Related resources
Search for item elsewhere
University researcher(s)
Modelling Gender Pay Gaps
Olsen, W.K., and Walby, S
Manchester: Equal Opportunities Commission; 2004.
Access to files
-
FULL-TEXT.PDF (pdf)
Abstract
EXECUTIVE SUMMARYIntroductionThere has been little change in the full-time gender pay gap since the mid 1990s andin the female part-time/male full-time pay gap since the mid 1970s. The gender gapin hourly earnings for those employed full-time in Britain in 2003 was 18 per cent,while that between women working part-time and men working full-time was 40 percent.This research uses statistical methods to identify how much of the gender pay gap isassociated with different factors. The data set analysed is the British HouseholdPanel Survey, a sample of around 10,000 adults. The data are weighted to benationally representative.Broadly, the research finds that gender differences in life-time working patternsaccount for 36% of the pay gap. Rigidities in the labour market, including those thatconcentrate women into particular occupations and mean that they are more likely towork in smaller and non-unionised firms, account for a further 18% of the pay gap.38% is due to direct discrimination and differences in the labour market motivationsand preferences of women as compared with men. The remaining 8% is due towomen's lesser educational attainment in the past.In many instances, these factors will of course be related to each other. For example,the occupations with higher female participation in which women are concentratedwill sometimes also be those where part-time work is particularly prevalent.The importance of indirect discrimination and systematic disadvantage is noted. Theycan affect the labour market motivations and preferences of women; they are part ofthe causes of labour market rigidities; and they are part of the reasons that particulartypes of working patterns result in lower wages. It is therefore incorrect to make asimplistic assumption that gender wage differences due to variations in educationand working patterns are legitimate because they reflect skills, qualifications andexperiences that are relevant to employers.KEY FINDINGSFactors affecting wagesThe first stage of the analysis was to model how different factors impact on wages forboth women and men. Because the BHPS is used, the regression model is able toinclude the impact of work histories and a particularly wide range of variables. Keyfindings from this part of the research show that:iiiMODELLING GENDER PAY GAPS??? for each year of full-time education, hourly wages increase by 6%;??? for each year of full-time employment, hourly wages increase by 3%;??? for each year of part-time employment, hourly wages decrease by 1% (inaddition to missing out on the 3% gain that each year of working full-timebrings);??? for each year of interruptions to employment for childcare and family care work,hourly wages decrease by 1% (again, in addition to missing out on the 3% gainfrom each year of full-time employment);??? for every ten percentage points higher the proportion of men working in anoccupation, hourly wages are boosted by 1% (in other words, on average, thoseoccupations with more women working in them are valued less in terms of thewages paid):??? other factors associated with being female have a particularly large impact,reducing hourly wages by 9%. These factors include direct discrimination. Theyalso include the different preferences, motivations and attitudes to the labourmarket of women as compared with men, which may in part be attributable toindirect discrimination (or systematic disadvantage).Although some of these percentages sound small, the cumulative effect can be great.For example, ten years spent as a part-time worker would leave someone with hourlyearnings more than a third below that of someone who had worked full-time for thesame period.The size of the components of the gender wage gapThe gap in wages between men and women occurs because, on average, theposition of women and men in relation to the above factors that affect wages aredifferent. For example, on average, the occupation a man is employed in is 68%male, while that for a woman is 32% male. As stated above, the research shows thatthe higher the proportion of males in an occupation, the higher the wages, so the factthat women are more commonly in occupations with fewer males means that theiraverage wages are lowered by this factor. Similarly, the fact that women spend moretime out of the labour force caring for their family or working part-time and feweryears working full-time also lowers their wages relative to men who spend less timedoing so.ivEXECUTIVE SUMMARYAnother statistical model (simulation) is used to identify how much of the total paygap between women and men (the gender wage gap) is accounted for by each of thefactors. The most important are:Employment experience:??? less full-time employment experience (19%);??? more part-time employment experience (3%);??? more experience of interruptions to employment for childcare and other familycare(14%);Where women work as a result of rigidities in the labour market:??? the concentration of women into occupations with high proportions of femaleworkers (10%);??? other institutional factors, including the greater proportion of women working forsmaller firms and the smaller proportion in a union or staff association (8%);Direct discrimination and different labour market preferences and motivations:??? other factors associated with being female, including direct discrimination anddifferent preferences and motivations (some of which will be attributable toindirect discrimination or systematic disadvantage) (38%);Education:??? women's lesser education (8%).Discrimination can affect all components of the pay gapThe components of the gender pay gap listed above include factors that havetraditionally been associated with either the development of knowledge and skills(education, employment experience) or with discrimination (occupationalsegregation; some of the other factors associated with being female). Thedevelopment of knowledge and skills (human capital) has been seen as a legitimatesource of earnings differences because it is made up of skills, qualifications andexperiences that are relevant to employers. Moreover, the attainment of humancapital has been seen primarily as being determined by an individual. By contrast,discrimination has been seen as a failure in the working of the labour market, andthus a legitimate site of public policy intervention.This simple distinction between human capital and discrimination is overdrawn andcan have misleading implications for policy. This is because women can face indirectvMODELLING GENDER PAY GAPSdiscrimination and systematic disadvantage in acquiring human capital. Theacquisition of human capital depends upon women's place in the labour market, aswell as on individual decisions.This can be seen in the case of part-time employment. While working full-time isassociated with increased wages, working part-time is not - not even pro-rata.Rather, experience of part-time employment is associated with a slight reduction inwages. The typical assumption in human capital theory, that experience ofemployment increases human capital and thus increases wages, is challenged bythis finding. Rather, whether or not employment experience leads to increases inwages depends on the location of that experience within a differentiated labourmarket. Some of the reasons that women find themselves in a different labourmarket, i.e. competing for a different range of jobs than equivalently qualified men,may be thought of as indirect discrimination.The differential impact of years spent working part-time, as compared with full-time,has serious implications for both women???s wages and UK productivity and is worthyof policy intervention, whether or not discrimination is the whole cause of thedifference or not. Discriminatory practices, both direct and indirect, may be found notonly embedded in factors such as occupational segregation, but also within theprocesses by which human capital is acquired.Implications for policyThese findings have implications for two main types of government policy: first,policies concerned with gender justice; and second, policies concerned with theproductivity of the UK economy and its capacity for economic growth.In relation to the latter, this research has found that the acquisition of skills and otherforms of human capital is associated not only with education and length ofemployment experience, but also by the context of that employment experience(whether the experience is full-time or part-time) and by interruptions to it. There is agendered dimension to the acquisition of human capital, which is affected by theinstitutional context.In relation to competition, the research has found that labour markets are notperfectly competitive, and that they contain significant rigidities (such as occupationalsegregation) and forms of discrimination, which affect women???s potential in the labourmarket.viEXECUTIVE SUMMARYviiFuture research modelling policy scenariosThe next phase of this research could be to identify specific policies and to modeltheir implications for the gender pay gap and other policy relevant concerns.Examples of such policies are:??? universal childcare;??? training for returners;??? improved flexibility in the workplace;??? anti-discrimination policies.The analysis will identify the implications of such policies for a range of stakeholdersand beneficiaries including the Exchequer and employers as well as the UK economyas a whole and society as a whole.