Danish gender wage studies

WOMEN S MEN S & WAGES Danish gender wage studies Danish gender wage studies.... side 76

4. Danish gender wage studies Chapter 4 provides an overview of the most important economic analyses of wage differences between women men in Denmark over the last 10 years. Most of the analyses take their point of departure in the Oaxaca decomposition of the wage gap, based on the premise that wages to a significant degree reflect the individual person's background characteristics such as education, job experience, etc.; cf. also Chapter 2. DA's in Labour Market Report of 2001 is not part of this presentation, as the description of methods results to a wide extent form part of the main report. In 1994, Michael Rosholm Nina Smith published an of trends in gender wage differences in Nationaløkonomisk Tidsskrift. The purpose of the was to find explanations for the stagnation of the general wage gap between women men. Data in the contains a rom sample of one per cent of the Danish population for the years 1980-1990. As explanatory variables, job experience, length of education, sickness indicators degree of unemployment are included. Added to this are several demographic variables (no. of children, their ages, marital status, residence, etc.). The includes only whitecollar workers unskilled workers. Skilled workers are excluded, as there are relatively few skilled female workers. The hourly wage concept in the is calculated on the basis of ATP figures (supplementary pension payments) for the total wage income number of hours worked. Analysis of the background characteristics in the period shows that there has occurred a reduction of the differences in the background characteristics. This applies especially to white-collar workers. Among unskilled workers in the public sector, there has occurred a significant reduction in the difference in job experience among women men. During the period surveyed, the gender wage difference in the public sector decreased for both white-collar workers unskilled workers by, respectively, 4.5 1.3 percentage points, reaching 18.3 per cent for white-collar workers 8.5 per cent for unskilled workers in 1990. For private 76

sector white-collar workers, the wage difference decreased by 3.4 percentage points to 29.7 per cent. Most of the wage difference for white-collar workers in 1980 1990 is unexplained. For the group of unskilled workers, the entire difference is unexplained. For the group of white-collar workers in the public sector to lesser degree white-collar workers in the private sector, it is possible to explain wage trends in terms of the narrowing of the differences in education job experience. For the unskilled group, the effect of the narrowing of the differences in men women's background characteristics is virtually zero. Education can even have a negative effect. The higher degree of unemployment for women, especially among unskilled, cannot immediately explain the stagnation in the general gender wage difference. The causes of stagnation in wage differences among unskilled workers must be sought in factors other than the development of background characteristics. Here the authors point to successive governments' wage restraints in the public sector, which has especially affected women due to their high proportion of public sector employees. Also the transition to more decentralised wage negotiations, the elimination, in 1982, of automatic wage indexing, which has played a large role historically for the annual wage increases plays a role. In The Danish Gender Wage Gap in the 1980's: A Panel Data Study Rosholm Smith continue their work outlining the causes of stagnation in the general wage gap between women men. In their, the same group of persons is included as was included in the in Nationaløkonomisk Tidssskrift, the is here, too, undertaken for white-collar workers unskilled workers in the public private sectors. Methodologically, this study is a panel data which covers the years 1980-90. The shows that variables such as education job experience have relatively great significance for the remuneration of white-collar workers in the public private sectors. It is further shown that female white-collar workers 77

receive lower remuneration for these characteristics than do men. For the unskilled workers, the picture is quite different. Both education experience can in certain cases have a negative effect on remuneration. Generally speaking, the rewards for production characteristics are of limited size for the unskilled. The degree unemployment is significant for the wage, though not for unskilled women in the public sector. The effect is greatest for male private sector whitecollar workers. In this too it is pointed out that the explanation of stagnation in the general wage gap must be sought in changes in the process of wage setting in the labour market. In 2000, the Danish National Institute for Social Research (SFI) issued a report on wage inequality, 'Wage differences between Men Women in Denmark' by Lisbeth Pedersen Mette Deding. The is based on 940,000 observations from Statistics Denmark 1996 wage statistics. As explanatory variables, the authors use sectors occupations at an aggregate level. Hence, there are nine occupations 9-27 sectors. Besides sectors they also include several human capital demographic variables. In order to avoid statistical problems, they do not include public private sector together with the main private sectors. The use of information from the wage statistics provides the opportunity to analyse the significance of various wage concepts. The following wage concepts are used in the : direct remuneration, total earnings/paid hour (sum of working time absence), total earnings/unit worked. The wage differences in the private sector, in contrast to the public sector, are relatively unaffected by the choice of wage concept; wage differences between women men are 16-17 per cent, of which 12 percentage points are unexplained. A special of wage differences within nine sectors shows that the gross wage differences between men woman are greatest within the financial sector, 23.0-25.6 per cent, in retail, hotel restaurant, 21.9 per cent- 23.6 per cent. Differences are lowest within agriculture raw materials extraction, 6.2-7.8 per cent, transport telecommunications, 9.1-10.3 per cent, again dependent on 78

the wage concept selected. The largest unexplained wage differences are found within retail, hotel restaurant, 14.6-15.3 percentage points in building construction, 12.4-14.3 percentage points. In 'The Danish Gender Wage Gap the Labour Market' Nabanita Datta Gupta, Ronald L. Oaxaca Nina Smith analyse the consequences of the change of wage structure in relation to the gender wage difference. The takes its point of departure in the Oaxaca decomposition Juhn-Murphy-Pierce decomposition, which have been developed to analyse the significance of the wage structure for the wage gap. The data derives from a rom sample of 0.5 per cent of the Danish population in the period 1983-1994. The covers only white-collar workers in the public private sector. The hourly wage is calculated on the basis of ATP information. Estimates of wage functions for the years 1983, 1989 1994 show that there occurs a slight narrowing of the wage gap in the public sector of less than one per cent over the years, to 25.1 per cent. In the private sector, the wage gap increased by one percentage point, to 36.4 per cent. The decomposition shows various explanations for the stagnation in the respective wage gaps in the public private sectors from 1983-1994. In both sectors, women improved their qualifications in relation to men. In the private sector, however, the change in qualifications was offset by the change of the relative prices for qualifications (especially experience education). In the public sector, the effect of improved background characteristics was offset by the greater wage dispersion. The further shows that if women men in the public sector had been paid in prices according to the wage dispersion that applied in the private part of the labour market, the total wage gap in 1994 would have been 2 percentage points less. The thus supports the assumption that women have been affected by the transformed wage structure on the Danish labour market. 79

In 'Children Career Interruptions', Nabanita Datta Gupta Nina Smith analyse the correlation between wages childbirth. The is based on a rom sample of 0.5 per cent of the population for the years 1980-1995 includes all full-time employees. The point of departure is again a panel data. In contrast to regression analyses based on cross-sectional data, the estimation results from the panel data analyses show that children marriage have either a positive or an insignificant effect on men's women's salaries. Births still have a negative effect on women's wages, in that women's return on their experience after having borne children is lower than before the first birth. If women have also been on leave after birth, there will be a further effect of lacking accumulation of experience. However, this is only a temporary effect on wages. Around the age of 35 years, women with children will have recovered their wage disparity in the subsequent years obtain a wage which, all things being equal, is higher than that obtained by women without children. Helene Skyt Nielsen, Marianne Simonsen Mette Verner similarly analyse the family wage gap in 'Does the Gap in Family-Friendly Policies Drive the Family Gap'. The point of departure for the is that public sector workplaces generally have more family-friendly rights in connection with maternity, etc. that this factor may be significant for women's choice of sector. The share of women in the public sector who have children is larger than in the private sector. Among 20-40-year-old women working in the public sector 62 per cent have children (0-17 years) compared to only 48 per cent of those in private sector employment. The proportion of women with two or more children is significantly greater among public-sector employed women than among those working in the private sector. The distinguishes itself from earlier analyses by explicitly taking account of the possibility that choice of sector can be associated with motherhood with maternity benefits. This occurs in the form of an endogenous switch- 80

ing model, where the probability to work in the public sector wage relations in the private public sectors are estimated simultaneously. The shows that women in the private sector who have children experience a relative salary reduction following birth compared to women without children. The effect, which is permanent, is greatest for women with 13-14 15-16 years of education, least for women with very short educations higher educations. Also, women in the public sector experience a relative wage reduction in connection with births, but here the effect is of a temporary character. 'Women at work: who are they how are they faring' in OECD Employment Outlook, July 2002, focuses on various aspects of gender differences in the labour market, including the gender segregated labour market gender wage differences. In this connection, several international statistical analyses are carried out which also cover the Danish labour market. The analyses concentrate on the following issues: 1) The gender segregated labour market 2) Decomposition of the wage gap 3) 'Family wage gap' ( of whether women with children are affected in terms of wages) As concerns the gender segregated labour market, it appears that 75 per cent of the women in the labour market are employed in 19 out of 114 occupations. Within these occupations, there are an average proportion of 70 per cent women. The corresponding figures for men are 30 occupations out of 114 73 per cent. The Danish figures are at level with the average figures for the OECD. It appears, however, that there is apparently a positive association between the degree of gender segregation in the labour market women's labour participation rate. 81

In several countries, including Denmark, there is a tendency for the youth's choice of employment to be less genderdetermined than the elder generations. Moreover, it is a sign that the labour market for persons with a short education is more gender segregated than for the remainder of the labour market. This trend is less prominent in Denmark than in most other OECD countries. The international comparisons show that the gross wage difference between female male full-time employees is 11 per cent in Denmark versus an international average wage gap of 16 per cent. The wage difference is also decomposed (using the Juhn-Murphy-Pierce decomposition) for the OECD countries. This decomposition occurs in relation to a benchmark constructed on the basis of unweighted salaries of men in 13 OECD countries. This decomposition provides the opportunity to break down the gender wage differences into a part, which can be explained by the wage structure, a part, which can be explained by different remuneration of productive characteristics. The OECD concludes, referring notably to Portugal, that less wage dispersion is not always so beneficial for women. Furthermore, it is indicated that the relatively low salaries among public sector employees in Belgium, Denmark, Finl Holl compared to the other OECD countries, where public employees are often better salaried, contributes to an increase in gender wage differences in these countries. An investigation of the family wage gap, here rendered as the difference in monthly wages for women with without children, shows that in Denmark the wages of women with children are apparently not affected. For the OECD countries as an aggregate, women without children average 5 per cent higher salaries than women with children. This figure, however, hides great deviations from country to country. 82

Overview of genda wage studies Purpose of Author Analysis Data Hourly wage concept Main result Naur, To explain the Cross- Rom Hourly wage The wage gap between men Rosholm causes of section sample of 1 calculated on women in the public sector, both stagnation in per cent of the the basis of white-collar workers unskilled Smith the gender for 1980 Danish total wage blue-collar workers, narrowed in (1994) wage differ- population for income the period by 4.5 percentage points ences in the 1990. 1980-1990 number of 1.3 percentage points, respec- 1980s. (14,000- employed tively to 18.3 per cent among 17,000 cases). hours (ATP). white-collar workers 8.5 per Skilled blue- cent for unskilled blue-collar collar workers workers in 1990. For white-collar not included. workers in the private sector, the wage difference was reduced by 3.4 percentage points to 29.7 per cent. The wage difference among unskilled blue-collar workers, however, increased by 1 percentage point to 18.4 per cent. Only for the group of public sector white-collar workers to a certain degree white-collar workers in the private sector is it possible to explain the trends in wage difference on the basis of a change in characteristics. For unskilled blue-collar workers, the effect of the changed background characteristics is close to zero. Several explanations are given to account for the stagnation: 1) changing governments wage conflict policy aimed at limiting growth in public sector wages compared to the private sector; 2) decentralisation of wage setting; 3) elimination of automatic wage indexing. 83

Purpose of Author Analysis Data Hourly wage concept Main result Rosholm To explain Cross- Rom Hourly wage Even though a significant decline causes of section sample of 1 calculated on has taken place in the differences in Smith stagnation in per cent of the the basis of background characteristics (educa- (1996) the gender for 1980 Danish total wage tion, job experience, etc.) between wage differ- population for income men women, only a small ence in the 1990. 1980-1990 number of reduction in the wage gap occurs 1980s. (14,000- employed during the period 1980 to 1990. 17,000 cases). hours (ATP). For the group of unskilled workers Skilled blue- in the private sector, wage differ- collar workers ences between men women not included. increased. Generally, women receive a lower return on education, job experience, etc. Explanations offered for the lack of reduction in the wage difference between men women are similar to those cited above. 84

Purpose of Author Analysis Data Hourly wage concept Main result Pedersen To describe Cross- Approx. Three hourly The wage gap on the private labour factors affect- section 940,000 cases wage concepts market is between 16.1 per cent Deding ing the from DST are used: 1) 17.0 per cent depending on the (2000) differences in for 1996. wage statistics direct remu- salary concept utilised, of which the hourly wage for 1996. neration; 2) unexplained segment comprises income to Persons with total earn- between 12.1 12.6 percentage summarise loose connec- ings/paid points. possible tion to the hours 3) A special of wage difference unexplained labour market total earn- within nine sectors shows that gross wage differ- are excluded. ings/unit time wage differences between men ences. Branch worked. women are greatest within the job functions financial services sector, 23.0 per used as cent-25.6 per cent, in the retail, explanatory hotel restaurant branch, with variables at the 21.9-23.6 percentage points. The aggregate level. largest proportion of unexplained wage differences is found within retail, hotel restaurant sector, 14.6-15.3 percentage points, in building construction, with 12.4-14.3 percentage points. 85

Purpose of Author Analysis Data Hourly wage concept Main result Gupta, To analyse the Cross- Rom Hourly wage Estimates of wage functions for the Oaxaca effect of the section sample of 0.5 calculated on years 1983, 1989 1994 show wage forma- per cent of the the basis of that there occurs a small narrowing Smith tion process on for 1983, Danish total wage of the wage gap in the public sector (2001) the wage gap. 1989 population for income of less than 1 percentage point in 1994. 1983-1994. number of this period to 25.1 per cent. In the The employed private sector, the wage gap in- includes only hours (ATP). creases by 1 percentage point to white-collar 36.4 per cent. The decomposition workers in the shows that the stagnation of the public wage gap in the public private private sectors. sectors from 1983 to 1994 has different explanations. In both sectors, women improve their qualifications compared to men. In the private sector, however, the changes in qualifications are offset by the changes in the relative prices for qualifications (especially experience education). In the public sector, the effect of improved background characteristics was dissolved by the greater wage dispersion. Gupta Analyse Longitu- 0.5 per cent Hourly wage Births have an immediate negative association dinal- sample of the calculated on effect on women s wages, in that Smith between wages. Danish the basis of women s return on job experience (2002) childbirths population of total wage following childbirth is lower than (family wage 0.5 per cent income before the first birth. If women gap). for 1980- number of have also been on childcare leave 1995. Covers employed after the maternity period, there full-time hours (ATP). will be a further effect of lacking employees. accumulation of experience. 86

Purpose of Author Analysis Data Hourly wage concept Main result Skyt Analyse Cross- Rom Hourly wage The shows that childbirth Nielsen, association section sample of the calculated on has varying consequences for Simonsen between wages Danish the basis of women employed in the private childbirths for 1997. population of total wage versus the public sectors. Women in Verner (family wage 0.5 per cent in income the private sector who have children (2002) gap) in the 1997. Also number of obtain a lower salary than do private analyses a sub- employed women without children. For public sectors. sample of hours (ATP). women working in public sector women aged employment, the effect of births on 20-40 working their salary is of transitory charac- in public ter. private sectors. 87

Purpose of Author Analysis Data Hourly wage concept Main result OECD (2002) To describe associations between wages gender, analysed according to age, education, etc. Crosstabulations crosssectional analyses Data come from the ECHIP for 1998. Hourly wage calculated as gross salary divided by hourly norm in the given country. Seventy-five per cent of the women in the labour market in the OECD countries are employed in 19 occupations out of the total 114 job functions. In these occupations, the average proportion of women in the respective positions is 70 per cent. Similar figures for men are 30 occupations 73 per cent. The Danish figures resemble the average figures for the OECD. In several countries, including Denmark, there is a tendency for the youth s choice of employment to be less gender-determined. International comparisons show that the gross wage gap between men women for full-time employees is 11 per cent in Denmark, versus an international average of 16 per cent. In connection with a decomposition of the wage difference between men women, it is indicated that the relatively lower wages in the public sector in Belgium, Denmark, Finl Holl compared to other OECD countries, where public employees are often better salaried, contributes to an increase in the wage differences between men women in these countries. A survey of the family wage gap among Danish women with children shows no effect on their wages. 88