Presentation on theme: "Earnings Differences Between Ethnic Groups: Evidence from the LFS * Ken Clark University of Manchester Stephen Drinkwater University of Surrey November."— Presentation transcript:
Earnings Differences Between Ethnic Groups: Evidence from the LFS * Ken Clark University of Manchester Stephen Drinkwater University of Surrey November 2006 * Funding from the Joseph Rowntree Foundation is gratefully acknowledged
Some optimistic signs in terms of the labour market position of ethnic minorities Data for 1991 (SARs), 2001 (SARs), 2002-5 (LFS) shows some convergence in employment rates (excluding students), especially for males Also a faster rate of growth for each male minority group in the latter period, apart from Others However employment rates for some female ethnic minority groups remain very low Employment rates also fell for all female ethnic minority groups in 2002-5, whereas they rose for Whites Background (1)
However, ethnic minorities (especially males) have experienced persistent earnings disadvantage Blackaby et al. (1994, 2002) report percentage earnings differentials of 7% for the 1970s, 12% in the 1980s and 10% in the 1990s LFS data for 2002-5 reveal that it was still 11% Also large differences between ethnic minority groups, especially for males The majority of these differences cannot be explained in terms of characteristics => discrimination? Background (2)
Occupation has typically been ignored by economists in examining wage differentials Mainly because of its potential endogeneity Some exceptions: De Beyer and Knight (1989) argue that occupation is important in determining wages in general terms Stewart (1983) compares ethnic wage differences using occupational earnings Elliott and Lindley (2006) analyse earnings differences by occupation for white and non-white immigrants We investigate the importance of occupation in generating ethnic wage differences because of possible occupational segregation and the lack of ethnic minorities in top executive jobs (DTI, 2004) Background (3)
Pooled LFS from 2002 to 2005 Our sample consists of: Wave 1 respondents only Working age employees excluding full time students We use (deflated) hourly wages as our measure of earnings We examine the position of 7 ethnic minority groups relative to Whites, separately for males and females Data
We estimate standard (augmented Mincerian) wage equations: We can then calculate percentage wage differentials for each ethnic group relative to Whites using: We then observe what impact including occupational controls (at the 1 and 4 digit levels) has, i.e.: We also estimate the first equation separately for three occupational groupings (Professional/Managerial, Skilled and Routine/Semi-Routine) Methodology
Large earnings differentials for males when personal and job characteristics are controlled for Variables such as region exert an important effect compared to the raw differences Controlling for 1 digit SOC reduces the differentials quite a bit for most male ethnic groups (especially Black Afr.) But replacing these with 4 digit controls does not reduce them much further and even increases the differential for Indians Smaller differentials for females But controlling for 1 digit SOC roughly halves the differential for virtually all of the groups Again 4 digit controls dont make much of a difference Results (1)
Percentage Differences in Male Earnings Relative to Whites No Occ. Vars1 Digit Occ.4 Digit Occ. Black Caribbean-14.0-10.4-8.6 Black African-26.3-15.3-10.9 Indian-15.1-10.4-11.0 Pakistani-20.5-14.1-12.8 Bangladeshi-27.0-20.9-19.8 Chinese-9.6-8.8-8.7 Other-16.2-11.1-9.9 N54934 Models also include controls for education, experience, marital status, region, year, immigrant cohort, industry, firm size, part-time, tenure and public sector.
Percentage Differences in Female Earnings Relative to Whites No Occ. Vars1 Digit Occ.4 Digit Occ. Black Caribbean-5.4-2.5-1.5 Black African-18.0-8.5-7.0 Indian-10.9-5.7-5.4 Pakistani-11.0-6.0-4.8 Bangladeshi-14.4-7.0-6.1 Chinese-1.4-0.2-0.9 Other-5.1-3.3-2.5 N56358 Models also include controls for education, experience, marital status, region, year, immigrant cohort, industry, firm size, part-time, tenure and public sector.
The largest earnings differentials exist for the higher level occupations for all male ethnic minority groups, apart from Indians Differentials also tend to fall by occupation Particularly noticeable for Black Africans and Pakistanis This pattern is less pronounced for females Professional and Managerial Chinese even have an earnings advantage over Whites But again relatively large occupational variations for Black Africans Results (2)
Percentage Differences in Male Earnings Relative to Whites by Occupational Group Prof./Man.SkilledRout./Semi Black Caribbean-12.0-9.2-6.5 Black African-24.8-10.0-2.5 Indian-8.6-12.5-11.1 Pakistani-19.4-15.1-7.9 Bangladeshi-24.7-19.2-17.1 Chinese-13.82.1-1.7 Other-14.4-7.0-7.8 N260491384615039 Models also include controls for education, experience, marital status, region, year, immigrant cohort, industry, firm size, part-time, tenure and public sector.
Percentage Differences in Female Earnings by Occupational Group Models also include controls for education, experience, marital status, region, year, immigrant cohort, industry, firm size, part-time, tenure and public sector. Prof./Man.SkilledRout./Semi Black Caribbean-6.3-0.8 Black African-16.0-11.3-1.5 Indian-6.5-6.0-4.3 Pakistani-0.7-9.6-5.8 Bangladeshi-16.17.6-1.2 Chinese2.31.8-7.2 Other-6.0-2.43.4 N234231610316832
There continues to be large earnings deficits relative to Whites for most ethnic minority groups, in particular for males Controlling for occupation reduces ethnic wage differences quite substantially for both males and females, even by just including 1 digit controls More detailed occupational controls do not have much of an additional effect Earnings differentials are largest for higher-level occupations for most groups and the impact of occupation is very large for some e.g. Black Africans Conclusions