LFS/APS user meeting 2 Dec Is ethnicity or religion more important in explaining inequalities in the labour market? Jean Martin Anthony Heath University of Oxford Karin Bosveld Office for National Statistics
LFS/APS user meeting 2 Dec Previous research Lots of studies have looked at disadvantage by ethnic group Ethnic penalty – differences which persist after controlling for differences between ethnic groups in age, qualifications etc. Few studies have looked at religion as well Not based on recent data/restricted to certain ethnic groups
LFS/APS user meeting 2 Dec Why might religion be important? An indication of cultural differences within an ethnic group Direct impact through religious beliefs &/or practices Lack of social networks linking to wide range of employment Chill factor Basis for discrimination by employers
LFS/APS user meeting 2 Dec Research questions Are there differences between religious groups that are consistent across ethnic groups? Are effects less marked among those born in UK than among first generation immigrants? Are there differences between men and women?
LFS/APS user meeting 2 Dec Identification problem Cant always distinguish ethnicity and religion Eg All Pakistanis and Bangladeshis are Muslim Wont know whether differences are due to culture or religion narrowly defined
LFS/APS user meeting 2 Dec Why use the APS? Interest in minority ethnic groups and religious differences within them means need very large sample Census/LS data now quite old – not enough of new immigrant groups like Black Africans APS has year of arrival in UK But doesnt have a measure of English language proficiency and doesnt code foreign qualifications
LFS/APS user meeting 2 Dec APS under special licence Need for fullest detail of ethnic and religious groups Plans for including detailed geographical area information (but not done yet) Disclosure risks
LFS/APS user meeting 2 Dec Data for analysis Combined 2005 & 2006 APS Working age population Excluding FT students Excluding 2005 boost sample which wasnt asked about qualifications men and women
LFS/APS user meeting 2 Dec
10
LFS/APS user meeting 2 Dec Women
LFS/APS user meeting 2 Dec Women
LFS/APS user meeting 2 Dec Identification problem
LFS/APS user meeting 2 Dec Outcome variables Proportion economically inactive Proportion unemployed of the economically active Not employment rates because affected by ratio of inactive to unemployed which varies by ethnic group
LFS/APS user meeting 2 Dec Independent variables Separate analyses for men and women Age and age squared Qualifications: –degree, other, none Whether born in UK; if not, age at arrival (<16, 16-30, over 30) For women: whether married, whether has dependent children 2 nd generation = born in UK or arrived before age 5
LFS/APS user meeting 2 Dec Logistic regression models Model 1: ethnicity alone Model 2: reduced ethnic group and religion variables to deal with sparse cells – 7 ethnic and 3 religion groups Model 3: combined ethno-religious group variable All with the same control variables
LFS/APS user meeting 2 Dec Women' s economic inactivity: log-odds
LFS/APS user meeting 2 Dec Women' s economic inactivity: log-odds
LFS/APS user meeting 2 Dec Women' s economic inactivity: log-odds
LFS/APS user meeting 2 Dec Economic inactivity Muslim women very much more likely to be economically inactive Other Asian non-Christians less likely to be active than Christians Black African very diverse: Christians more likely to be active than white British Christians but Muslims less likely
LFS/APS user meeting 2 Dec Women' s unemployment: log-odds
LFS/APS user meeting 2 Dec Women' s unemployment: log-odds
LFS/APS user meeting 2 Dec Women' s unemployment: log-odds
LFS/APS user meeting 2 Dec Womens unemployment Muslims significantly more likely to be unemployed in all groups (except Other White) Among Indians Muslims, Hindus and Sikhs more likely to be unemployed than Christian or no religion Coeffs for Black Caribbean and Black African show strong ethnic effect on unemployment
LFS/APS user meeting 2 Dec Difference in log-odds between all women and 2nd generation: economic inactivity
LFS/APS user meeting 2 Dec Second generation women economic inactivity 2 nd gen less different from women overall for most groups Muslims still more likely to be economically inactive but coefficients reduced Economic inactivity increased for Indian and Black African Christians
LFS/APS user meeting 2 Dec Difference in log-odds between all women and 2nd generation: unemployment
LFS/APS user meeting 2 Dec Second generation women unemployment 2 nd generation women generally less likely to be unemployed than women in general but some exceptions Muslims still more likely to be unemployed but coefficients reduced Black Caribbean and Black Africans still more likely to be unemployed
LFS/APS user meeting 2 Dec Conclusions Both ethnicity and religion are important in explaining differences in economic inactivity and unemployment rates among women Muslim women more likely to be inactive, or if active, to be unemployed than other religions Black Caribbean and Black African women more likely to be unemployed, irrespective of religion Main differences still apparent in the second generation
LFS/APS user meeting 2 Dec But…. We dont know what the mechanisms are for the differences Havent controlled for all the factors which might be relevant (eg language) Cultural differences might explain differences in economic activity but not clear how they would explain differences in unemployment Clear that there are major barriers to employment for Muslim women