Originally published here.

Labour market inequalities between ethnic groups and between areas are a priorityfocus for governments whose intention is to reduce the disadvantage which thoseinequalities may represent. However, evidence shows that inequalities between ethnicminorities persist and the ethnic minorities face unfair disadvantage anddiscrimination in the labour market (Berthoud 2000; Cabinet office 2003; Dale et al. 2002; Esmail and Everington 1997; Heath and Cheung 2006, Modood et al. 1997).

Average inequalities between populations or areas are the result of a variety ofsocietal and personal investments in training, provision of jobs, success in turningqualifications into employment, and the impact of demographic composition such asage structure and birthplace, reviewed for Great Britain by Heath and Yu (2005). Raw differences between populations can be seen as the cumulative impact ofdisadvantages, including but not limited to discrimination in the labour market. Theraw differences between groups are an important indicator of social cleavages.

Many social scientists attempt to quantify the extent to which raw differences betweengroups can be identified with specific factors that create them (Berthoud, 2000;Borjas, 1995; Heath and McMahon, 1997; Cheung and Heath, 2005).Demographic composition and human capital are the two factors that past studiesconsistently identify as influences on labour market outcomes. If a population is particularly young then a relatively large proportion will be studying rather thanworking; a high proportion born overseas might on average suggest inexperience andlack of confidence in the labour market.

Such demographic compositional factors arenot easy to change, although their impact may be ameliorated for example by provision of English classes for those whose first language is not English. Humancapital – principally the skills an individual brings to the labour market – is alsoknown to have a major impact on success in the labour market. Generally, the greateran individual’s qualifications the more likely he or she is to gain a job and remaineconomically active (Heath and McMahon, 1997, show this for England and Wales in1991, as this paper does for 2001; Hirschman and Snipp 1999 show similar findingsfor the United States).

Thus the extent to which ethnic minorities have gainedqualifications may account for some of the average labour market inequalities found between ethnic populations. That component of inequalities can be targeted withresources to improve and equalise the educational opportunities of each population.The evidence in Britain suggests that at least one half of the average differences inlabour market outcomes for ethnic groups can be attributed to their composition, suchthat those with lower outcomes are on average younger and have fewer qualificationsor have other individual circumstances that disadvantage them (Leslie et al., 2001;Carmichael and Woods, 2000).

The remaining differences are often termed ‘ethnic penalties’.The database described and used in this paper measures ethnic penalties nationallyand for each neighbourhood of England and Wales (we occasionally use the shorthand‘national’ to refer to England and Wales). The evidence shows whether a localoutcome that is different from the national is consistent with local demographiccomposition and human capital. The part of local outcomes which is not consistentwith local composition can be thought of as a neighbourhood ethnic penalty and istermed a ‘neighbourhood effect’.

It indicates local factors that create a worse or betteroutcome than would be expected from the usual impact of demography and humancapital. The nature of these local factors remains hidden, but the identification of areasand ethnic groups where such factors are having an impact is a means of targeting and prioritising the need for investigation and remedial measures. Durlauf (2004) reviewsthe concepts and varied approaches to analysis of neighbourhood effects.

Here we areconcerned not with the mechanisms by which neighbourhood effects operate, but toestimate the size of their impact relative to the impacts of ethnic penalties andindividual characteristics that operate across all areas of residence. Clark andDrinkwater (2002) examine neighbourhood effects for ethic minorities in Britain fromdata for the early 1990s and find they are similar to those for the White majority,across differet levels of neighbourhood ethnic concentration.

The paper begins with examples of the raw neighbourhood labour market outcomeswhich motivate the analyses of this paper. Data estimation methods for the evidence base and jobs deficits are specified in a section on data and methods. The paper thendescribes the influence on employment rates of age, sex, qualifications and country of birth, for each ethnic group defined in the Census, using data for England and Wales as a whole.

These patterns are used to account for variation in local outcomes that areconsistent with local demographic composition and human capital. Local jobs deficitsare estimated in the following section, taking into account neighbourhoodcomposition in order to derive the remaining neighbourhood effects. The geographical pattern of these neighbourhood effects is compared across ethnic groups throughmeasures of correlation and regional values.

The empirical starting point of this paper is the national impact of qualifications anddemographic characteristics on the labour market outcome of each ethnic group. Wehave measured this impact for England and Wales with the latest 2001 Censusmicrodata. The results confirm other studies’ findings that young people, women and people without qualifications are less likely to be employed, and that there arevariations between ethnic groups, principally the relatively high employment rates ofCaribbean women and the relatively low employment rates of Pakistani andBangladeshi women; each ethnic minority group has lower rates of male employmentthan the White Briton average, but to differing extents. Qualifications raise theemployment rate of each group, but the impact of qualifications is less for those bornoutside the UK, the majority of whom will also have been educated outside the UK.The analysis proceeds to new ground by applying these national relationships tointerpret local labour market outcomes from the latest Census. It has assessed thecontribution of human capital and demographic characteristics to the geography ofemployment, and distinguished it from the remaining neighbourhood effects.

The paper has defined and used the jobs deficit as a measure of the impact of lowemployment in a locality, for each ethnic group. The estimation of jobs deficits has anumber of benefits. It provides expected employment outcomes to compare withobserved values not only for each locality but for each sub-population defined by age,sex, ethnic group, birthplace and qualifications. It is based on the fully saturatedmodel, so that each interaction of those variables with employment is fully used toassess local expected values. The jobs deficit neatly combines the size of a local population with low rates of employment in a way that can also be summed acrossneighbourhoods. Finally, the jobs deficit is a measure that is readily understood inrelation to policy objectives equalising employment outcomes across social groupsand localities.Overall 1.1 million jobs are needed to bring employment for each population in eachneighbourhood up to the current England and Wales average. 1.4 million jobs would be needed to bring employment up still higher to reach to the White Briton average,for every group in every neighbourhood. For some ethnic minority groups, jobsdeficits are large relative to their population of working age, implying an addition of thirty percentage points to Pakistani and Bangladeshi employment, and an addition of between 10 and 20 percentage points to Chinese and African employment.

However,in absolute terms, half the total jobs deficit is among local White Briton populations.The jobs deficit highlights local White lack of employment in the same way and onthe same scale as other groups.A population’s jobs deficit is greatly reduced when measured against its own averagein England and Wales. Thus local replication of national inequalities accounts formost of the jobs deficits among ethnic minority groups. In the extreme case, thePakistani and Bangladeshi local jobs deficits would each drop to one ninth of theircurrent value if local employment rates were increased by the national deficit.

This local impact of national inequalities overshadows but does not eliminate the impact oflocal demographic composition and qualifications. The geography of qualifications, birthplace, sex and age structure does impose further local jobs deficits for each ethnicgroup: many areas with low unemployment have that condition partly because theirresidents are les well-qualified, or younger than the residents of other areas. Amongethnic minorities, local deficits of 32 thousand jobs are accounted for in this way.

The remaining local disadvantage is not related to the measured individualcharacteristics of residents and has been termed the neighbourhood effect. It isconsiderable in extent and larger than the impact of local composition measured bysex, age, qualifications and birthplace. Its nature however may be structural,contextual, or compositional (Blalock 1984; Curtis and Rees, 1998). A structuraleconomic effect might impose a lack of local jobs. A contextual effect would suggestthat the local area’s composition affects the employment of residents irrespective oftheir own characteristics.

Thus a generally low employment level may make it harderto find jobs because of poorer social networks and lower expectations of work. On theother hand a compositional effect would simply be the concentration of individualswith poor labour market outcomes for reasons that have not been measured. Thesemight include selection effects: the unemployed tend to concentrate in areas of poorerhousing. Thus neighbourhood effects are not necessarily structural rather thancompositional in nature, but do identify where there are particular problems faced bysignificant numbers of people, that are not accounted for by their level ofqualifications or demographic characteristics.

You can read the complete article here.