Economists often use decomposition analysis to factor the gender pay gap into two components: one because men and women have different qualifications, skills, experience, etc.; and the other because men and women earn different wage returns for those qualifications, skills, and experience. The former is typically referred to as the ‘explained’ gap because it can be explained by differences in men’s and women’s characteristics. The latter is typically referred to as the ‘unexplained’ gap and is often what people have in mind when they think of labour market discrimination.
Unfortunately, a variety of econometric challenges imply that it is exceedingly difficult to make comparisons between workers who are equally qualified. As a result, statistical estimates of discrimination can be often quite biased. This has lead most economists to use decomposition analysis – not to measure discrimination per se – but to highlight the extent to which gender pay gaps stem from the observable characteristics of workers, for example educational qualifications, experience, job characteristics, and so on. This allows us to draw indirect inferences about how discrimination might have changed over time and how it might vary across groups.
Barón and Cobb-Clark (2010), for example, conduct an analysis of the gender pay gap in both public and private sector employment in Australia using data from the HILDA survey. They find that in private sector employment, approximately two thirds of the median gender pay gap can be attributed to the differences in men’s and women’s characteristics. Only one third of the median gap is the result of differences in the return to those characteristics. At the same time, the gender pay gap for low-paid workers – those at the 10th percentile – is completely explained by gender differences in characteristics, while at the top of the pay distribution – the 90th percentile – less than 60 per cent is explained.
In public-sector employment nearly all – 98 per cent – of the median gender pay gap is unexplained once we account for the differences in men’s and women’s occupations. This unexplained component falls slightly at the bottom and the top of the pay distribution, but remains sizeable.
The differences in the size of the gender pay gap in public versus private sector employment are generally consistent with differences in the wage-setting mechanisms in the two sectors. In particular, the smaller pay gap in the public sector is consistent with more intensive antidiscrimination enforcement and faster occupational integration in public sector employment (see Gregory and Borland, 1999). However, it is surprising that almost none of the gap in the relative wages of highly-paid, public sector employees can be attributed to differences in their qualifications, experience, skills, etc. Finally, irrespective of sector, the gender pay gap among low-paid workers is almost entirely explained by their characteristics, while the gap among highly-paid workers is largely unexplained. This suggests that the issue in Australia is not sticky floors, but rather glass ceilings.
What are the challenges and where do we go from here?
These results advance our understanding of the gender pay gap in Australia. At the same time, they leave open a number of important puzzles yet to be resolved. Why do highly-paid women employed in the Australian public sector continue to fall behind their male colleagues? To what extent is this evidence of discrimination? Given the degree of segregation in men’s and women’s work, why do wage differences across occupations play so little role in explaining the persistent wage penalty faced by Australian women? More specifically, why does occupational segregation seem to improve rather than undermine the relative wages of women in Australia, when the opposite appears to be the case in other countries (see Blau et al, 2010).
These are important questions that must be addressed if we are to fully understand – and rectify – gender pay imbalances in the Australian labour market. Perhaps the biggest challenge of all is to understand why the gender pay gap is so persistent. Does it result from the hiring, promotion or pay decisions of employers or rather from disparity in men’s and women’s skills or preferences that we have failed to measure?
Barón, J and Cobb-Clark, D, 2010, ‘Occupational segregation and the gender wage gap in private- and public-sector employment: a distributional analysis’, Economic Record, vol. 86 (273), June, pp. 227-246.
Blau, F.D, Ferber M.A, Winkler A.E, 2010, The Economics of Women, Men, and Work, 6th Edition, Upper Saddle River, NJ: Prentice Hall.
Blau, F.D and Kahn, L.M, 2006, ‘Wage structure and gender earnings differentials: an international comparison’, Economica, pp. S29-S62.
Cobb-Clark, D.A and Tan, M, 2011, ‘Noncognitive skills, occupational attainment, and relative wages’, Labour Economics, vol. 18(1), pp. 1-13.
Eastough, K and Miller, P, 2004, ‘The gender wage gap in paid- and self-employment in Australia’, Australian Economic Papers, vol. 43, pp. 257-276.
Gregory, R.G and Borland, J, 1999, ‘Recent developments in public sector labor markets’, Handbook of Labor Economics, volume 3C, Chapter 35, North-Holland, New York, USA, pp. 3573-3630.
Kee, H.J, 2006, ‘Glass ceiling or sticky floor? Exploring the Australian gender pay gap’, Economic Record, vol. 82(259), pp. 408-427.
Lee, Y and Miller, P, 2004, ‘Occupational segregation on the basis of gender: the role
of entry-level jobs’, Australian Journal of Labour Economics, vol. 7 (3), pp. 355-374.
Miller, P.W, 2005, ‘The role of gender among low-paid and high-paid workers’, The Australian Economic Review, vol. 38, pp. 405-417.
A condensed version of an Alumni Refresher Lecture delivered at the University of Melbourne on 21 September 2010.