Can religious affiliation explain ‘ethnic’ inequalities in the labour market?

Abstract

The disadvantage experienced in the labour market by the ethnic groups currently in Britain has long been established. This study builds on earlier research by exploring how far religious affiliation can explain these inequalities. Distinguishing the effects of ethnicity and religion is difficult for both conceptual and technical reasons, due to the ‘identification’ problem. We therefore estimate the effects by focusing on cases where ethnicity and religion cross-cut each other, excluding those groups where ethnicity and religion coincide. The results demonstrate a strong ‘Muslim penalty’ for both women and men from different ethnic groups with respect to economic activity and unemployment. Furthermore, the apparent ‘ethnic’ effects estimated for Pakistanis and Bangladeshis in models that do not control for religion prove to be similar in magnitude to the effect of Muslim religion estimated in models from which Pakistanis and Bangladeshis have been excluded.

Keywords:

• Ethnicity
• religion
• labour market
• inequalities
• Muslims
• black Caribbeans

Acknowledgements

The authors thank the UK Data Archive for provision of the APS data used in this paper. We also thank Karin Bosveld for her help in the early stages of the work and the anonymous reviewers for their helpful comments.

Notes

1. We also excluded a small number of individuals who, while not coded as full-time students, gave as their reason for being economically inactive that they were in fact students – although presumably not full time).

2. We found that arriving before the age of five had no impact on economic activity compared with those born in the UK and therefore we include them with the second generation.

3. We also checked our results assigning excluded respondents in Models 2 and 3 to an additional residual category and thus employing the same n throughout as in Model 1. The results were effectively identical to those reported in the text and tables.

4. We need to recognize, however, that the smaller n for the unemployment analysis can be expected to be associated with greater variability in the magnitude of the estimates.