Richard Berthoud - iser.essex.ac.uk · Author: Richard Berthoud, Institute for Social and Economic...

INSTITUTE FOR SOCIAL AND

ECONOMIC RESEARCH

Richard Berthoud

The Incomes of EthnicMinorities

Author: Richard Berthoud, Institute for Social and Economic Research

The Joseph Rowntree Foundation has supported this project as part of itsprogramme of research and innovative development projects, which it hopeswill be of value to policy makers and practitioners. The facts presented and theviews expressed in the report, however, are those of the authors and notnecessarily those of the Foundation.

Readers wishing to cite this document are asked to use the following form of words:

Berthoud, R. (1998) The Incomes of Ethnic Minorities. ISER Report 98-1. Colchester:University of Essex, Institute for Social and Economic Research.

October 1998

ISSN: 1464-8350ISBN: 1 85871 2009

© October 1998

All rights reserved. No part of this publication may be reproduced, stored in a retrieval systemor transmitted, in any form, or by any means, electronic, mechanical, photocopying, recordingor otherwise, without the prior permission of the Institute for Social and Economic Research.

ISER Report 98-1 The Incomes of Ethnic Minorities

The Incomes of EthnicMinorities

Richard Berthoud

Table of Contents

INTRODUCTION ..............................................................1

PURPOSE................................................................................1ETHNIC MINORITIES AND ECONOMIC DISADVANTAGE ............1TWO NEW SOURCES OF INFORMATION....................................6

The Fourth National Survey of Ethnic Minorities............7The Family Resources Survey..........................................7

1. THE COMPOSITION OF FAMILYINCOMES...........................................................................9

BACKGROUND .......................................................................9INTRODUCING THE ANALYSIS.................................................9STAGE IN THE LIFE-CYCLE ...................................................11THE INCOMES OF WORKING FAMILIES ..................................13NON-WORKING FAMILIES BELOW 60 ....................................20PENSIONERS ........................................................................28COMPARING FAMILY TYPES .................................................30

2. LEVELS OF HOUSEHOLD INCOME......................33

BACKGROUND .....................................................................33INTRODUCING THE ANALYSIS...............................................33COMPARING ETHNIC GROUPS...............................................34‘POVERTY’ ..........................................................................35

Mixed households ..........................................................37Alternative measures of poverty.....................................38Adjusting for household size ..........................................39

GEOGRAPHICAL VARIATIONS IN THE DISTRIBUTION

OF INCOME...........................................................................42

CONCLUSIONS...............................................................47

REVIEW OF THE FINDINGS ....................................................47Chinese ..........................................................................47Indians ...........................................................................48Caribbeans.....................................................................48Africans..........................................................................49Pakistanis and Bangladeshis .........................................49

DISCUSSION.........................................................................49Poverty among Pakistanis and Bangladeshis ................50The role of the social security system ............................52Inequality in multi-cultural Britain................................53

APPENDICES...................................................................56

A: COMPARING THE FAMILY RESOURCES SURVEY

WITH THE FOURTH NATIONAL SURVEY OF ETHNIC

MINORITIES .........................................................................56Composition of the samples ...........................................56Response rates for the income questions .......................56Distributions of earnings and of income amongwhite households............................................................57Comparisons between ethnic groups .............................59

B: BASELINE NEEDS AND ‘AVAILABLE’ INCOME.................61REFERENCES .......................................................................62

Acknowledgements

This research was funded by the Joseph Rowntree Foundation. It is one of four detailedstudies of minority groups undertaken with JRF support, to follow up the broad findings ofthe major Fourth National Survey of Ethnic Minorities (Modood, Berthoud and others 1997).The other three studies are:

Ethnic minority families (Tariq Modood, Sharon Beishon and Satnam Virdee)Ethnic minorities in the inner city (Richard Dorsett and Richard Berthoud)Young black men in the labour market (Richard Berthoud)

The data from the Family Resources Survey were provided by the Department of SocialSecurity. Thanks to Jo Semmence and Margaret Frosztega for providing and explaining datafiles and computer programs, and to Simon Gault for an analysis of non-response.

The research started when the author was at the Policy Studies Institute. Bernard Caseycontributed to the design of the study and the early stages of analysis. John Forth made amajor contribution by extracting data from the huge survey and transcribing it into a formwhich could be easily analysed.

Thanks to John Hills, Tariq Modood, Ceri Peach, Liz Tadd and John Veit-Wilson forcomments on earlier drafts.

As always, the analysis and interpretation of the data are the author’s responsibility, and donot represent the views of the Joseph Rowntree Foundation, the Department of SocialSecurity or the other contributors just mentioned.

1

The Incomes of Ethnic Minorities

INTRODUCTION

Purpose

The disadvantaged economic position ofminority ethnic groups has been a matter ofconcern since the main post-war migrationstarted in the 1950s. W. W. Daniel’s pioneer-ing survey of ethnic minorities (Daniel 1968)showed, for example, the extent of un-employment among West Indian and Asianmen — in spite of the fact that it was labourshortages which had attracted migrants to thiscountry in the first place. Those who had jobsgenerally occupied low skill and low statuspositions — and such data as were availableabout earnings suggested that they were lowpaid as well.

Subsequent research, especially the second andthird national surveys of ethnic minorities(Smith 1976, Brown 1984) showed that un-employment and low pay persisted, althoughthe pattern of disadvantage was not uniform.But until recently, one vital piece of inform-ation has been missing: data on employmentand earnings may have suggested that certainminorities probably experienced low incomesand poverty, but there has been no directmeasure of household incomes to show exactlyhow minorities compared with the rest of thepopulation.

That gap has been filled recently by two majornew enquiries.

• The Fourth National Survey of EthnicMinorities, conducted in 1994, covered arepresentative sample of 3,315 house-holdsof Caribbean, South Asian and Chineseorigin, and 2,867 white house-holds. Itincluded a question in which one personreported the household’s total net income.

• Meanwhile the Department of SocialSecurity has launched a Family Resources

Survey covering 25,000 households eachyear. Over the two years 1994/5 and 1995/6there were 2,520 households from minoritygroups. Every member of these house-holdswas asked detailed questions about allelements of their incomes.

These surveys are described more fully later inthis introduction, and a direct comparison of theresults of key questions is in Appendix A.

We therefore have a specialist survey of ethnicminorities with a simple income question; and aspecialist survey of incomes, which includesethnic minorities. Confidence in the findings ofeach survey will be enhanced if they both leadto the same conclusions. The results of theFourth Survey income question have alreadybeen analysed in the main report on that study(Modood, Berthoud and others 1997, Chapter5)1. This new report analyses the more detailedincome data from the Family Resources Survey.

The objective is to complement the findings ofthe Fourth Survey, and complete the picture ofthe incomes of ethnic minorities. The report isin two main sections: a detailed analysis of thecomposition of the incomes of ethnic minor-ities in different types of family; and an analysisof total incomes, focusing especially on lowincome households.

Ethnic minorities and economicdisadvantage

The 1991 Census, the first to include a directquestion on ethnicity, recorded that non-whiteminorities made up 5½ per cent of the

1 To avoid repetition of the full reference, the report ofthe Fourth Survey will be referred to as EMiB (EthnicMinorities in Britain) from now on.

2

ISER Report 98-1

Table 1: Demographic characteristics of ethnic groupsColumn percentages

Source: Fourth National Survey of Ethnic Minorities

population of Great Britain. The numbers in theprincipal groups analysed in this study were:

Caribbean2 500,000African 212,000Black other2 178,000Indian 840,000Pakistani3 477,000Bangladeshi3 163,000Other4 488,000

2 A very large proportion of members of the Censuscategory ‘black other’ are British born people withparents of Caribbean origin. In this report, as in theFourth Survey, a single group will be analysed.3 In this report, Pakistanis and Bangladeshis are alwayscombined, because there are too few of them forseparate analysis. The Fourth Survey showed that theireconomic positions were very similar.4 Includes Census categories ‘Asian other’ and ‘Other’.This category covers a very wide range of origins, andcannot be considered an ethnic group in its own right.They are shown in tables in this report for the sake ofcompleteness, but are not commented upon in the text.

Some basic facts about the ethnic minoritypopulations are summarised here to provideessential background to the analysis of theirincomes. Table 1 repeats some key demo-graphic findings from the Fourth Survey (seeEMiB, Chapter 2). The first two lines of thetable illustrate the medium-term consequencesof a period of migration. While almost a quarterof white adults were over the age of 60, theproportion of elderly people was much lower inminority communities, mainly because few ofthose who migrated as young men and womenhad lived in Britain long enough to attain thatage. Nevertheless half of Caribbean adults, anda quarter of South Asian adults, were born inthis country. Both groups are in a stage oftransition between a migrant and a long-termresident population. Caribbeans, having beenhere longer, are in a later stage of that transition— almost of all their children were born inBritain, many of them now of the secondgeneration.

There have been substantial changes in familystructures in Britain over the past 25 years. The

White Caribbean Indian Pakistani/Bangladesh

i

Chinese

Proportion of adults who wereover 60 24 15 12 7 7Proportion of those aged 16 to 59who were born in Britain na 53 26 24 14Proportion of those aged 16 to 59who were married 60 39 72 74 62Proportion of families with childrenwhich were lone parents 21 45 8 8 naProportion of families with childrenwhich had four or more 4 7 8 35 3Average number of adults perhousehold 1.9 1.9 2.8 3.0 2.4Proportion of adults who wereMuslims 0 1 9 96 0

3


two main blocks of minority groups each deviate from the white pattern, but in differentdirections: Caribbeans might be said to beahead of the white trend, while South Asiansare behind it. These points are illustrated inlines three to six of Table 1. Caribbeans weremuch less likely to be married than their whitecounterparts. A very high proportion ofCaribbeans with children were in one parentfamilies. In contrast, South Asians were morelikely to be married than a white person ofabout the same age. There were few Asian oneparent families. Among Pakistanis and Bangla-deshis, family sizes were much larger than isnow normal in Britain. Another feature ofSouth Asian communities is that many adultslived with their parents, even after they them-selves have married, and this leads to relativelylarge households, even without counting thechildren.

These variations in family structure will lead todifferences in households’ income and expend-iture. In particular, the majority of one parentfamilies in Britain claim basic social securitybenefits, and have low incomes (McKay andMarsh 1994); this is likely to affect the welfareof Caribbean families. The large householdsizes in Asian communities could increase thepotential number of workers bringing in anincome; on the other hand, it also means anincrease in the number of people depending onthat income.

4

ISER Report 98-1

Another demographic point worth mentioningis people’s religion. Most Indians were Hindusor Sikhs, but about one in ten was Muslim.Almost all Pakistanis and Bangladeshis wereMuslim. This is relevant to an analysis offamily incomes for two reasons — Islam isassociated with large family sizes, and with lowlevels of employment among women.

Table 2 (also derived from the Fourth Survey)summarises some of the key facts about theemployment and earnings of ethnic minoritygroups — facts which have an obvious anddirect link with the interest of this study inhousehold incomes. Looking first at men, in thetop half of the table, 15 per cent of

economically active white men were unem-ployed. The proportion was similar for Indianmen, and actually lower for Chinese men. Butthe rate of unemployment was twice as high forCaribbeans as it was for whites; and higher stillfor Pakistanis and Bangladeshis. In these moredisadvantaged groups, there are some sub-groups — young Caribbean men withoutqualifications; middle-aged Pakistanis orBangladeshis — for whom the number un-employed exceeded the number with a job.

There were also some variations in men’searnings, which broadly reflect the pattern ofeconomic success indicated by the employmentrates. Chinese men averaged more than their

Table 2: Employment characteristics of ethnic minoritiesColumn percentages, pounds

White Caribbean Indian Pakistani/Bangladeshi

Chinese

MenUnemployment rate 15% 31% 17% 39% 9%Average earnings of full-timers £331 £311 £317 £220 £368WomenProportion in work (FT or PT) 51% 58% 49% 15% 59%Average earnings of full-timers £244 £270 £260 £189 £274Source: Fourth National Survey of Ethnic Minorities. The proportion of women in work counts part-timers as half

5


white counterparts; Indians, slightly less, withCaribbeans a little further behind. It wasPakistanis and Bangladeshis who stood out onthis measure, however: the average earnings ofthose with a job were only two-thirds of thosereceived by white men.

Table 2 showed that there were also ethnicvariations in women’s employment, though thepattern is rather different than that for men,because of differences in the proportion ofwomen who remain outside the paid labourforce, looking after the home or family. BothChinese women and Caribbean women wererather more likely to have a job than whitewomen; and those who did have a job,commanded higher earnings. Indian womenwere about level with white women on bothmeasures. All women had lower rates ofemployment and of earnings than men, but forthese groups, the minority women were noworse off, sometimes better off, than their whiteequivalents. The exceptions were Pakistani andBangladeshi women: very few (15 per cent)were in work, and their earnings wereexceptionally low.

The reader may have noticed that the ethnicgroups discussed in Tables 1 and 2 do notinclude ‘Africans’. The Fourth National Surveyof Ethnic Minorities (from which the tables arederived) did not include Africans, on the groundthat the term covered too wide a range of ethnicorigins to be meaningful. There were too few ineach specific African sub-group to provide anadequate sample. Africans will be included inthe FRS analysis in this report, but the resultsshould be treated with great caution. Two of themost important sources of migration fromAfrica are West African students who decidedto take jobs in Britain after they graduated, andrefugees from war-torn Ethiopia and Somalia.Measures of the incomes of such a disparategroup are of doubtful value.

Thus there are substantial variations betweenethnic groups in their experience in the labourmarket. It can be seen that these variations musthave important consequences for theirhousehold incomes. It will be helpful toconclude this introductory discussion with areview some of the key issues that have to betaken into account in the search for anexplanation for these differences.

The first is to distinguish between the con-ceptsof ‘discrimination’ and ‘disadvantage’.Discrimination was the primary issue ofconcern during the 1950s and 1960s (Daniel1968). In those days discrimination wascommon, conscious and overt — it was notunusual then to see signs and advertisementswhich specified ‘no coloureds’. Since it hasbeen outlawed by successive Race RelationsActs, it is now not overt, and may be lessconscious, though there is evidence that it didnot necessarily become less common (Brownand Gay 1985).

Direct discrimination occurs when members ofan ethnic minority applying for a job (orpromotion, or a place at a university and so on),receive less favourable treatment than whitepeople whose claim on the position is no better.It is very difficult to measure. Individuals maythink they have been discriminated againstwhen in fact they have not; equally, they maynot suspect discrim-ination when it has actuallyoccurred (Smith 1976). It can only be identifiedfor certain if the researcher knows as muchabout the candidates as the person ororganisation making the decision. Aneconometric approach has been to analysevariations in earnings, controlling for such‘legitimate’ influences as educationalqualifications and years of experience (egDenny and others 1997). Any differencesbetween groups remaining after taking accountof these influences is attributed todiscrimination. While this is certainly avaluable form of analysis, employers know far

6

ISER Report 98-1

more about their workers and job-candidatesthan the crude measures of education andexperience available in surveys. The extent ofunfair discrimination cannot therefore bemeasured with any precision by that route. Inthe absence of natural situations in which theanalyst knows as much as the decision-maker, ithas been necessary to create artificial ones.Experimental studies in which identicallyqualified candidates have applied for real job-vacancies have conclusively shown that theproportion offered an interview or a post islower for members of minorities than for whitepeople (Daniel 1968, Smith 1976, Brown andGay 1985).

While discrimination probably remains animportant influence on minority communities’economic positions, it has proved too narrow aconcept. Minority groups may be worse off thantheir white neighbours for a wide range ofreasons. Indirect discrimination (where a set ofcriteria or procedures places minorities at adisadvantage, without explicitly singling themout), factors associated with recent arrival in thecountry, changes in the occupational orindustrial structure of the economy — all ofthese could also have adverse consequences forminorities, as well as, or instead of,racist discrimination. Racial disadvantage is abroader term designed to cover the outcomes ofall these potential influences.

The important point in the current context isthat this study identifies serious disadvantage insome minority groups; but it makes no pretenceat an explanation for that dis-advantage in termsof discrimination.

The other key conceptual issue concerns thenature of people’s position as a ‘minority’ in awhite-dominated Britain. The early evidencesuggested that all the main minority groupsexperienced a similar level of disadvantage.This encouraged analysts to perceive each ofthe specific minority groups as members of an

overarching category whose primary character-istic was that they were not white. This way oflooking at things, characterised by the use of theword ‘black’ to embrace all minorities,emphasised the sub-groups’ shared experienceof racist exclusion (Modood 1994). Thatremains an important common factor, but morerecent theoretical and empirical research haspointed to a diversity between groups whichcannot be subsumed in a single classification.At a theoretical level, it has been argued that amerger of all non-white groups deprives themof any character of their own — they aredefined by their relation to the white majority.‘Multi-culturalism’ is about acknowledging andvaluing the diverse contributions of each groupto the new Britain. In empirical terms, Table 1has already shown that the family structures ofCaribbean and South Asian communities aremore different from each other than from thewhite pattern. Table 2 and the remainder of thisreport show that there are substantial variationsbetween groups in employment, earnings andincome. Chinese and Indian households areclose to their white counterparts; on somemeasures, they are ahead. But Pakistanis andBangladeshis are the poorest groups in thecountry. These patterns cannot possibly beexplained on the basis that all are members ofthe same group, differentiated only from thewhite population by the colour of their skin. Amore sophisticated explanation is required forthese diverse outcomes.

Two new sources of information

Research on ethnic minorities is hindered bythe difficulty of locating a large and represent-ative sample, and by the fact that manymembers of some minority groups do notspeak English fluently. Research on earningsand incomes5 is hindered by the difficulty of

5 Throughout this study, we refer to ‘earnings’ as thegross pay received by an individual from his or her job,while ‘income’ is the total amount of money received by

7


obtaining accurate information from surveyrespondents about topics which are oftenregarded as confidential. In the absence of asource of information which addresses both ofthese problems at the same time, it isnecessary to look for either a specialist surveyof minorities which has adequate data onearnings and incomes, or a specialist incomesurvey which has an adequate sample ofminorities. Neither of these options hasprovided a very satisfactory solution in thepast, but two new sources have becomeavailable recently.

The Fourth National Survey of EthnicMinorities

Substantial specialist surveys of ethnic minor-ities were undertaken by the Policy StudiesInstitute in each of the past three decades(Daniel 1968, Smith 1976 and Brown 1984).All three surveys collected data about thewages and salaries of those in work, andevidence about ethnic bias in the distributionof earnings has been an important outcome ofthe series of studies. None of them, though,provided information about the total income ofhouseholds, combining earnings, socialsecurity receipts and other sources for allhousehold members.

The Fourth National Survey of Ethnic Minor-ities was carried out by PSI and SCPR in 1994(EMiB; Smith and Prior 1997). The samplewas fully representative of people ofCaribbean, South Asian and Chinese originliving in England and Wales. (People ofAfrican, Other Asian and Other origins werenot covered by the survey.) 5,196 minorityadults were interviewed, in 3,315 households.2,867 white adults (in 2,867 households)provided a comparison group. The question-naires for minorities was translated into

all members of a household from all sources. Earningsis therefore one component of income.

six Asian languages, and interviewers werematched to respondents by ethnic origin andlanguage.

The Fourth Survey is unquestionably the mostthorough specialist survey of ethnic minoritiesavailable. Like its predecessors, it includedquestions about employees’ earnings and theprofits of self-employed workers. Unlike theprevious three surveys, there was an additionalquestion about overall household income. Allthree questions were asked in the same format:a card listing income ranges was shown to therespondent, who was asked to indicate whichof 16 bands included their own earnings orincome. This form of question is commonlyused in surveys where earnings and income arenot the principal subjects of enquiry. Directcomparison between the crude question andmore detailed questioning has suggested thatthe banded income question providesreasonably accurate and robust results (Fosterand Lound 1993).

The basic results of the Fourth Surveyquestions have been analysed in outline in themain report on the survey (EMiB, seeespecially Chapters 4 and 5).

The Family Resources Survey

For many years, the primary household-basedsource of information about earnings andincomes was the Family Expenditure Survey.It did not identify minorities, but in any case,the sample size — 5,000 households each year— was too small. The General HouseholdSurvey included some basic earnings andincome data, and its larger sample allowedsome analysis by ethnic group (eg Denny andothers 1997, FitzGerald and Uglow 1993). Ithas, though, never provided a serious base forthe analysis of ethnic minority incomes; itssample and its income questions just aboutmet the grade in each dimension, without

8

ISER Report 98-1

offering the quality in either dimension thatwould be expected of a specialist survey.

The situation has been transformed by theintroduction of the Family Resources Survey(FRS), commissioned by the Department ofSocial Security. The two years of dataanalysed here (1994/5 and 1995/6) cover50,000 households, including a large enoughsample of ethnic minorities for reliableanalysis.

The FRS is the most comprehensive survey ofincomes ever undertaken in this country. Allmembers of the household are asked detailedquestions about each of their sources ofincome. It is as far in advance of thecompetition in its field as the Fourth Survey isin its own. Of course some individuals do notanswer every question about their income.Someone who refused to answer any questionswould not appear in the survey at all, but aproportion of households are missing theanswer to one detailed question or another.The data being used here include answers thathave been filled in by the Department ofSocial Security’s analysts, using asophisticated procedure for working out(‘imputing’) what the amounts would havebeen, given the other known characteristics ofthe person or household concerned. Thiscannot be expected to produce exactly accurateresults for each member of the sample, butshould be reliable in the analysis of largegroups of households.

The DSS provided two sets of data derivedfrom the FRS6. The first was the basic data(but including imputed values for missing

6 To be precise, both sets of data were provided for1994/5, but for 1995/6 we received the basic data plusthe computer program by which the HBAI variables hadbeen derived the previous year. To the extent that theDSS refined their programs for 1995/6 and incorporatedminor changes made to the data, our 1995/6 derivedvariables may not be exactly identical to the DSS’s.

variables, as just described). This is the dataused for the annual base report on the FRS(DSS 1997a). The other data set was theversion derived by DSS analysts for theDepartment’s annual analysis of HouseholdsBelow Average Income (DSS 1997b). Thisincludes a number of derived variables (suchas total household net income) which havebeen checked by the DSS for consistency.

This report is based primarily on the FamilyResources Survey, but makes some com-parison with the Fourth National Survey ofEthnic Minorities. A more detailed assessmentof the differences between the two sources,including a direct comparison of some keyquestions, is provided in Appendix A.

9


1. THE COMPOSITION OF FAMILY INCOMES

Background

This first chapter focuses on the compositionof the income available to families in differentcircumstances: in work, out of work or retired.For workers, the most important source ofincome will be their earnings but, as will beseen, benefits make an important contributionat the margin. For non-workers belowretirement age, most income comes frombenefits, but the main client groups —unemployed, lone parents, disabled — facerather different regimes. For pensioners, acentral issue is the relative importance of stateand occupational or private provision.

The recent restrictions on the availability ofbenefits to migrants and asylum seekers maybe seen to discriminate against ethnicminorities, but among those resident in thiscountry, the rules do not distinguish directlybetween ethnic groups. There are neverthelessa number of ways in which the ‘impartial’operation of the system may place minoritiesin a different, and potentially worse, positionfrom that of white people. For example7:

• A high proportion of some minority groupsare unemployed. They will therefore beespecially sensitive to the less generousarrangements for unemployed people,compared with other client groups.

• Members of some minorities receive lowwages when in work. Many Asians haverelatively large families. Low pay and large

7 The following list of points is a précis of part of a noteabout Social Security and Race written for the SocialSecurity Advisory Committee at the time of the Fowlerreforms of the benefit system (Berthoud 1988). It hastaken nearly ten years for data to become availablewhich would test the validity of this analysis; but thefindings in this chapter confirm the relevance of each ofthe factors listed.

families are the conditions for entitlementto Family Credit.

• A large number of women of Caribbeanorigin bring up children on their own. Theyshare with other lone parents the difficultyof maintaining themselves and theirfamilies in the labour market; a very highproportion of them have to claim IncomeSupport.

• Migrants who arrived in this country asadults are unlikely to have built up fulloccupational pension entitlements. Nor willthey have the full qualifications for theState Earnings Related Pensions Scheme.Even among those who were born in thiscountry, some groups’ life-timeemployment and earnings records are likelyto be worse than those of white workers.These facts indicate poor pension provisionfor minority groups, and a high probabilityof having to claim Income Support in oldage.

These are the conditions which might createethnic disadvantage, even though the socialsecurity rules were apparently fair and fairlyapplied. It has also been suggested thatminorities are less well treated by the staff ofbenefit offices (Gordon and Newnham 1985,NACAB 1991), or that they need more adviceto find their way about the system (Bloch1993, Law and others 1994) but the dataavailable for the current study do not allow usto examine those problems on this occasion.

Introducing the analysis

The Fourth National Survey of EthnicMinorities provided some information abouttotal household income (EMiB), but no detailabout its make-up — earnings, benefits and soon. The Family Resources Survey provides a

10

ISER Report 98-1

much more detailed account of thecomponents of family incomes.

Although it is individual adults who takeemployment, and receive earnings, it is morehelpful to analyse total incomes at the level ofthe ‘family’. This is defined as either a singleadult, or a couple (married or living asmarried), including their dependent children.8

The assumption is built into the analysis thatthe income of any one member of a family isavailable to be spent on all the others. Eventhough many families do not share theirresources fairly (Pahl 1989, G. Wilson 1987),there is probably still a legal obligation on amarried couple to maintain each other, andparents are certainly obliged to maintain theirchildren. Whatever happens on the ground,many social security benefits are assessed andpaid on the basis of information about thecircumstances of a whole family9, and it wouldnot make sense to analyse claims as thoughindividual adults were not members of familygroups.

The ‘family’ defined for this purpose is not thesame as the ‘household’, which consists ofpeople who live and eat together, but who maynot be related, and may have no mutualobligations. The incomes of whole householdswill be considered in the next chapter, but forthe moment it is the family that forms the unitof analysis.

A family’s income can come from three broadtypes of source:

• Income from work: wages, salaries andearnings from self-employment.

8 A ‘dependent’ child is defined, following theconvention embodied in social security rules, as agedless than 16, or aged between 16 and 18 and still in full-time education. Sons and daughters who are older thanthat are each treated as a separate ‘family’, even if theyare living in the same household as their parents.9 The technical term is ‘aggregation’. For a discussionof alternative approaches, see Esam and Berthoud 1991.

11


• Other sources of income derived in one wayor another from the market — occupationaland personal pensions, investment incomeand so on.

• Social security benefits paid by the state.

All the income figures quoted in this sectionare net of tax and other compulsorydeductions. But they are absolute weeklyfigures, and have not been adjusted by anequivalence scale.

For each family, we also calculated how muchbasic Income Support they would have beenentitled to if they had had to claim in the yearin question, including Housing Benefit andcouncil tax rebate. A deliberately crudecalculation of the amount has been used toassess these basic needs, independent of thetortuous set of rules devised by government togive a bit more money to this group, a bit lessto that. The details are explained in AppendixB. Available income is the family’s actualincome after subtracting their basic needscalculated in this way. As will be seen, somefamilies’ available income was negative: thismeans that their net income was below theirbasic needs.

Stage in the life-cycle

People have very different expectations abouttheir principal sources of income, dependingon their age and family structure. Acomparison between ethnic groups has to takefamily structure into account, because theminorities have much younger age-profilesthan the white majority (see Table 1, page 2).This is largely because most of the people whomigrated to Britain came as young adults —initially from the West Indies in the fifties,most recently from Bangladesh and Uganda inthe eighties — and it will take several moredecades before enough of them have reached

old age to form a full cross-section of the life-cycle.

12

ISER Report 98-1

Table 1: 1: Age and structure of familiesColumn percentages

Whites Caribbean African Indian Pakistani/Bangladeshi

Chinese Others

Pensioners 27 11 3 11 6 7 6Over 60 5 4 2 5 6 1 3Lone parents 6 23 25 4 7 7 10Teenagers 4 3 3 4 6 5 4Others 58 58 67 75 76 80 77Sample size 59319 768 263 722 501 142 709

Source: FRS 1994/5 and 1995/6

The most important factor from our point ofview is that a quarter of our FRS sample ofwhite families were pensioners (Table 1.110).But among minority groups, the proportionvaried between only 3 per cent (Africans) and11 per cent (Caribbeans and Indians). Becausemost pensioners did not work, and because (aswe will see) many of them had low incomes,the white population might be expected tohave lower incomes than the minorities.

There are other population groups, besidespensioners, among whom relatively fewpeople had jobs — whatever their ethnicgroup. Only about half of families where theman was aged between 60 and 65 were inemployment; and a similar proportion ofsingle teenagers (Table 1.2). There was littlevariation between ethnic groups, though, in thenumber of such families. Less than half of loneparents had a job, but this time they were

10The precise definitions of the groups in Table 1.1 wereas follows:Pensioners: single people over pensionable age; coupleswith the man over 65;Over 60: men over 60; couples with the man over 60Lone parents: Men or women with dependent children,not married or living as married;Teenagers: Single people aged less than 20. Thedefinition of a dependent child in footnote 1 means thatthis group did not include 16-18 year olds still in full-time education;Others: All other family units.

much more common in some ethnic groupsthan others — a quarter of the two groups of‘black’ family were lone parents. This is a wellknown feature of these minority groups (seeTable 1, page 2), and will help to explain lowlevels of employment, and of incomes, inthose communities.

The most direct comparison between ethnicgroups, then, is among families which do notfall into any of these low-employmentcategories — the bottom line, labelled ‘others’in Tables 1.1 and 1.2. The majority of allfamilies fell into this general category —three-fifths of whites and Caribbeans, two-thirds of Africans and three-quarters or moreof the Asian groups. It will be seen from thebottom row of Table 1.2, however, that not allethnic groups had the same chance ofemployment: the proportion of the ‘other’category of families with no job at all was onlyone fifth for whites, but rose to a quarter forIndian families, one third for Caribbeans andChinese, and nearly half for Africans,Pakistanis and Bangladeshis. This is a hugerange of variation when you consider that theeffects of age and family structure have beentaken out of the analysis: comparing like withlike, Africans, Pakistanis and Bangladeshiswere two and a half times more likely to haveno earner in the family than white people.Whatever the reasons for this lack ofemployment, some minority groups are likely

13


to be poorer, andTable 1: 2: Proportion of families with any worker, by family structure

Cell percentagesWhites Caribbean African Indian Pakistani/

BangladeshiChinese Others

Pensioners 8 10 5Over 60 51Lone parents 37 43 32 28Teenagers 54Others 81 68 51 73 53 68 66Sample size 59319 768 263 722 501 142 709

Source: FRS 1994/5 and 1995/6. Cells with fewer than 50 respondents have been left blank

more reliant on social security, than others,even among the core group who wouldnormally be expected to have a job.

• The following three sections divide thesample from each ethnic group into threeprincipal categories11. Working families:those families with any job (but excludingthe few workers above pensionable age).

• Non-working families: those families withno job, but below the age of 60.

• Pensioners: those over pensionable age,plus those over the age of 60 who were notin employment.

The incomes of working families

There were enough working families in theFRS sample for a full analysis by ethnic group.

Most of the analysis in this chapter will beconcerned with the total income of allmembers of a ‘family’ — adding together theearnings and other sources of income of

11 Single teenagers (whether in or out of work) and full-time students below the age of 25 have been removedfrom the analysis altogether. Their sources of incomeare very different from the rest of the population, butthere were not enough of them in the minority groupsamples to analyse separately.

husband and wife, depending on whether eachof them has a job, and how many hours s/heworks. Before turning to such family-basedaccounts, it is important to look specifically atearnings rates. The Family Resources Surveybroadly confirms the findings of the FourthNational Survey (EMiB, see Table 2 of thisreport) and of the Labour Force Survey(Employment Policy Institute 1998)

Chart 1.3 shows that, among men, Chinese andIndian full-time employees were on a par withthose of whites — the Chinese were actuallyslightly ahead of the majority group.Caribbean and African men averaged ratherless than whites. Pakistanis and Bangladeshiswere further behind again, averaging only£246 in comparison with whites’ £380 perweek.

Women averaged less than men for full-timework in all groups. In general the ethnicvariations among women were much less thanamong men. Chinese and Caribbean womenwere slightly ahead of the white average. Thelatter in particular is now a well-establishedfinding based on several sources of data. Thesmall number of Pakistani and Bangladeshiwomen full-time employees were again theworst-off group, though their disadvantage incomparison with their white comparison groupwas less than that experienced by Pakistaniand Bangladeshi men.

14

ISER Report 98-1

Obviously, all members of the categorydefined as working families had earnings toreport.

15


Chart 1.3: Average weekly earnings of full-time men and women employees

£0 £50 £100 £150 £200 £250 £300 £350 £400 £450

O ther

Chinese

Pakistani Bangladeshi

Indian

African

Caribbean

W hite

M enW om en


When we use the ‘family’ as the unit ofanalysis, there are three distinct sources ofvariation in the amount of earnings: whether(in the case of couples) one or both partners isin employment; how many hours each workeris employed; and the wage or salary s/he ispaid.

Table 1.4 shows the averages and distributionsof total earnings for working families in eachethnic group. The Chinese and Indians had thehighest totals, followed by white families.

Caribbeans and Africans were significantlybelow the white average. Pakistanis andBangladeshi working families were muchworse off than any other group, though: onaverage £124 per week below whites, and £65below their nearest neighbours, the Africans. Ifthe quarter of whites earning less than £150are considered ‘low paid’, and the quarterearning £380 or more are ‘high paid’, half ofPakistani and Bangladeshi working familieswere in the low group, less than one in ten inthe high group. This is clearly the result of a

Table 1.4: Net earnings of working familiesMeans, column percentages

White Caribbean African Indian Pakistani/Bangladeshi

Chinese Other

Averageearnings

£298 £253 £239 £305 £174 £314 £303

Up to £149 24% 28% 32% 22% 51% 24% 23%£150 to £379 51% 53% 52% 53% 41% 49% 51%£380 or more 25% 19% 16% 25% 8% 27% 26%Sample size 31069 406 115 430 216 84 401


16

ISER Report 98-1

Table 1.5: Other sources of income received by working familiesMeans, column percentages


combination of the low rates of pay earned byPakistani and Bangladeshi men, and low ratesof economic activity among women.

We would expect earnings to be the primarysource of income among these workingfamilies, and this is confirmed, below, for allethnic groups. But other sources of incomewere also reported by the great majority ofworking families (Table 1.5). The mostcommon consisted of unearned incomederived at first or second hand from themarket. This covered a variety of sources, ofwhich the most frequent was interest anddividends from savings and investments.Many people received unearned income of thissort, though the amounts may not have beenvery large, in comparison with earnings: three-quarters of whites, Indians and Chinese; abouttwo-thirds of the two black groups; but onlyhalf of Pakistani and Bangladeshi workingfamilies. The signs are, therefore, that theavailability of unearned income followsroughly the same distribution as of earnedincome.

Although the social security system is aimedmainly at families with no earnings, benefitswere also of some importance to theseworking families, especially to some groups.The most common, of course, was ChildBenefit, which (with its linked One Parent

Benefit) is paid to all parents without regard totheir employment status or income (Table 1.5).The distribution of Child Benefit betweenethnic groups simply mirrors the presence ofchildren: two fifths of white families, rising totwo-thirds of Pakistani and Bangladeshifamilies, received it. The line in the tablelabelled ‘Other benefits’ covers a wide varietyof particular schemes, and although there werevariations between ethnic groups in thenumber of people receiving them, it is difficultto identify a clear pattern.

A more striking pattern can be seen in the useof means-tested benefits. A small number ofpeople were recorded as ‘working families’even though they were working part-time andclaiming Income Support. Other benefits,though, are available to full-time workers,provided their earnings are low enough:Housing Benefit and Council Tax Rebates arebased on a calculation linking rent/taxliabilities, household size and income; FamilyCredit is designed specifically for low paidworkers. For whites, Indians and Chinese, lessthan one family in ten received any means-tested benefit while working. For Caribbeansand Africans, approaching two in ten; forPakistanis and Bangladeshis, three in ten.

Table 1.6 shows more clearly how the means-tests affected different types of family.


Chinese Other

Other marketincome

78% 65% 69% 77% 51% 79% 72%

Child and oneparent benefits

38% 44% 47% 62% 68% 49% 50%

Means-testedbenefits

8% 16% 19% 9% 29% 7% 10%

Other benefits 14% 20% 21% 13% 23% 6% 14%Sample size 31069 406 115 430 216 84 401

17


Table 1.6: Working families’ receipt of means-tested benefits, by family typeColumn percentages

Familieswith nochildren

Loneparents

Couples with children

Allgroups*

Allgroups*

Pakistanis/Bangladeshis

Africans Othergroups*

Family Credit nil 34 29 6 5Housing Benefit 2 30 15 15 3Council Tax Rebate 3 22 17 18 3Any means-tested benefit 4 56 40 24 8Average amount £1 £39 £33 £7 £4Sample size 19756 1539 144 33 11249

Source: FRS 1994/5 and 1995/6. Income Support is included under ‘any means-tested benefit’, but is not shownseparately

* Note that ‘All groups’ and ‘Other groups’ includes whites

• Among single people and couples withoutchildren, hardly any (from any ethnicgroup) claimed such benefits — FamilyCredit is confined to families with children,Housing Benefit and Council Tax Rebatescale rates are geared towards them.

• A high proportion of lone parents claimedmeans-tested benefits, especially FamilyCredit and Housing Benefit. Means-testedbenefits, and Family Credit in particular,are an important element of policy to enablelone parents to work in spite of the lowrates of pay and short hours available tomany of them (McKay and Marsh 1994).These benefits were of equal importance tolone parents from all ethnic groups. Butbecause lone parenthood is very commonamong Caribbean and African families, andvery rare among Indian, Pakistani andBangladeshi families, the overall impact ofthe benefits varied between communities.

• Very few couples with children from mostethnic groups received any means-testedbenefits. Pakistanis and Bangladeshis wereexceptional in this respect: 29 per cent oftwo-parent families in this group claimedFamily Credit — as many as among white

or black lone parents. Housing Benefit andCouncil Tax Rebates were also of someimportance for Pakistanis and Bangladeshiswith children, and also to African coupleswith children.

The impact of means-tests, and especially ofFamily Credit, on Pakistani and Bangladeshiworking families with children is striking, andan important outcome of recent policydevelopments. In-work means-tested benefitshave been a major element of governmentpolicy designed to ‘target’ help on those ingreatest need — Family Credit was claimed asthe ‘jewel in the crown’ of the social securityreforms introduced in 1988. It is directlyaimed at families with low earnings and largenumbers of children; previous research hadindicated that it was achieving significantpenetration among lone parents, but itsachievements among two-parent families wereless clear (Marsh and McKay 1993).Pakistanis and Bangladeshis are now found tobe right in the middle of the target group, anda very high proportion of them were receivingthe benefit. They are therefore a litmus groupfor the success or failure of the policy. It willbe judged a success, to the extent that theadditional £75 or so for each claimant family

18

ISER Report 98-1

Table 1.7: Components of net income among working familiesMeans


Chinese

Other

Net earnings £298 £253 £239 £305 £174 £314 £303Other marketincome

£23 £15 £10 £19 £9 £29 £25

Child and oneparent benefits

£7 £8 £9 £11 £17 £8 £9

Means-testedbenefits

£4 £8 £10 £5 £22 £3 £7

Other benefits £7 £7 £4 £7 £4 £0 £6(Adjustments) -£18 -£14 -£11 -£17 -£9 -£19 -£3Total net income £321 £277 £261 £330 £217 £335 £337Benefits as % oftotal

6% 8% 9% 7% 20% 3% 7%

Sample size 31069 406 115 430 216 84 401Source: FRS 1994/5 and 1995/6. ‘Adjustments’ consist of taxes and other deductions not already accounted for in the

calculation of net earnings. They are not taken into account in Charts 1.8 and 1.9

improves their living standards compared withwhat it might have been if no means-testedbenefit had been on offer. On the other hand, ithas been argued (Marsh and McKay 1993) thathaving to depend on means-tested supportplaces a family on the disadvantaged side ofBritish society, even if it is combined withearnings. Nor can Family Credit claim to havesucceeded in some of its other objectives, asfar as Pakistanis and Bangladeshis areconcerned. They still have very high rates of

unemployment, so the incentive effect has notovercome their other disadvantages. And itwill be seen that they are still very poor, so thewelfare effect is not strong enough either.

Whatever benefits are on offer, earningscontinue to be the primary source of incomefor the working families being considered inthis section. As Table 1.7 and Chart 1.8confirm, variations between ethnic groups inaverage earnings are closely reflected by

Chart 1.8 Components of total net income of working families

£0 £50 £100 £150 £200 £250 £300 £350 £400

Other

Chinese


Indian

African

Caribbean

White

Net earnings Other market income Child and one parent benefitsMeans-tested benefits Other benefits

19


Chart 1.9: Composition of total net income of working families

variations in their total incomes, after addingin unearned income and benefits. Benefitsnever- theless had an important contribution tomake for some groups of working families —the lower their earnings, the more importantbenefits were. To the extent that one-fifth ofthe income of Pakistani and Bangladeshiworking families was accounted for by socialsecurity (Chart 1.9). More than half of theirbenefits were means-tested.

How do these incomes compare with the‘needs’ of the families concerned. The currentconvention (DSS 1997b) is that total incomeshould be adjusted for household composition,and then compared in some way with thenational average. That convention will beadopted for the overall analysis of householdincomes, in the next chapter. At this stage theanalysis of family units compares currentincomes with a simple measure of basic needsderived from the system of means-testedbenefits which is intended to provide a floorbelow which no-one should fall. This offers aconvenient baseline against which to comparepeople’s actual incomes. ‘Available’ income isdefined as the amount of income received bythe family in excess of the baseline measure.

The details of the calculation are explained inAppendix B.

The ranges of income ‘available’ to workingfamilies are shown in Chart 1.10. The firstpoint to note is that there was a widedistribution of income within each ethnicgroup, and that they overlapped considerably.You could not say that all working Pakistanisand Bangladeshis were poorer than anyworking members of the more prosperousgroups such as whites, Indians or Chinese.Indeed, some groups were characterised by thewidth of their range of incomes: Indians,especially, included many of the best offfamilies, but other members of the same groupwere very close to the baseline derived fromsocial security rates. Nevertheless, workingCaribbean families were distinctly below thetop three groups; Africans below them; andthe Pakistanis and Bangladeshis at the bottomof the scale. More than a quarter of workingPakistanis and Bangladeshis had net incomesbelow their family needs baseline. Even thecombination of earnings and in-work benefitshad not protected this group from poverty.

0 % 2 0 % 4 0 % 6 0 % 8 0 % 1 0 0 %

O th e r

C h in e s e

P a k is ta n i B a n g la d e s h i

In d ia n

A fric a n

C a rib b e a n

W h ite

N e t e a rn in g s O th e r m a rk e t in c o m e C h ild a n d o n e p a re n t b e n e fitsM e a n s - te s te d b e n e fits O th e r b e n e fits

20

ISER Report 98-1

Chart 1.10: Quantiles (10, 25, 50, 75, 90) of the distribution of available income of workingfamilies

£28£15

-£19£8

-£73

-£2£12

£80£52

£36£63

£1

£58 £58

£153

£118£106

£138

£52

£134 £134

£259

£203£182

£248

£129

£266 £266

£398

£318£301

£419

£230

£381 £381

-£100

-£50

£0

£50

£100

£150

£200

£250

£300

£350

£400

£450

W hite Caribbean Indian Chinese OtherAfrican

Pak Bang

Note: See Appendix B for definition of ‘available’ income

Non-working families below 60

The analysis now turns to the second of thefamily types defined at the beginning of thischapter. This consists of families (benefitunits) none of whose members was inemployment, and whose head was below theage of 60. Single teenagers have beenexcluded, though, as have full-time studentsbelow the age of 25. There were fewer than 50non-working families in the FRS sample ofChinese people, and they are therefore notexamined in this section.

There are three main reasons why a familymight be out of work before they reach the ageat which everyone retires from the labourforce: disability, lone parenthood andunemployment. The three are treated quitedifferently by the social security system.

• The national insurance scheme covers mostpeople who leave work on grounds of ill-health and disability. It provides a long-term benefit roughly equivalent to the basic

retirement pension, although the benefitwas curtailed when it was converted fromInvalidity Benefit to Incapacity Benefitin 1995 (Berthoud 1998). Many disabledpeople can also claim the DisabilityLiving Allowance, another non-means-tested benefit intended to meet the extracosts of disability.

• There is no national insurance provision forlone parents (unless they are widows) andmost of this group have to claim IncomeSupport and other means-tested benefits.

• The national insurance scheme providedUnemployment Benefit until 1995, and thenJob-Seeker’s Allowance. (All of the dataused here refer to the period before thechange.) The rates of benefit are lower thanthose available to pensioners and disabledpeople; the insurance scheme is limited to12 months (UB) or 6 months (JSA); andcarries no allowance for children. Mostunemployed people therefore have to claimIncome Support.

21


Table 1.11: Sources of income of non-working families, by client typeColumn percentages

Disabled Lone Parents UnemployedUnearned income 47 81 46Insurance benefits 52 3 13Any means-test 75 96 80 Income support 56 93 75Other benefits 78 99 52Sample size 3016 2234 3263Source: FRS 1994/5 and 1995/6. Students and ‘pre-retired’ are excludedNote: This analysis covers all ethnic groups combined

These three categories of non-working familyare often referred to as ‘client groups’ and arethe basis for many of the social securitystatistics published by the DSS. For presentpurposes they will be referred to as ‘clienttypes’, to avoid confusion with ethnic groups.Although political debate focuses first on oneand then on another, a long run view showsthat there have been substantial increases inthe numbers of all three types of claim overthe past 25 years, even though the totalnumber of people of working age in Britainhas remained fairly stable (Berthoud andothers 1998).

The consequences of the different benefitregimes are illustrated in Table 1.11 and Chart1.12, which refer to non-working familiesfrom all ethnic groups combined. The resultsare very much in line with the policyinfluences just outlined.

• A high proportion of lone parents receivedsome unearned income — much of it inmaintenance payments. Only about half ofdisabled people received unearned income— mainly occupational pensions. But theamounts of disabled people’s unearnedincome were relatively large, so that in totalthis group obtained rather more than loneparents from non-state sources — and muchmore than the unemployed.

• Disabled people had a clear advantage overthe other two client types in their claims onthe insurance system. It accounted for morethan a quarter of their total income, butmade a negligible contribution to loneparents’ and unemployed people’sresources.

• More than half of the non-working familiesin each client type had to claim IncomeSupport — just over half of disabledpeople, three-quarters of the unemployed,and nearly all one parent families.

• The main other benefits were Child Benefitfor one parent families, and DisabilityLiving Allowance for disabled people.

The total incomes of unemployed people inChart 1.12 were significantly lower than thoseof the other two groups. That might have beena consequence of their family structures — ifthey had smaller families, they would beentitled to less benefit, but that should havereflected their needs. However Chart 1.13shows the three client types’ ‘available’incomes in relation to the estimate of needsbased on means-test scale rates.12 As expected,all three types of non-working family weremuch worse off than the working familiesreported in the previous section (Chart 1.10).

12 As defined in Appendix B.

22

ISER Report 98-1

Chart 1.12: Composition of income of non-working families, by client type

• The great majority of disabled people were,nevertheless, receiving a total income ratherhigher than their baseline needs, and aquarter of them had at least £70 per weekclear of the baseline. Note that for thisanalysis the calculation of baseline needsdoes not include an allowance for the extracosts experienced by disabled people, andwhich are intended to be met by DLA. Ifextra costs were allowed for, disabledpeople would appear worse off than Chart1.13 indicates. But the better position ofthis client type is also explained by theirrelatively generous provision of insurance-based benefits.13

• Non-working lone parents were quitetightly bunched near the level of incomecalculated to be their baseline needs.

• But unemployed families were much worseoff than the other two groups: more thanhalf reported an income below theirbaseline needs, and a few were well belowthis implicit poverty line.

13 For a more detailed analysis of FRS data on theincomes of disabled claimants, see Berthoud 1998.

Two main conclusions from this brief reviewof the incomes of non-working families fromall ethnic groups need to be taken forward tothe comparison between ethnic groups. First,the importance of means-tested benefits. Thesebenefits were originally intended as a minorback-up to the main social security system(Deacon and Bradshaw 1983). They arecheaper than insurance or contingent benefits,but they suffer a number of disadvantages.Many commentators regret the over-relianceon means-tests; many others argue strongly for‘targeted’ benefits as a way of minimising thestate’s role. What the foregoing analysis showsis that means-tests are now the main deter-minant of non-working families’ incomesbelow the age of 60. £4 in every £10 of benefitexpenditure on disabled people consisted ofmeans tests; £8 in every £10 for lone parentsand the unemployed.14 For the majority in allgroups who receive means-tested as well asother benefits, it is the means-test whichultimately determines total income.

14 These proportions derived from the FRS are verysimilar to the calculations derived from the publicaccounts (DSS 1997c).

£0 £20 £40 £60 £80 £100 £120 £140 £160 £180

Unem ployed

Lone parents

D isabled

Unearned incom e Insurance benefits M eans-tests O ther benefits

23


Chart 1.13: Quantiles (10, 25, 50, 75, 90) of the distribution of ‘available’ income ofnon-working families, by client type

-£ 9-£ 1 9

-£ 5 6

£ 1 2

-£ 2

-£ 2 1

£ 3 1

£ 5-£ 6

£ 7 0

£ 1 2£ 1

£ 1 1 8

£ 2 6£ 1 8

-£ 1 0 0

-£ 5 0

£ 0

£ 5 0

£ 1 0 0

£ 1 5 0

D is a b le dL o n e p a re n ts

U n e m p lo y e d

Note: This analysis covers all ethnic groups combined

The second important point is the differencesin ‘available’ income between theunemployed, lone parents and disabledclaimants. This is not an accident; benefitlevels for the unemployed have deliberatelybeen allowed to fall behind provision for otherclient types because, historically, one parentfamilies and disabled people were notexpected to look for work, and did not requirethe harsh incentive regime imposed on thosewho had no specific reason, other than scarcityof jobs, not to be in the labour force.

All three types of non-worker were found ineach ethnic group’s sample, plus additionalcategories of students and early retired peoplewho were of some importance to some groups(Table 1.14). Compared with whites:

• Caribbeans and Africans had more oneparent families, while Asians had fewer;

• Indians had more disabled people;

Table 1.14: Client type of non-working families below 60Column percentages

White Caribbean African Indian Pakistani/Bangladesh

i

Others

Disabled 34 19 18 41 28 17Lone parent 24 38 33 13 16 24Unemployed 35 40 41 41 53 43‘Retired’ 6 1 nil 5 3 2Student 1 2 8 2 nil 14Sample size 8252 228 122 128 186 195Source: FRS 1994/5 and 1995/6. Non-working Chinese families have been excluded because there were only 30 in the

sample. ‘Retired’ is defined as non-workers over the age of 40, but below 60, who did not report that they were

24

ISER Report 98-1

‘unemployed’ or ‘disabled’, and whose unearned income was at least £50 per week. Widows have also beenincluded in that category.

25


Table 1.15 Sources of income of non-working familiesColumn percentages


Others

Unearnedincome

58 54 59 56 62 62

Insurancebenefits

26 13 6 34 18 12

Any means-test 78 88 83 76 89 78 Income Support 68 81 77 71 84 67Other benefits 70 73 66 70 77 66Sample size 8252 228 122 128 186 195Source: FRS 1994/5 and 1995/6. Non-working Chinese families have been excluded because there were only 30 in the

sample

• Pakistanis and Bangladeshis had moreunemployed people than other groups.

These variations in the composition of thesample will help to explain why some minoritygroups were more or less likely to receivecertain types of benefit, and so to have higheror lower incomes. In practice, the effects werenot very great. The following series of tablesand charts show the actual incomes of eachethnic group; at the end of the section, this willbe compared with what the differences wouldhave been if each ethnic group had had thesame proportions of disabled people, loneparents and unemployed among its non-working families.

Table 1.15 and Charts 1.16 and 1.17 comparethe sources of income of non-working familiesby ethnic group. There was virtually no earnedincome, and this is omitted from the analysis.About half of all the non-workers hadunearned income, in every ethnic group. It wasbenefits, though, which provided theoverwhelming proportion of income. Nationalinsurance was of some importance to Indians(who had a high proportion of disabled peopleamong their non-working families) and whites(who are most likely to have built up aninsurance record). The ‘other’ benefits,including Disability Living Allowance and

Child Benefit, were of greatest importance toIndians, Pakistanis and Bangladeshis. But, asthe earlier analysis showed, most non-workingfamilies depended, in the end, on means-tests.There was some variation between groups inthe number claiming Income Support, butmeans-tested benefits were the biggest singleitem for every ethnic group — ranging fromhalf the total income of whites and Indians, tothree quarters of the income of non-workingAfricans.

Chart 1.16 showed that, in absolute terms,Pakistanis and Bangladeshis had the highestaverage incomes of non-working families. Butthis was largely because, with many children,they attracted higher Income Supportpayments. If incomes are compared with eachfamily’s basic needs, calculated from means-test scale rates, then Chart 1.18 shows that themedian levels of ‘available’ income rangedonly between -£6 and +£6 per week. All thefamilies being considered here were living onsocial security, and most of them, from allethnic groups, were claiming means-testedbenefits. The general conclusion of Chart 1.18is not the extent of differences, but the broadsimilarity of outcomes. The conclusion of thischapter will be that some ethnic minorities,especially Pakistanis and Bangladeshis, weremuch poorer than others, and poorer than the

26

ISER Report 98-1

white majority. But this was because more ofthem were on social security, not because

Pakistani

Chart 1.16: Components of total income of non-working families

Chart 1.17: Composition of total income of non-working families

£0 £20 £40 £60 £80 £100 £120 £140 £160

Other

Pak Bang

Indian

African

Caribbean

White

Unearned income Insurance benefits Means-tests Other benefits

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Other

Pak Bang

Indian

African

Caribbean

White

Unearned income Insurance benefits Means-tests Other benefits

27


Chart 1.18: Quantiles (10, 25, 50, 75, 90) of the distribution of available income ofnon-working families

-£34-£41

-£55 -£53-£60

-£87

-£7 -£10

-£30

-£15-£22

-£27

£6£0

-£6£2 £0 -£4

£31

£13£8

£38

£17 £16

£86

£44

£22

£83

£51 £51

-£100

-£50

£0

£50

£100

W hiteCaribbean

African Indian Pak Bang

Other

Chart 1.19: Proportion of non-working families with available incomes below zero: actualcompared with weighted figures

and Bangladeshi social security claimantswere much poorer than white ones.

There were, nevertheless, some differencesbetween ethnic groups in the proportion ofnon-workers whose available incomes fellbelow their estimated needs — 38 per cent of

whites, rising to 62 per cent of Africans (Chart1.19, left-bars). This result is partly explainedby the composition of each ethnic group’ssample of non-working, families. The right-bars of Chart 1.19 use weighting to estimatehow many non-workers would be below theneeds scale if each minority group had the

0

10

20

30

40

50

White Caribbean African Indian Pak Bang Other

Per

cen

tag

es

Raw Weighted

28

ISER Report 98-1

Table 1.20: Sources of income of pensionersColumn percentages

White Caribbean Indian Pakistani/Bangladeshi

Other

Earnings 7 9 4 nil 9Any unearned income 83 56 58 44 70

Occupational pension 58 42 39 14 45Investment income 71 33 48 19 59

Insurance benefits 95 92 61 60 73Any means-test 39 70 55 83 52

Income support 17 42 41 71 32Other benefits 45 65 48 81 50Sample size 17327 103 101 52 56Source: FRS 194/5 and 1995/6. Africans and Chinese have been omitted from the analysis

same ratio of disabled people, lone parents andunemployed as whites.15 In every ethnicminority group, the number of low-incomenon-working families was partly explained bythe composition of the group by client type.Once the adjustment has been made, mostgroups were only slightly worse off thanwhites on this measure, with the exception ofAfricans.

Pensioners

The third main group defined at the beginningof the chapter was pensioners. These havebeen defined as families whose ‘head’ wasover pensionable age, whether they hadactually retired or not; plus those over the ageof 60 who were not in work. There were only11 Chinese and 12 African pensioner familiesin the sample, and they are not thereforeanalysed. There were 52 Pakistani andBangladeshi pensioners; their results are worthquoting, but will not be very accurate.

The issues for this section of the populationare different again. The state pension fundedby National Insurance contributions isintended to provide a base for all pensioners.

15 The weighted figures exclude retired and student non-working families (see Table 1.13).

Occupational pensions, incomes from savingsand so on, can be added, to provide a higherstandard of living in old age derived fromresources which are more directly linked to theindividual’s earnings and savings habitsduring their working life. The growth ofoccupational pensions, in particular, meansthat a larger number of elderly people are nowretiring to a period of prosperity than everbefore (Bone and others 1992). But for thosewho have not had access to such schemes, thebasic pension provides a very restrictedstandard of living. Its level has always beenrather lower than the scale rates of means-tested benefits, and elderly people with only asmall income from non-state sources generallyhave to claim Income Support, HousingBenefit and/or Council Tax Rebates.

These points are well illustrated by the resultsfor white pensioners in Table 1.20 and Chart1.21. Almost all white pensioners receivedinsurance benefits, which can be seen toprovide far better coverage for pensioners thanfor those out of work for other reasons (seeTable 1.11). Five out of six white pensionerfamilies had some form of unearned income,including more than half with an occupationalpension. On the other hand, one in six claimedIncome Support, and as many as two in fivedepended on one or another of the means-

29


Chart 1.21: Components of income of pensioners

tested benefits. For white people, nevertheless,43 per cent of total income was provided bythe NI pension; nearly as much — 40 per cent— by unearned income; and only 8 per cent bymeans-tested benefits.

Each of the three main minority groups had itsown pattern of income in retirement.Caribbeans — the group which has lived inthis country for the longest period — were justas likely to receive a state pension as whitepeople were, though the amounts they receivedwere a little less. But the Caribbeancommunity, having experienced low levels ofemployment and of earnings over the pastdecades, had less than half as much unearnedincome from occupational pensions andinvest-ments. As a result, Caribbeans weremuch more reliant on means-tests in general,and Income Support in particular.

As for Indians, only 60 per cent of‘pensioners’ appeared to be receiving the statepension. This must be because many of themhad arrived in this country too late to build uptheir entitlement. Since many of them havebeen relatively successful in the labour market,though, their non-state incomes were a littlehigher than those of Caribbeans, if less than

those of whites. The net result was that Indianpensioners were just as dependent on IncomeSupport and other means-tests as Caribbeanswere — about a quarter of their income, inboth cases.

For Pakistanis and Bangladeshis, only a smallproportion have reached pension age, and thesample is correspondingly small. But the signsare that, like the Indians, only about halfclaimed the state pension. Unlike Indians,though, few of them had any significant non-state resources: the average non-state incomewas £8 per week, compared with £79 for whitepensioners. No less than 71 per cent ofPakistani and Bangladeshi pensioners had toclaim Income Support — compared with 17per cent of white pensioners. More than half oftheir total income came from means-tests.

Chart 1.21 showed that the pensioner incomesof each of the three main minority groups hada different composition, but the overall totalswere rather similar. Clearly means-testedbenefits were effectively compensating forvariations between minorities in theavailability of the state pension and of non-state incomes. Only among white people was

0 20 40 60 80 100 120 140 160 180 200

Other

Pak Bang

Indian

Caribbean

W hite

Earnings Occupational pension Other unearned income Insurance benefits Means-tests Other benefits

30

ISER Report 98-1

there enough income as of right to overcome the means-testChart 1.22: Quantiles (10, 25, 50, 75, 90) of the distribution of available income of pensioner

families

£7

-£9-£19 -£23

-£32

£19£11 £9 £8

£0

£46

£24 £23 £21 £22

£102

£66£56 £56

£78

£192

£138 £142

£87

£185

-£50

£0

£50

£100

£150

£200

W hite

CaribbeanIndian Pak Bang

Other

poverty trap, and give pensioners theadvantage of their occupational pensions andinvestments. The effects on the distribution of‘available’ income for individual pensionerscan be seen in Chart 1.22. In every ethnicgroup (including whites), the less well-off halfof pensioners’ available incomes wereclustered either side of the basic needs scale— the median above the scale; the lowestdecile below it. There was a rank orderingfrom the less poor whites to the more poorPakistanis and Bangladeshis, but thedifferences here were not great — becausevariations in income among the poorer half ofpensioners have been ironed out by means-tests. In the better-off half of the distribution,on the other hand, the availability of basicpension plus unearned income has created aclear gap between whites at one extreme andPakistanis and Bangla-deshis at the other. Thewhite pensioner at the top decile had morethan double the ‘available’ income of theequivalent Pakistani or Bangla-deshi, withCaribbeans and Indians in between.

Comparing family types

This chapter has contained a great deal ofdetailed analysis, comparing the sources andamounts of income of six ethnic groups withinthree family types. Table 1.23 offers asummary, by comparing the ‘available’incomes reported by each family type. Thecolumns of the table have been re-ordered sothat the group with the highest incomes is onthe left, the group with the lowest on the right.Although the median incomes shown in thetable do not capture all the detail of thevariation, the overall picture is clear enough.

Among working families, whites appearsignificantly better off than any other group.Chinese and Indians come next, about £20 perweek lower than white working families.Caribbeans, followed by Africans, were someway behind again — about £40 per week lessthan whites. Pakistani and Bangladeshiworking families averaged only about £50 per

31


week above the social security entitlements

32

ISER Report 98-1

Table 1.23: Median ‘available’ income by type of family£ per week

White Indian Chinese Caribbean African Pakistani/Bangladeshi

Working families £153 £138 £134 £118 £106 £52Pensioners £46 £23 na £24 na £21Non-working familiesunder 60 £6 £2 na £0 -£6 £0All families £82 £76 £88 £66 £51 £27

Note: The first three lines of the table are the 50th percentiles shown in Charts 1.10, 1.18 and 1.22

from which ‘available’ income has beencalculated.

Pensioner families typically had far lowerincomes than working families. Whitepensioners were better off than the minorities,but there was little difference among thesegroups. All non-working families were closeto the social security ‘poverty line’. There waslittle difference between ethnic groups.

But among non-working non-pensionerfamilies, most members of all ethnic groupsdepended on social security benefits. Therewas very little difference between groups.

Taking all three types of family together, thewhite position is reduced somewhat in relationto ethnic minorities by the relatively largenumber of white pensioners. But the Pakistaniand Bangladeshi position is worsened, relativeto the more prosperous groups, by theirnumber of non-working families. There is aclear ordering of groups: whites, Indians andChinese with the best position; Caribbeans andAfricans in the middle; and Pakistanis andBangladeshis clearly worst off.

33


2. LEVELS OF HOUSEHOLD INCOME

Background

This second chapter is concerned with the totalincomes available to ethnic minority house-holds, in relation to the number of people to besupported. It was pointed out in theintroduction that it has long been clear thatethnic minorities, or at least some of them,must have lower average incomes than thewhite population, but it has not been possiblebefore now actually to measure the gap. Somepreliminary conclusions were reported fromthe Fourth National Survey (EMiB, Chapter 5)— showing, especially, a huge gap betweenPakistanis and Bangladeshis and all otherethnic groups. This chapter consolidates thatanalysis with information from the FamilyResources Survey, which enables ethnicminority incomes to be set in the nationalcontext using exactly the same analyticalapproach as is adopted for the ‘official’ figurespublished by the DSS (DSS 1997b).

The chapter also looks at the geographicaldistribution of income. In parts of America,the majority of black people live in areaswhere the majority of the population is black(Peach 1996). These ‘ghettos’ are closelyassociated with poverty and with notions of‘underclass’ (W. J. Wilson 1987). We do nothave anything like that degree of concentrationin this country, but there are neverthelesssubstantial concentrations of minorities in, andwithin, the major conurbations (Ratcliffe1996). Many of the areas with a sizeableminority population are ‘run down’ (acharacteristic of the area itself) and are oftenthought of as ‘poor’ (a characteristic possiblyof the area, possibly of the people who livethere). Indeed, many area classifications usethe number of ethnic minority residents as oneof the key indicators of ‘deprivation’. Thesurvey being used for this study contains someinformation about the areas where people live,

and it will therefore be possible to make adirect comparison between geographical andindividual measures of income and ‘poverty’.

Introducing the analysis

The analysis of the sources of income ofworking and non-working families, and ofpensioners, in the previous chapter used the‘family’ as the unit of analysis, largely becausesocial security benefits are often assessed andpaid on the basis of the family unit. Now thatthe components have been assembled, we turnto the ‘household’ as the unit of analysis oftotal incomes. The assumption is that although‘families’ who live together in householdsmight separate from each other to form newhouseholds, they are likely to share manyaspects of their living standards as long as theyremain together.

Of course, the conversion from ‘families’ tohouseholds is not neutral between ethnicgroups. It has been well-established that adultsof Asian origin are more likely to live withtheir parents, and/or with their children, orwith brothers and sisters. Table 2.1 shows thathouseholds in both the Caribbean and Africanminority groups were similar to whites, in thatabout a one-sixth of ‘households’ containedmore than one ‘family’; these multiple house-holds accounted for a third of all families.Among Indians, Pakistanis and Bangladeshis,multi-family households were about twice asfrequent, and accounted for a majority of all‘families’. This meant that Asian householdsmight have had quite high incomes measuredin terms of pounds per week, but this did notnecessarily leave them well placed when thesize of the household was taken into account.

In this chapter, a number of analyticalconventions will be adopted which have beenset down by the Department of Social

34

ISER Report 98-1

Table 2.1: Multi-family households, by ethnic groupPercentages


Other

Proportion of‘families’ 33 39 34 66 59 53Proportion ofhouseholds 15 17 14 27 24 24

Security’s official analysis of HouseholdsBelow Average Income (HBAI) (DSS 1997b).The total income in the household has beenadded together using the DSS formulae:earnings, plus other market income, plusbenefits, minus direct taxes. No allowance ismade for the value of benefits or otherresources received in kind, rather than as acash income16. No deductions have been madefor indirect taxation, nor (except in one table)for housing costs. The equivalence scale bywhich income is adjusted for the needs oflarger and smaller households is also the same.The national average equivalent income hasbeen calculated, and each household’s incomehas been expressed as a percentage of theaverage. The data have been weighted by thenumber of individuals in each household, sothat the results reflect the number of people —adults and children — at each income level.

Comparing ethnic groups

Chart 2.2 shows the full distribution ofhousehold incomes for each ethnic group, inrelation to the national average. Since much ofthe rest of this section will be looking at lowincomes in some detail, it is appropriate tostart with a consideration of high incomes. Forwhite people the median income is about four-fifths of the average (mean). This is a commonfeature of income distributions, because themean is affected by the very high incomes ofthe ‘rich’, while the median is not. The top

16 Although Housing Benefit paid direct to the landlordis counted as income.

decile among whites — the income of the 90th

person up the scale — is about two-thirdshigher than the average. This represents a netincome of about £23,000 per year for a twoadult family.

There were some important differencesbetween ethnic groups at the top of the incomedistribution. All the minorities had mediansbelow that of whites, but the Chinese upperquartile and (especially) their top decile wereboth higher than whites. This suggests a widestretch among Chinese household incomes, buta number of them seem to have gained ahigher level of prosperity than their whiteequivalents. (‘Seem to have’ because thesample of Chinese households is too small forthe conclusion to be reliable.) But, as the chartshows clearly, the other minorities were allless well-represented than whites among highincome households. Indians and Caribbeanswere in a very similar position to each other,with rather fewer well-off households thanamong the majority group. Africans had fewerstill. Pakistanis and Bangladeshis, arguably,had none at all: the top decile for that groupwas well below the national average income.

At first sight, the variations between ethnicgroups were much narrower at the bottom endof the income scale. All groups’ lowest decilespenetrated below 50 per cent of the nationalaverage income, which is commonly used as aconventional benchmark to identify thepoorest households. A closer look, though,shows that while most ethnic groups’ lower

35


quartile — representing the poorest quarter of theirChart 2.2: Quantiles (10, 25, 50, 75, 90) of the distribution of household income in relation

to the national average

44% 41% 37% 38%30%

37% 37%

57% 53%47%

52%

38%47% 50%

83%75%

64%72%

46%

69%76%

121%

107%

89%

104%

59%

128%117%

168%

151%

136%

154%

86%

188%

165%

0%

50%

100%

150%

200%

W h iteC arib b ean A frican Ind ian

P ak B an gC h in ese O ther

Note: Net equivalent household income before housing costs

households — was close to the 50 per centbenchmark, for Pakistanis and Bangladeshisnot only the lower quartile but the median wasbelow the poverty line — that is, more thanhalf of that group were poor.

‘Poverty’

This section concentrates on a detailedanalysis of households whose income wasbelow 50 per cent of the national average.There has been a long debate in this countryabout what level of income is so low thathouseholds cannot participate in the activitieswhich their fellow citizens consider so normalas to be essential (Journal of Social Policy1987). The debate about a poverty line hasbeen inconclusive — at least, those who arguethat there is much poverty have failed toconvince those who argue that there is little.One element of the debate of particularrelevance in the present context is thereference group whose ‘normality’ each

household aspires to — is it the nationalaverage, or the average of the localcommunity, including perhaps many membersof the minority group? For migrants, does thereference group include ‘home’ where peoplemay be poorer or more prosperous than theyare in this country?

While recognising the importance of thesetheoretical issues, the analysis here adopts anentirely conventional approach, in which thosebelow half the national average are identified— and labelled ‘poor’ simply on the basis thatthey have less income than other households.The findings need to be interpreted in the lightof the general trend in Britain (DSS 1997b).Low-income households had roughly the sameresources at the end of the 1980s as theirequivalents had ten years earlier. All thegrowth in the national income over that periodwent to high-income households. The lowincome household fell behind in relativeterms: where only 8 per cent were below half

36

ISER Report 98-1

the national average in 1979, 20 per cent werein that position by 1992/93. This trend seemsto have levelled off in recent years; but the rateof poverty indicated by the analysis isprobably at a post-war high. Our own figuresfrom the Family Resources Survey show that17.3 per cent were below half the nationalaverage in 1994/5/6. It is this group which willbe analysed in detail.

The headline figures for each ethnic group areshown in the first row of Table 2.3. The 16 percent of whites below half the national averagewas, of course, very close to the overall rate.But it was also the lowest of any ethnic group.Caribbeans and Indians both had higher levelsof poverty. The Chinese had a higher rate too,though Chart 2.2 showed that they were wellrepresented at the top of the income scale aswell. 31 per cent of Africans were below thebenchmark poverty line. This is an importantfinding, because Africans could not beincluded in the sample for the Fourth NationalSurvey of Ethnic Minorities (EMiB), and wenow find that they were significantly worse offthan the Caribbeans with whom they are oftencombined. But — as every page of this reportwould lead us to expect — it was Pakistanisand Bangladeshis who were the poorest groupin Britain. Three out of five were below thethreshold. This was nearly four times the rateobserved among white households.

One way of assessing the impact of poverty onethnic minority groups is to show how the riskof falling below half the national averagevaried between types of household in eachgroup, and between groups in each type ofhousehold. In the white population, only 9 percent of households containing any worker fellbelow the benchmark poverty line (firstcolumn of Table 2.3). For white households allof whose members were above pensionableage, the risk was 23 per cent. For white non-pensioner households containing no worker,the risk was higher again, at 43 per cent. Thisis exactly what would be expected from whatwe know of the workings of the labour marketand the social security system, and runs inparallel with the findings of the earlier sectionbased on a similar analysis of families.

In the other ethnic groups, there were fewpensioners, and this, on its own, ought to havereduced the overall amount of poverty. Thedifferences in risk between working and non-working households was always in the samedirection as for whites. The levels of risk ineach category were sometimes similar towhites, and sometimes higher. For mostminority groups, it appeared that increasedlevels of poverty were partly attributable to

Table 2.3: People in households below 50% of the national averageCell percentages


Chinese Other

All households 16 20 31 22 60 28 26Working 9 8 9 15 50 18 14Pensioner 23 29Non-working 43 42 54 54 72 54Sample size(households) 49864 651 232 529 368 114 626Source: FRS 1994/5 and 1995/6. Cells with fewer than 50 households have been left blank. Income is net equivalent

income before housing costs

37


larger numbers of people of working agewithout employment, and partly due to lowerlevels of equivalent income once in that state.

The exceptions, again, were the Pakistanis andBangladeshis. Half of the working householdsand three-quarters of the non-workinghouseholds fell below the poverty line. Therisk of poverty among working Pakistani andBangladeshi households was higher than therisk among non-working white households —a result very similar to that derived from theFourth National Survey (EMiB).

Another way of looking at the same figures isto compare the composition of each ethnicgroup’s poor population in terms of householdtype (Chart 2.5). Although working house-holds had a relatively low risk of poverty,there are so many of them that they accountedfor as many as 40 per cent of all poor whitehouseholds. In the white population, there wasa significant contribution from poor pensionerhouseholds (though the popular impressionthat most poor people are elderly is clearlyinaccurate).

There were few pensioners in the minoritypopulations, and, therefore, few poorpensioners. In the two black groups, the extrapoor consisted mainly of people with noworker in their household. In the three Asiangroups, a relatively high proportion of the poorcame from working households.

Mixed households

All this analysis is based on classifying ahousehold under the minority heading if anyperson living there was not white. Otherresearch has showed that black and Chinesepeople have quite high rates of inter-marriagewith white people, while the South Asiancommunities have remained more exclusive(Berrington 1996, EMiB). The number ofmixed households in the FRS was consistentwith those findings. Table 2.6 compares ratesof poverty for households all of whosemembers were of the same minority group,with those who lived in the same household aswhite people. In every case, the mixedhouseholds were less likely to report lowincomes than the members of the same ethnic

Chart 2.5: Composition of poor households

0%

W hite Caribbean African Indian Pak/Bang Chinese Other

W orking Non-working Pensioner

38

ISER Report 98-1

Table 2.6: Risk of poverty in mixed minority/white householdsCell percentages

Caribbean African Indian Pakistani/Bangladeshi

Chinese

% of households whichwere mixed 20 18 11 4 30Risk of poverty amongunmixed households 21 33 23 61 33Risk of poverty amongmixed households 16 27 11 (5) 18Sample size (mixed) 129 41 56 14 34Source: FRS 1994/5 and 1995/6. Households which contained combinations of more than one minority ethnic group

have been assigned to the ‘other’ category, and are not included here. Note the very small cell size for mixedPakistani or Bangladeshi households

group who lived exclusively with members ofthe same community. In most minority groups,indeed, the risk of poverty was lower formixed households than for all-white house-holds (16 per cent). Much more detailedanalysis would be required to explain thisfinding (in terms, perhaps of the ages of theindividuals concerned, the duration of theirperiod of residence in Britain, and so on). Butit is at least a sign that social and economicintegration in a white society may be linked.

Alternative measures of poverty

The measure of income used so far is the oneknown as ‘before housing costs’ (BHC). Allincome, including Housing Benefit and othersupport for housing costs is included in thecalculation. The assumption is that theaccommodation occupied by the household isof value to it, and should be included in theestimate of its resources and consumption.

An alternative formulation is to measureincome ‘after housing costs’ (AHC). Theassumption here is that housing expenditure isan unavoidable necessity over which house-holds have no short-term and little long-termcontrol, and the AHC measure is a betterreflection of discretionary income. Much ofthe argument over the relative merits of these

two measures has been about their validity inmaking comparisons over time: during the1980s’ period of rapid increases in councilhouse rents the AHC measure provided abetter picture of the effects on the poor thanthe BHC measure could. That historicalcomparison is not relevant to the current study.

Another point is that an AHC measureincreases the apparent degree of inequality, byeliminating from consideration an element ofconsumption which is relatively equal. Thatwould be cheating. In order to make thecomparison as fair as possible, the low-incomethreshold for our AHC analysis has beencalculated to produce exactly the sameproportion of ‘poor’ people as the BHCanalysis. Instead of 50 per cent of average, wecount the number of people below 42.3 percent of average (Table 2.7, upper panel). Sothe measure is forced to produce the sameanswer for the whole population, and enablesus to look specifically at the effect of housingcosts on different ethnic groups.

For Asian households, the proportion belowthe threshold was very similar ‘after’subtracting housing costs, as it was ‘before’.For black households, both Caribbean andAfrican, housing costs had a much moreimportant effect. In both cases, the poverty

39


Table 2.7: Household incomes below a proportion of the national average: comparison ofthree measures

Column percentagesWhite Caribbean African Indian Pakistani/

BangladeshiChinese Other

Before housingcosts (50%) 16 20 31 22 60 28 26After housingcosts (42.3%) 16 28 46 23 56 32 32Sample size(households) 49864 651 232 529 368 114 626Fourth nationalsurvey 28 41 na 42 83 34 naSample size(FNS) 2457 778 na 757 737 115 na

estimate was higher, using the AHC measure.Nearly half of Africans were below thebenchmark, and this suggests that housingcosts were a significant drain on theirresources.

Another basis of comparison is with theFourth National Survey of Ethnic Minorities,conducted in 1994 (EMiB). As explained inthe introduction, the Fourth Survey used amuch cruder question about income than theexhaustive enquiry for which the FRS wasdesigned; but it had other potentialadvantages. One difference between the twosurveys is that the Fourth Survey used a muchsimpler equivalence scale than is adopted bythe HBAI conventions, but in fact the FRSproduced almost identical poverty estimatesusing the simple scale, as it did using theMcClements version.

The Fourth Survey produced much higheroverall estimates of the proportion below halfthe national average (Table 2.7, lower panel).This is directly consistent with the basiccomparison of the income distributions of thetwo surveys discussed in Appendix A: theFourth Survey seems to have under-stated the

incomes of poorer households and overstatedthe incomes of richer households.

The important point of comparison is betweenthe minority groups and the white figure. Herethe results of the two surveys were verysimilar. Similar in both, Caribbeans andIndians were rather worse off than whites. Inboth, Pakistanis and Bangladeshis were muchworse off than any other group. The FRSsuggested that the Chinese had a higherpoverty rate than Indians; the Fourth Surveygave them a lower poverty rate, though stillabove that of whites; but the sample ofChinese people in both surveys was too smallfor much weight to be placed on this finding.

Adjusting for household size

Analysts of the distribution of income andof poverty have considered a number ofalternative ways of adjusting householdincome to take account of the number and agesof the people among whom the income has tobe shared (Atkinson 1992). In principle, it canbe assumed, first, that many people requiremore income to maintain a given standard ofliving than one person on his or her own. Onthe other hand, it can also be assumed that

40

ISER Report 98-1

Chart 2.8: Households below half average income, using a range of equivalence scales

0

20

40

60

80

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Weight attached to extra household members (theta)

Per

cen

tag

es

Caribbean

Indian

Pak Bang

African

Chinese

White

some costs are fixed per household, which allmembers enjoy the benefit of. Thus, at thelimit, the extra needs associated with an extraperson are probably more than zero, but lessthan the needs of a single person.

The McClements scale used by the DSS hasthe advantage that it was based on empiricalanalysis (McClements 1977), and that it iswidely adopted in this country. Itsdisadvantages may be that the data analysedare now out of date; that it is unrealisticallycomplex; that the needs associated with veryyoung children are much lower than otherestimates suggest; and that it is not widelyused in international comparisons. Actually,most analysts conclude that the overall level ofinequality, or of poverty, is very similar,whichever equivalence scale is used. Butresults are more sensitive when comparisonsbetween types of household are introduced —small and large households will swappositions as greater or lesser weight is attachedto extra persons.

As already noted, a simple equivalence scale(1.00 for a single person, 0.6 per extra adult,0.3 per child) produced almost exactly thesame results from the FRS as the McClements

scale did, and that may suggest that theestimates are not sensitive to the scale used,within the range of ‘reasonable’ scales. Amore systematic test has been undertakenusing a generalised form of equivalence scalesuggested by Buhmann and others (1988, seealso Coulter and others 1992). This takes theform:

Scale = (N)θ

where N is the number of people in thehousehold,

and θ (theta) is a power between 0 and 1.

If N is raised to the power of 0, the scale is 1,however many people there are in the house-hold. At a power of 1, the scale equals thenumber of people in the household. At apower of 0.5 (for example) the scale is thesquare root of the number of people: 1.0 for 1person, 1.41 for 2, 1.73 for 3, 2.00 for 4, andso on. The advantage of this formula is that acontinuous family of scales can be introducedin between the logical maximum andminimum.

Although the DSS/McClements scale isdifferently formulated (reducing the ‘value’ ofadditional people according to their age, rather

41


than their number), in practice its effect issimilar to a value of θ of about 0.7517.

Chart 2.8 shows the results. (The ‘other’ethnic group has been omitted for the sake ofclarity.) The proportion of white householdsbelow the ‘poverty line’ remained just under20 per cent, whichever equivalence scale wasused — because the white population containssuch a wide range of household sizes that anyrise in poverty in one sector would be offset bya fall in another, as the sensitivity of theequivalence scale increased or decreased18.

There was some slight variation in the povertyestimates for Caribbean households, accordingto the value of θ. At the left of the chart, theywere in a very similar position to whitehouseholds; but they were more likely thanwhite people to appear poor on the right handside of the scale.

Indian and Chinese households had rates ofpoverty very similar to the white population,except that they were both fairly sensitive tothe equivalence scale. The more weight wasattached to additional household members, themore of them fell below the poverty line. Theeffect was sufficient to move both groups frombelow the white rate with ‘weak’ scales, toabove the white rate when the scale was‘strong’.

17 This is established by calculating the correlationbetween the DSS scale and the experimental scale forvarious values of θ, on the actual sample of householdsin the FRS. The correlation peaks at between θ = 0.7and 0.8.18 Coulter and others (1992) using 1986 FES data,suggest that poverty rates appear high at large and smallvalues of θ, with the lowest rates observed at valuesbetween 0.6 and 0.8. The difference in the shapes of thecurves may be caused by technical differences in thecalculation of the ‘national average’. The main point inthe analysis presented here is a comparison between thelines for each ethnic group, which should not beaffected.

African households were also fairly sensitiveto the equivalence scale. But they were alwaysmore likely to fall below the poverty line thanany of the four groups mentioned so far, andthis is therefore a very stable result.

For Pakistanis and Bangladeshis, the povertyrate was hugely sensitive to the equivalencescale used. If the number of people amongwhom income is shared is virtually ignored(low θ) this group were no more likely to befound below half the national average thanothers. If extra people (including children) areassumed each to require almost as muchincome as a single person living alone (highθ), three-quarters were found to be in poverty.The finding that 62 per cent of Pakistanis andBangladeshis were below the HBAIbenchmark is because the McClements scale isroughly equivalent to a θ of 0.7.

At first sight it is highly surprising thatPakistanis and Bangladeshis were no morelikely to experience poverty than whites, ifhousehold size is heavily discounted (low θ).Does that mean that Pakistanis and Bangla-deshis would have adequate incomes, if theyhad only one or two persons per household?What about all the evidence onunemployment, economic activity andearnings, which would tend to place thesegroups low in the income scale even if familysize was not an issue? The answer is that theapparent adequacy of Pakistani andBangladeshi incomes if the needs of childrenwere minimal is an artefact of a social securitysystem which accepts that their needs are notminimal. As has been shown earlier in thischapter, a very high proportion of familiesfrom Pakistan and Bangladesh were out ofwork; a high proportion of those wereclaiming means-tested benefits; and evenamong those in work, many were claimingfamily credit. All these payments are stronglyrelated to family size. They would make someof the recipients reasonably well off, if they

42

ISER Report 98-1

spent none of it on their children, and thisexplains the position of Pakistanis and Bangla-deshis low down at the left of Chart 2.8.All the evidence suggests that children doimpose costs (Bradshaw 1993, Berthoud 1984,Berthoud and Ford 1996, Middleton andothers 1997), and that social security pay-ments err if anything on the low side.

Geographical variations in the distributionof income

It is well known that Britain’s ethnicminorities are not spread evenly across thecountry but tend to live in certain areas,especially London, the West Midlands andother large cities (Ratcliffe 1997). At theextreme, they represented less than half of oneper cent of the population of the County ofCumbria at the time of the 1991 Census; but asmany as 45 per cent of the population of theLondon Borough of Brent. Even within adistrict (such as Brent) there are pockets ofhigher and lower concentration. The pattern ofsettlement may be explained by a combinationof historical, social, economic andgeographical consider-ations which are outsidethe scope of this enquiry. It is relevant, though,to consider some of the possible consequencesof the uneven distribution of minorities for thecurrent study’s interest in patterns of income.Variations between areas in the availability ofemployment or of low-cost housing mean thatthe geographical distribution of income willalso be uneven, and it is widely assumed thatethnic minorities tend to be found in lowincome areas.

A separate study is being undertaken of thedetailed relationship between ethnic minoritypatterns of settlement and measures of urbandeprivation (Dorsett 1998); that study is basedon the Fourth National Survey of EthnicMinorities, linked to detailed Census dataabout sample members’ local areas ofresidence. Another small study is planned of

the geographical distribution of income amongthe population as a whole; that will be basedprimarily on the Family Resources Survey.The intention on this occasion is simply to addgeographical factors to the analysis of ethnicminorities’ incomes.

Each year’s Family Resources Survey sampleis based on 1,752 postal sectors which havebeen selected at random19. The data availableto us records simply the name of the localauthority district20 within which eachhousehold lives. The sample can not beinterpreted as representative of householdsliving in any particular district. But it isrepresentative of all households living in anytype of district, provided the classification bytype is broad enough to include a large numberof the sampling points. We have looked up the1991 Census results for each of the districtsrepresented in the sample, to establish twothings:

• The proportion of all residents in thedistrict who were members of minorityethnic groups. This is referred to as ethnicminority density. The national average was5½ per cent.

• The proportion of all economically activemen who were unemployed. This is referredto as the unemployment rate. The nationalaverage was 11½ per cent.

As expected, the ethnic minorities interviewedin the Family Resources Survey wererelatively common in the main conurbations

19 In formal terms, the sampling points were selectedwith probability proportional to size, from a stratifiedlist of postal sectors.20 The term ‘district’ is used here in its official sense,not simply as another word for area or locality. Thedistrict is the lower tier of local government, where twotiers exist: a district within a county; a metropolitandistrict within one of the former metropolitan counties;a London borough.

43


— especially London and the West Midlands.This has been

44

ISER Report 98-1

Table 2.9: Characteristics of district of residence

Note: Big cities include London, Birmingham, Manchester, Sheffield, Leeds, Liverpool, Newcastle and Glasgow. Thelast three of these have quite low levels of minority representation

well established elsewhere (eg Ratcliffe 1997),but the first column of Table 2.9 illustrates theseparation. Only 15 per cent of white peoplelived in one of the big cities; half of the Asiansin the sample, and higher proportions ofCaribbeans and Africans, were found there.

It stands to reason that ethnic minorities tendto be found in areas of high minorityconcentration. The direction of this associationis simply tautologous; but its strength is ameasure of the extent to which the minoritypopulation is segregated from the majoritygroup. The middle column of Chart 2.9 shows,as would be expected, that minorities wereoften found in districts of high minoritydensity, while whites were not. At least two-thirds of Caribbeans, Africans and Indianslived in districts where more than 10 per centof their fellow council tax payers were also ofminority origin. The concentration at districtlevel is slightly less for Pakistanis andBangladeshis, and less again for Chinese. Inspite of this evidence of concentration, though,members of every minority lived in districtswhere they were genuinely in a minority. (Thehighest recorded density, in Brent, is justunder half — 45 per cent.) The degree ofconcen-tration would be higher if the analysisfocused on a much smaller grain ofgeographical measurement such as the ward or

the enumeration district. Even so, few black orAsian people live in wards or EDs where theyoutnumber whites (EMiB, Chapter 6, Dorsett1998.)

The striking thing about minority patternsof residence is that so many live in areas ofhigh unemployment. Roughly one-fifth of allindividuals (and of all whites) lived in districtswhere the 1991 male unemployment rateexceeded 15 per cent (third column of Table2.9). More than half of black people, and morethan a third of Asians, lived in the districtswith the highest rates of unemployment. Forthe minorities, only about one in twenty livedin an area of low unemployment (comparedwith one fifth of whites, not shown in thetable). Thus you might expect minorities toexperience relatively high levels ofunemployment, and of poverty, simply as acorrelate of the types of area in which theytend to live.

Preliminary analysis of household incomesamong white pensioners was very similar,whatever the characteristics of the area inwhich they lived. This is not surprising, sincethe universal state pension makes up such ahigh proportion of the incomes available toelderly people; they would not be expected tobe especially sensitive to local effects. There

Row percentagesBig cities Ethnic minority

density greaterthan 10 per cent

Unemploymentrate greater than

15 per cent

Sample size

White 15 13 20 116620Caribbean 66 68 53 1550African 82 78 60 588Indian 48 67 37 1828Pakistani/Bangladeshi 48 54 44 1679Chinese 52 39 33 324Other 53 51 33 1987

45


were, in any case, very few pensioners amongthe samples of ethnic minorities. The analysiswhich follows therefore excludes all-pensionerhouseholds, and concentrates on householdswhere at least one adult was belowpensionable age.

It is commonplace to observe, or assert, thatmembers of minority groups tend to live in‘run-down’ areas. From this is often assumedthat the area, and the people living in it, are‘poor’. In practice, the variation in incomesbetween the households within an area are sowide that it is far from clear that the ‘poor’area is a very meaningful concept (Berthoud1976). The question now is the extent to whichlow incomes are associated with ethnicconcentration. To the extent that there is anassociation, there are three possible ways inwhich this might work:

• Minorities, having high rates of unem-ployment and poverty, can only findaccommodation in areas of cheap housing,where their neighbours are also likely to beunemployed and/or poor.

• Racial discrimination has excludedminorities from prosperous areas, so theyhave to live in deprived ones.

• Minorities tend to live near each other,because of chain migration and communityties. Because as a group they have highrates of unemployment and poverty, theplaces where they live register as deprived.

However this simple set of hypotheses has tobe adjusted for the fact that some minoritiesnow enjoy economic positions no worse, onaverage, than those of white people; and thatthere is substantial variation in incomes withineach group. It is not possible, though, tounravel all these effects on the basis of thematerial available here.

Because most white people live in one set ofdistricts and most minorities in another set, itis not easy to make direct comparisons of theeffects of area characteristics on householdincomes. Instead of the usual tables, therefore,we have calculated, for each ethnic group, howmuch higher the average household income is,in a district where the level of unemploymentis 10 percentage points higher than in someother district. It is to be expected that averageincomes should be lower in areas of highunemployment and in general this is thecase. The first line of Table 2.10 shows, forexample, that white households in districtswhere the unemployment rate was (say) 15 percent were £61 worse off than those in districtswhere only 5 per cent of men were unem-ployed. The relationship between unemploy-ment and income was much stronger forIndian and Chinese households. But it wasrather weaker for Caribbeans, and weaker stillfor Pakistanis and Bangladeshis. For Africans,the local rate of unemployment made nodifference to average incomes.

The middle line of the table shows that theaverage incomes of Indians and Chinesehouseholds living in a district where the ethnicminority density was (say) 25 per cent wouldbe £26 or £29 lower than in districts whereonly 15 per cent of the population were non-white. This relationship holds good even whenthe rate of unemployment has also been takeninto account, and suggests some polarisationbetween poorer members of these com-munities, who live in high density areas; andthe better-off families living in areas of lowerconcentration. For Caribbeans there was only aweak association between income and density.For Africans, Pakistanis and Bangladeshis, noassociation at all: they were equally poor,wherever they lived.

If white people were strongly antagonistic tothe idea of living among members of minoritygroups, one might expect only lower income

46

ISER Report 98-1

Table 2.10: Association between equivalent household income and characteristics of thedistrict of residence (non-pensioners only)(The table shows the size of the difference in household incomes associated with achange of 10 percentage points in the characteristic of the district)

Regression coefficientsWhite Caribbean African Indian Pakistani/

BangladeshiChinese Other

Unemployment rate -£61 -£33 +£3 -£101 -£14 -£106 -£60Ethnic minoritydensity +£16 -£13 £0 -£26 +£1 -£29 -£4Variance explained 2.6% 2.7% 3.0% 5.9% 2.3% 7.0% 4.3%Note: The first two lines are coefficients x 10, derived from simple regression equations. ‘Variance explained’ is 100 x

R2 from a multiple regression equation in which the independent variables were density, unemployment, thesquares of density and unemployment, and a dummy variable distinguishing London from elsewhere

whites to move to or remain in high densityareas — because they could not afford toexercise their preference for living amongwhite people. It is therefore important to findthat white families in high density areas weresignificantly better off than those who lived inareas of low concentration. This is asurprising, and perhaps an encouraging, result,though it would be a mistake to place toomuch weight upon it. Few whites actually livein areas of high concentration, and most of thedata contributing to the analysis is reallydistinguishing between people in low or verylow density districts.

So the FRS data suggest that there are somesystematic geographical influences onminority incomes. These are strongest forIndians and Chinese, and weakest forAfricans, Pakistanis and Bangladeshis. On theother hand, these area effects do not explainmuch of the very wide variations in incomelevels within ethnic groups (third row of Table2.10). Nor much of the differences betweenethnic groups: in every type of area, whites,Chinese and Indians had the highest levels ofincome; in every area, Africans, Pakistanis andBangladeshis had the lowest. Thus incomeseems to be determined more by who you are,and where you came from, than by where youlive now.

47


CONCLUSIONS

Review of the findings

Analysis of the distribution of income, andstudies of the economic position of ethnicminorities, have been two of the strongestareas of social research in Britain over the pastthree decades. So it is surprising that it hastaken so long for the two traditions to cometogether, to provide sound data about theincomes of ethnic minorities. Now that twosources have become available, the findingscomplement the results of existing studies ofminority employment. This analysis has notshown anything startling or unexpected; but ithas provided valuable additional informationon the composition of family incomes, the roleof the benefit system and the extent of poverty.Some of those findings may not be surprising,but they are still shocking.

One of the themes emphasised in the Intro-duction (page 6) was ‘diversity’. Britain’sethnic minorities have very different startingpoints — the colours of their skin, theirlinguistic, religious and other cultural back-grounds, and their pre-migration educationaland economic experiences. The fact that thevarious groups arrived at different periodsmeans that varying proportions of them wereborn and have been brought up in this country,and should no longer be seen as ‘migrants’.These differences between the minority groupsmay be at least as important as any commonfactors which distinguish them all from thewhite majority. Of course there are commonfactors too: all are at risk of discrimination orharassment; all live in urban areas with highunemployment rates (though the detailedpatterns of settlement vary); all include at leastsome adults who, as migrants, may beexperiencing the short-term consequences ofchanging their country of residence. Never-theless, it is to be expected that groups withdiverse origins have reached different

positions in Britain’s economic structure,rather than they should all occupy a slotlabelled ‘ethnic minority’.

The findings of this analysis of family andhousehold incomes have confirmed thediversity of outcome which had beensuggested by previous studies based on theLabour Force Survey (Jones 1993) and theFourth National Survey (EMiB). It may,therefore, be useful to summarise the findingsfor each minority group in turn, taking themroughly in order from best-off to poorest.

Chinese

This group is the hardest to characterise. Thesamples of Chinese people (in both surveys)are too small for detailed analysis, andestimates of their average income may not bevery reliable. But the Chinese do not seem tohave formed a group with so clear a profile assome of the other minorities. There are anumber of indications, though, that this is agroup which has prospered in Britain.

The average earnings of working Chinesefamilies were higher than for any other group,including whites. The same group oftenreported other market sources of income, andrarely received means-tested benefits. Whenwe analysed household incomes in relation tofamily size, the Chinese were well-representedat the top of the scale: the top quarter of theirhouseholds were better off than whites, andbetter off than any minority group. On theother hand, there was rather a wide gapbetween high and low incomes in thiscommunity, and the proportion of ‘poor’Chinese households was also fairly high: 28per cent, compared with 16 per cent for whitehouseholds.

48

ISER Report 98-1

Indians

The analysis of employment in the FourthSurvey placed ‘African Asians’ close to theChinese, while ‘Indians’ were rather worse off,close to Caribbeans. The Family ResourcesSurvey did not identify African Asiansseparately, and its ‘Indian’ group effectivelycovers both categories. Accordingly it tends tofall in between Chinese and Caribbeans —very close, in fact, to the white position.

A high proportion of Indian families containedat least one worker. Their average earningswere slightly higher than whites’. In fact thebest-off Indian working families had a higherlevel of income (in relation to basic needs)than any other group. Among non-workingfamilies, Indians were relatively well-providedfor by unearned income and insurance-basedbenefits, and depended less than any othergroup on means-tests. This allowed a fairproportion of non-workers to obtain an incomerather higher than the basic social securitylevel. Among pensioners, Indians had the bestoccupational pensions of any minority group,though they were less well-covered by thestate pension than whites or Caribbeans. Whentotal income was considered in relation to thenumber of people in the household, though,Indians fell somewhat behind white people,and came close to Caribbeans. This suggeststhat, while Indians were earning as much aswhites, the number of children and elderlypeople who had to be supported out of thesame income reduced this group’s overalllevel of prosperity.

Caribbeans

People of Caribbean origin, (including manyof those labelled in the Census as ‘Blackother’) have been in Britain longer than anyother of the non-white minority groups. Theyare therefore a test of the early hypothesis thattheir experiences might converge on those of

the white population after a generation or so.An alternative hypothesis, though, has beenthat Caribbeans might follow the previous in-flow of Irish migrants through an essentiallyworking-class trajectory, while Indians wouldbe more similar to previous generations ofJews through a professional trajectory (Peach1997b). In fact the Fourth Survey, and otherevidence, has suggested that Caribbeans are agroup with a substantial degree of internaldiversity — a polarisation between well-educated, successful black people on the onehand, and under-educated, unemployed peopleon the other. There were also signs that theexperiences of Caribbean men and womendiverge in ways which are not seen in othergroups. The FRS has not been an appropriatevehicle for a detailed exploration of thesecomplexities, because the sample ofCaribbeans was small, and because littleinformation is available on subjects other thanincome. The overall view is that Caribbeansare system-atically worse off than whitepeople (and Chinese and perhaps Indians),though they are much better off than otherminorities.

So, for example, the average earnings ofCaribbean families were £45 lower than thoseof white families, though still higher thanAfricans’, Pakistanis’ or Bangladeshis’. Thesame ordering could be seen when workingfamilies’ total incomes were expressed inrelation to their basic needs; Caribbeansappeared in the middle of the range of ethnicgroups. Caribbean non-working families werealso worse off than whites; but the importantconsideration here is the large number ofCaribbean one-parent families who depended(like their white equivalents) on socialsecurity. The Caribbean community is the firstof the post-war migrant groups to include asignificant number of pensioners — again,they were worse off than white pensioners.When it came to the overall analysis ofhousehold incomes, though, Caribbeans were

49


not so poor as might have been expected. Fewof them had high incomes, but the rate of‘poverty’ was only slightly higher than amongwhites, and slightly lower than the Indians andChinese who appeared more prosperous onother measures.

Africans

People have come to Britain from manydifferent parts of Africa. It is unlikely thatbackgrounds in places as far apart as Nigeria,Uganda and South Africa, or in non-Commonwealth countries such as Somalia orthe Congo, will contribute to the formation ofa homogenous ethnic group covered by thelabel ‘Africans’. Many of them came to Britainas students, and this makes them difficult tocompare with the long-term migrants from theCaribbean or South Asia. These were amongthe considerations which led to Africans beingomitted from the sample of the FourthNational Survey, and the FRS is therefore oneof the few sources describing their position. Ingeneral Africans’ incomes are similar to, butworse, than those of the Caribbeans to whomthey are sometimes compared. That is, theworst position of any of ethnic groupsanalysed, with the exception of Pakistanis andBangladeshis.

Thus, the average earnings of African familieswith at least one worker were nearly £60 lowerthan those of whites; though higher thanPakistanis and Bangladeshis by a similarmargin. Only about half of African familieshad a worker in any case. Measures ofhousehold income showed that few Africanshad penetrated into the upper reaches of thedistribution, but nearly one third of them werepoor — twice as many as among whitehouseholds.

Pakistanis and Bangladeshis

In some respects these two groups may berather different from each other. One camefrom the mainly Urdu- and Punjabi-speakingterritory to the west of India, the other fromthe Bengali- and Sylheti-speaking territory tothe east. The main period of Pakistanimigration occurred some years before thearrival of the Bangladeshis. And the twogroups moved to different parts of the country:the Pakistanis largely to Lancashire andYorkshire, Bangladeshis to London. The twogroups had one thing in common, though, atthe point of arrival — both communities areMuslim. And many of their post-migrationexperiences have been very similar. This hasnot been demonstrated in this report, but theFourth Survey (EMiB) showed that they hadsimilar family structures, similar rates ofemployment, similar earnings and similarhousehold incomes. The two groups cantherefore be considered together for manypurposes. They were strikingly — shockingly— the worst off ethnic groups in Britain.

This has appeared on practically every page ofthis analysis. Only half of potential workerfamilies actually included someone with a job.Among those with a job, average earningswere far below those in any other ethnicgroup. Even among working families, andeven after taking account of means-testedbenefit pay-ments, a substantial proportion ofPakistanis and Bangladeshis received less thanthe basic needs implied in social securitybenefit scales. Even the best-off Pakistani andBangladeshi households were around themiddle-income level experienced by whitefamilies. No less than 60 per cent of Pakistaniand Bangladeshi households were ‘poor’ onthe conventional definition — four times asmany as among white households, and twiceas many as among their nearest neighbours onthe poverty scale, the Africans.

Discussion

50

ISER Report 98-1

These findings about inequality of incomesbetween ethnic groups will contribute to widerdebates, both about social and economicstratification in Britain, and about the positionof ethnic minorities. It is clear that policymakers need to take account of these facts,although conclusions about the nature of thepolicy implications do not emerge directlyfrom the findings of this analysis. Threegeneral points are pursued here: first, about theextent of poverty among Pakistanis andBangladeshis; second, about the role of thesocial security system; and third about thenature of inequality between ethnic groups.

Poverty among Pakistanis and Bangladeshis

One of the conclusions of the analysis hasbeen that ethnic minorities are not necessarilyworse off than white people, and that povertyis by no means an automatic consequence ofnon-majority status. Another is that, at least onsome measures, people of Caribbean orAfrican origin are sufficiently worse off thanwhites to be a focus of concern. But thefindings about Pakistanis and Bangladeshisoverwhelm all others. Poverty is their mostcommon experience.

If we leave out social security benefitentitlements for the moment, the ‘equivalent’income of a household might be expressed inthe following simple terms:

the man’s wage rate x his chance ofemployment

+ the wife’s wage rate x her chance ofemployment

÷ the number of people in the family

At one level, Pakistanis’ and Bangladeshis’poverty can be explained by the fact that everyone of their terms in the numerator of thisformula (the first two lines) is exceptionally

low, while their term in the denominator (thethird line) is exceptionally high. Men in thesegroups have very poor employment prospects.Very few women have a paid job. Theearnings of those men and women who dowork are far lower than those of white people,or of any other ethnic group. Householdstypically contain more adults than is commonin Britain nowadays; and far more children.Thus simple arithmetic shows that Pakistanisand Bangla-deshis will have low averageincomes in relation to household size. And, asthe report of the Fourth Survey showed, allthese elements are directly implicated. IfPakistani and Bangladeshi employment rateswere improved, or their earnings were raised,or their household sizes were reduced, therewould of course be a reduction in poverty(EMiB, Chapter 5). But no one of thesechanges would be enough to return Pakistanisand Bangladeshis to the middle ranks ofeconomic prosperity. All of the changes wouldbe required, in combination.

The formula in the previous paragraph mightbe said to describe Pakistanis’ and Bangla-deshis’ low incomes, without necessarilyexplaining them. True explanations wouldneed to address the components of the formula— why do these groups have such lowemployment rates and earnings, and such largefamilies? Three types of explanation might beput forward: educational background, culturalcharacteristics and industrial structure.

Educational background should be interpretedin its widest sense, to include knowledge ofEnglish and relevant training and work-experience as well as formal educationalqualifications — the ‘human capital’ which isso easy to theorise but difficult to measure.There were certainly signs in the FourthSurvey that Pakistanis and Bangladeshis hadlower levels of educational attainment thanother groups; that many of them had limitedexperience of the range of work which might

51


be on offer in Britain; and that some wereless than fluent in English (EMiB, Chapter 3).These might be partial explanations of a weakposition in the labour market, though by nomeans enough to explain the whole of thedifference between Pakistanis and Bangla-deshis and other groups. The Fourth Surveyshowed, for example, that within each ethnicgroup, individuals with a degree were lesslikely to be poor than (at the other extreme)those without qualifications. But the differencebetween groups was so great that a Pakistanior Bangladeshi with a degree had the samerisk of poverty as a white person with noqualifications at all (EMiB, Figure 5.4).

Pakistanis’ and Bangladeshis’ position on twoof the five terms of the simple income formulaquoted above can probably be explained interms of cultural preferences. The very lowrate of economic activity among women, andthe large family sizes, are clearly associatedwith the family relationships valued in Islamicteaching. The question arises, therefore,whether their poverty is linked to theirreligion, rather than to the countries of originby which they have been labelled. The FourthSurvey included a question about religiousaffiliation, and showed that Indian Muslimshad a higher rate of poverty than Hindus orSikhs, though it was still lower than that oftheir co-religionists from Pakistan orBangladesh (EMiB). It is possible that thesedistinctive family structures may converge onwhite British norms over the generations,though the signs are that this movement maybe slower where the distinctive pattern islinked to strong religious beliefs. Some whitepeople may argue that Pakistani andBangladeshi women ‘ought’ to go out to work,and have fewer children, if they wish to avoidpoverty. But this is precisely the kind ofcultural discrimination which lies at the heartof the problem. The family values revered byPakistanis and Bangladeshis are not muchdifferent from those accepted and admired in

Britain in the recent past. The minority groupcharacteristic is not ‘wrong’; just ‘different’from the currently dominant pattern.

52

ISER Report 98-1

A more conventional economic explanation ofPakistanis’ and Bangladeshis’ disadvantage inthe labour market lies in occupational andindustrial structures. Pakistanis migrated inlarge numbers in the 1970s to work in thetextile industries of Lancashire and Yorkshire.They were needed then, but the industriescollapsed in the 1980s, and the workforce wasbeached by the ebbing tide. The bulk ofBangladeshis, on the other hand, moved toeast London in the 1980s. Many of them tooksemi-skilled work in the catering industry (egas waiters and kitchen hands in ‘Indian’restaurants), but there is a limit to the numberof jobs in such sectors, and Bangladeshis mayhave suffered from a local glut in the supply ofthe types of labour they were able to offer.

The role of the social security system

The social security system lays down the samerules for everyone. There is some evidencethat discrimination by officials (Gordon andNewnham 1985), or their own inexperience ofthe system (Bloch 1993, Law and others1994), may mean that ethnic minorities do notget as much out of the system as white peoplein similar positions. But we have notinvestigated that issue here. It is possible thatethnic minorities do not claim all the benefitsthey are entitled to; the samples of potentialclaimants were too small to measure take-upon this occasion, though it will be feasiblewhen more years of FRS data have beenaccumulated.

Nevertheless, the analysis has yielded someimportant findings about means-testedbenefits. The indignities associated with thehousehold means-test were a hated element ofunem-ployment assistance during thedepression years of the 1930s. The Beveridgesocial security plan was intended to minimisemeans-testing, but a small element remained,and has grown steadily ever since, to accountfor a third of social security spending in the

1990s. The balance between means-testing andcon-tingent or insurance-based schemes is atthe heart of current debates. The proponents ofmeans-tests argue that they are highly effectiveat maintaining incomes at a minimum level,without wasteful expenditure above that level.Their opponents argue that they create a senseof dependence on the state, and reduce workincentives; to them, the widespread use ofmeans-tested benefits represents the failure ofother, more dignified, schemes.

The striking finding of this analysis has beenthe extent to which ethnic minorities ingeneral, and Pakistanis and Bangladeshis inparticular, rely on means-tested provision.This was especially true of working coupleswith children. Only 8 per cent of white,Caribbean, Indian or Chinese working coupleswith children received any means-testedbenefit. The figure was 24 per cent forAfricans with this family structure, and 40 percent for Pakistanis and Bangladeshis. Ofcourse, the benefits were designed for familieswith low earnings and many children, soPakistanis and Bangladeshis are in the centreof the target population. The consequence,though, is that even for those able to findemployment, the family income is oftendetermined, in the end, as much by the scalerates of social security benefits as by the actualearnings.

When you also take account of the receipt ofmeans-tested benefits by the large number ofnon-employed families in some ethnic groups,the picture is even more striking. £7 in every£100 received by white families consists ofmeans-tested benefits — mainly IncomeSupport, Housing Benefit and council taxbenefit. This is widely considered — on boththe left and the right of the political spectrum— to be far too high a figure. It is a drain onpublic expenditure. It keeps families at povertylevels of income, and dependent on the state.

53


The proportion of income accounted for bymeans-tests was very similar for the tworelatively prosperous minority groups, theIndians and Chinese. It was rather higher forCaribbeans (£17 in every £100). It was muchhigher though for the two poorest groups,Africans and Pakistanis/Bangladeshis: forevery £100 they had to spend, £27 and £34respectively were accounted for by means-tested benefits. These are astonishingly highfigures, and speak volumes about the scarcityof other resources for these families.

Inequality in multi-cultural Britain

In one sense, the factual findings of thisanalysis speak for themselves. Some ethnicminority groups — Chinese and Indians —now enjoy incomes very similar to those of thewhite population. This is substantial progressfrom the days, in the 1960s and 1970s, whenall minorities appeared worse off than themajority. Many measures suggest that onesignificant minority — Caribbeans — is ratherworse off than the groups just mentioned, eventhough people of West Indian origin have beensettled in Britain for more than a generation.Most measures place a fourth minority group— Africans — further behind again. But thefinal pair of minorities — Pakistanis andBangladeshis — are much poorer than anyother group. It is worth saying again: they arefour times as likely to be in poverty as whitepeople.

The ‘facts’ derived from ‘objective’ surveyevidence nevertheless need some interpret-ation. The null hypothesis implicit throughoutthis study is that all ethnic groups might havesimilar levels of employment, of earnings andof income. At an empirical level, that hypo-thesis has been disproved. But it is worthquestioning the assumptions built into thehypothesis.

54

ISER Report 98-1

Chart 3.1: Proportion of total net income accounted for by means-tested benefits

First, should it be assumed that different ethnicgroups need similar incomes to achieve similarlevels of material welfare? A relative povertyline, of which ‘half average income’ is aconvenient example, is justified on thegrounds that ‘needs’ are determined by socialand cultural conventions, rather than by anyabsolute requirements for a minimum level ofconsumption. If so, cultural variations betweenethnic groups might have an effect onperceptions of need. To illustrate the pointwith a few examples:

• The meat-free diets adopted by manyHindus might reduce the cost of livingfor them; alternatively, the need to buyminority produce such as ghee or Halalmeat might increase the cost of living.

• Some Bangladeshi families were livingat third-world poverty levels beforethey migrated. Some African Asianswere wealthy before they were evictedfrom Uganda. Do these references toprevious experience affect the

benchmarks against which eachgroup’s incomes should be compared?

• The analysis of ‘equivalence scales’ atthe end of the previous chaptersuggested that measures of Pakistanis’and Bangladeshis’ poverty were highlysensitive to assumptions about the costof a child. How would the analysishandle the possibility that the cost of achild varied between communities?

Second, to what extent does the pattern ofinequality differ from what might be expected.An important distinction was made in theintroduction between ‘discrimination’ and‘disadvantage’. The former refers specificallyto unfair treatment at the hands of white-dominated institutions; the latter is a broaderconcept covering situations in whichminorities are worse off than whites, withoutdirect reference to the cause. Some forms ofethnic disadvantage may be entirelypredictable. You would not expect a landlesspeasant from Bangladesh to reach the sameeconomic position in Britain as a well-

0% 5% 10% 15% 20% 25% 30% 35%

Other

Chinese


Indian

African

Caribbean

White

Percentages

55


educated and prosperous business man fromUganda. Hence the need to distinguishbetween the character-istics the minoritiesbrought with them, and their experiences inthis country. A more exacting test of equitywithin Britain would lie in a comparison of theoutcomes for the grandchildren of migrantsfrom such different backgrounds.

These are questions both for politicalphilosophy and for empirical analysis. Theyare important issues for the future of multi-cultural Britain. Sophisticated issues ofinterpretation should not, though, be allowedto reduce the impact of the findings of thisstudy, and its predecessor. If some minoritiesare prospering in the period following theirmigration, two groups remain at astonishinglylow levels of income.

56

ISER Report 98-1

APPENDICES

A: Comparing the Family ResourcesSurvey with the Fourth NationalSurvey of Ethnic Minorities

Composition of the samples

Table A1 shows that the FRS’s distribution byethnic group was reasonably close to thatobtained by the 1991 Census. It also shows thesample sizes in each of the two surveys underconsideration: the Fourth National Surveyprovides larger samples of the minoritygroups, but the two years of the FamilyResources Survey offer a reasonable numberfor analysis.

Response rates for the income questions

A problem with all earnings and incomequestions is that a proportion of surveyrespondents are unable or unwilling to providean answer. Table A2 shows the rates of non-response to the two key questions in theFourth Survey. For whites, 6 per cent of bothmale and female full-time employees gave no

answer on earnings, while 17 per cent ofhouseholds provided no information on theirtotal income. The proportions were verysimilar for people of Caribbean origin, thoughrather more black women than whites declinedto report their earnings. But people of SouthAsian origin were much less able, or moreunwilling, to answer these questions. This isbound to cast some doubt on the reliability ofthe data that was provided.

If certain minorities were more reluctant thanothers to answer money questions, that mighthave affected their overall response rate to theFamily Resources Survey, but we have no wayof measuring non-response by ethnic group.The DSS has, however, been able to show theextent to which those who took part in the(1996-97) survey were unable or unwilling toanswer the key question about the earningsfrom their main job. Table A3 shows somevariation between groups, though it does notmatch the substantial level of earnings non-response by Asians in the Fourth Survey.

Table A1: Analysis of samples by ethnic group

1991 Census FamilyResources

Survey

FourthNationalSurvey

Percent ofpopulation

Percent ofpopulation

Number of adults Number ofadults

White 94.5 94.7 89,063 2,867Black Caribbean 0.9 1.0 906 1,205*Black other 0.2 0.2 na naBlack African 0.3 0.4 322 nilIndian 1.5 1.3 1,136 1,947Pakistani 0.9 0.8 605 1,232Bangladeshi 0.3 0.2 219 598Chinese 0.3 0.2 194 214Asian other/Other 0.9 0.9 841 nil

* The Fourth Survey combined Black Caribbean and Black other in a single category

57


Table A2: Non-response to earnings and income questions in the Fourth National Survey

Table A3: Non-response to the FRS question on earnings from head of household’s job

Column percentagesWhite Caribbean African Indian Pakistani/

BangladeshiOther

Non-response 4.4 9.7 7.4 5.4 1.6 11.6Sample size 10748 135 54 148 69 95

Distributions of earnings and of incomeamong white households

The earnings and income data from the twosurveys can be compared in two ways. Thissection asks whether the overall distributionswere the same, even though the questions bywhich the information was obtained weredifferent. Since the Fourth Survey selectedsamples of white people and of ethnic minor-ities which cannot legitimately be combined,this overall comparison is undertaken for thetwo samples of white people only. The nextsection will asks whether the two surveysshow the same differences between ethnicgroups.

The comparison should take account of thefact that the two surveys took place at slightlydifferent times. The Fourth Survey interviewswere spread between November 1993 andOctober 1994; the great majority of the whitesample (involved in the comparison in thissection) were covered in the winter of

1993/94. The FRS data analysed here wascollected continuously between April 1994and March 1996. Since the FRS was ratherlater, we would expect its earnings andincomes figures to be slightly higher.

Chart A4 compares the distributions ofearnings for men and women in both surveys.All the data refer to gross pay of employees intheir main job, and are confined to full-timeworkers. The graph is in a form in which, ifone survey tended to show lower earnings thanthe other, its curve would appear to the left ofthe other’s. In practice, the distributions ofwomen’s full-time earnings are so similar thatthe two curves are indistinguishable from eachother. For full-time men, the Fourth Survey(shown by the dotted line) recorded slightlylower earnings in the middle of thedistribution, so that the median from theFourth Survey was £284, compared with £309from the FRS. Over the full range of earnings,though, the two surveys are very consistent —

Column percentagesWhite Caribbean Indian/

African AsianPakistani/

BangladeshiProportion of FT employeesnot answering the earningsquestion: male 6 6 23 21

female 6 12 27 18Proportion of householdsnot answering the incomequestion 17 17 35 29

58

ISER Report 98-1

remarkably so when the differences in the method of asking the question are considered.Chart A4: Distribution of gross earnings of white full-time employees: two surveys

compared

0

25

50

75

100

£50 £150 £250 £350 £450 £550 £650 £750G ross weekly earnings

Cum

ulat

ive

perc

enta

ge

FNS m en FNS wom en FRS m en FRS wom en

M en

W om en

Chart A5 makes a similar comparison betweenthe two surveys’ measures of total nethousehold income — still focusing on thewhite samples. Here there are bigger differ-ences between the two distributions. TheFourth Survey had lower incomes at the footof the scale (the dotted line starts to the left ofthe solid line), but higher incomes at the upperend of the distribution (the dotted line crossesthe solid line, and ends up on the right). Themedian incomes are fairly similar (£236 forthe Fourth Survey, £220 for the FRS), but therange is much wider in the Fourth Survey: thelatter would indicate a greater degree ofinequality than the former. For example therange between the upper and lower quartileswas £317 in the Fourth Survey, but only £229in the FRS. We cannot be certain of thereasons for this difference, but if we assumethat the FRS provided the more reliablemeasure (at least among white people) it isconsistent with two well-known potentialproblems associated with crude income

questions: some high salary earners may havereported their pre-tax income even though thequestion asked for net income; and somehouseholds dependent on social security (or acombination of earnings and social security)may not have totted up all their benefits. TheFourth Survey’s question about total house-hold income therefore seems rather lessreliable than its earnings question, at least atthe upper and lower ends of the distribution21.

21 A technical conclusion is that the median may providea better indicator of ‘typical’ incomes in each ethnicgroup, than the mean. There are two ways of calculatingthe median (or other quantiles) from grouped data. Oneis to assume that all the cases in each group are situatedat the mid-point of the range of income covered. Theother is to assume that they are evenly spread acrosseach range. We have used the latter assumption inTables A6 and A7, so that ‘exact’ medians could becalculated by interpolation. This can be thought of, ingraphical terms, as the point where the cumulativedistribution crosses the 50 per cent line (see Figure A5).

59


Figure A5: Distribution of total net incomes of white households: two surveys compared

0

25

50

75

100

£50 £150 £250 £350 £450 £550 £650 £750

Net weekly household income

Cum

ulat

ive

perc

enta

ge

Family Resources Survey

Fourth National Survey

Comparisons between ethnic groups

For some purposes, the absolute value ofmeasures of income may not be so importantas the relative values for different groups inthe population. In this case, our primaryinterest may be whether specific minorityethnic groups are shown to receive more orless earnings or income than whites. Table A6compares the two surveys’ estimates ofmedian earnings for each main ethnic group.Again, the data refer to gross earnings of full-time employees, this time confined to men.There is some degree of consistency. Bothsurveys show Pakistanis and Bangladeshiswell below all other groups. Both placedCaribbeans and Indians below whites. But the

relative position of these two groups wasrather different: the Fourth Survey suggestedthat Caribbean men were very close to whites,and above Indians; the FRS placed Indiansrather higher than Caribbeans. The greatestinconsistency referred to the Chinese: theFourth Survey estimated a very low medianfor Chinese men, while the FRS gave a veryhigh figure. The samples of Chinese men weresmall in both surveys, and they wereexceptional in the range of their earnings, sothis problem should perhaps be consideredspecific to the Chinese group, rather thancasting doubts on the accuracy of the data as awhole.

Table A6: Median gross earnings of full-time male employees: two surveys compared

Fourth National Survey Family Resources SurveyWhite £284 £309Caribbean £279 £268Indian/African Asian £240 £279Pakistani/Bangladeshi £173 £230

60

ISER Report 98-1

Chinese £235 £342

61


Table A7: Median total net household incomes: two surveys compared

Fourth National Survey Family Resources SurveyWhite £236 £220Caribbean £196 £185Indian/African Asian £266 £283Pakistani/Bangladeshi £163 £210Chinese £308 £251

Table A7 does the same for total nethousehold income. Both surveys recordedIndians and Chinese above whites, but indifferent positions relative to each other.Again, Pakistanis/ Bangladeshis andCaribbeans were well below white householdsin both sources; but the two were inconsistentin the relative positions of these two groups. Ithas been seen that a large proportion ofPakistanis’ and Bangladeshis’ incomesconsists of social security benefits paid tolarge families, and it seems likely that thesewere under-counted in the Fourth Survey. Theapparent advantage of Pakistanis andBangladeshis over Caribbeans in the righthand column of Table A7 disappears oncefamily size has been taken into account.

B: Baseline Needs and ‘Available’Income

The ‘rediscovery of poverty’ in the mid 1960swas based on an assumption that the thenNational Assistance scale rates represented an‘official’ view of the standard of living belowwhich no-one should be expected to live (AbelSmith and Townsend 1965). National Assist-ance became Supplementary Benefit, andeventually Income Support. For many years,social assistance scale rates were used as aconventional ‘poverty line’, though no govern-ment ever accepted that they represented a truemeasure of minimum needs.

For the analysis in Chapter 1 of this report, amuch-simplified version of Income Supportrates has been used to provide a benchmark.

For each family (benefit unit) in the FRSsample, basic needs have been calculated as:

• the scale rates for an adult aged 25 to 59, orfor a couple in that age range

• plus the scale rates for their dependentchildren, according to age, including thefamily premium

• plus their housing costs (if this familyincludes the head of the household).

This formula is very similar to the simplestructure of national assistance, withoutcomplex variations in rates by age, norpremiums, nor detailed rules about whichhousing costs are allowed or not allowed atany particular stage in a claim. The implicitassumption is that these complexities do notreflect variations in actual need, but politicalresponses to a variety of issues such asincentives, deservingness and lobbying.

If a family’s actual net income is expressed inrelation to this baseline, it provides anindication of the level of resources which are‘available’ after its basic minimum needs havebeen met. Where ‘available’ income isnegative, this implies they do not have enoughto meet even their minimum needs, and are‘poor’.

It may be possible to develop measures of thissort, based on social assistance scales, formore systematic use in the analysis of low

62

ISER Report 98-1

incomes. For the moment, though, the formulais used simply as a convenient benchmark forcomparing the incomes of low incomefamilies.

References

Abel-Smith, B. and Townsend, P. (1965) ThePoor and the Poorest, Bell

Atkinson, A. (1992), ‘Measuring poverty anddifferences in family composition’,Economica vol 59

Berrington, A. (1996) ‘Marriage patterns andinter-ethnic unions’ in D. Coleman andJ. Salt (eds) Demographic Characteristicsof Ethnic Minority Populations, TSO

Berthoud, R. (1976) ‘Where are London’spoor?’ Greater London IntelligenceBulletin

Berthoud, R. (1984) The Reform ofSupplementary Benefit, Policy StudiesInstitute

Berthoud, R. (1988) Social Security and Race:an agenda, Policy Studies InstituteWorking Paper

Berthoud, R. (1998) Disability Benefits,Joseph Rowntree Foundation

Berthoud, R., Burghes, L. and Herbert, A.(1998 forthcoming) Dependence andIndependence, DSS

Berthoud, R. and Ford, R. (1996) RelativeNeeds: variations in the living standards ofdifferent types of households, PolicyStudies Institute

Bloch, A. (1993) Access to Benefits: theinformation needs of ethnic minoritygroups, Policy Studies Institute

Bone, M., Gregory, J., Gill, B. and Lader, D.(1992) Retirement and Retirement Plans,HMSO

Bradshaw, J. (1993) Budget Standards for theUnited Kingdom, Avebury

Brown, C. (1984) Black and White Britain,Heinemann

Brown, C. and Gay, P. (1985) RacialDiscrimination: 17 years after the Act,Policy Studies Institute

Buhmann, B., Rainwater, L., Schmaus, G. andSmeeding, T. (1988) ‘Equivalence scales,well-being, inequality and poverty:sensitivity estimates across ten countriesusing the Luxembourg Income Studydatabase’, Review of Income and Wealth,vol 34

Coleman, D. and Salt, D. (1996) DemographicCharacteristics of Ethnic MinorityPopulations, TSO

Coulter, F., Cowell, F. and Jenkins, S. (1992),‘Equivalence scales and the extent ofinequality and poverty’, The EconomicJournal, vol 102, no 414

Daniel, W. (1968) Racial Discrimination inEngland, Penguin

Deacon, A. and Bradshaw, J. (1983) Reservedfor the Poor: the means test in Britishsocial policy, Basil Blackwell and MartinRobertson

Denny, K., Harman, C. and Rooke, M. (1997)‘The distribution of discrimination inimmigrant earnings — evidence for Britain1974-93’, IFS Working Paper 97/19

Department of Social Security (1997a) FamilyResources Survey — Great Britain1995-96, TSO

63


Department of Social Security (1997b)Households Below Average Income: 1979to 1995/96, TSO

Department of Social Security (1997c) SocialSecurity Departmental Report — thegovernment’s expenditure plans, TSO

Dorsett, R. (1998) Ethnic Minorities in theInner City, Policy Studies Institute,forthcoming

EMiB, Ethnic Minorities in Britain, seeModood, Berthoud and others, 1997

Employment Policy Institute (1998)Employment Audit, Issue 8

Esam, P. and Berthoud, R. (1991) IndependentBenefits for Men and Women, PolicyStudies Institute

FitzGerald, M. and Uglow, G. (1993), ‘Ethnicminority income’, Home Office ResearchBulletin 33

Foster, K. and Lound, C. (1993) ‘Acomparison of questions for classifyingincome’, OPCS Survey Methods Bulletin 32

Gordon, P. and Newnham, A. (1985) Passportto Benefits: racism in social security,CPAG/Runnymede Trust

Jones, T. (1993) Britain’s Ethnic Minorities,Policy Studies Institute

Journal of Social Policy (1987) vol 16, no 2Special issue devoted to poverty measures

Law, I., Hylton, C., Karmani, A. and Deacon,A. (1994) Racial Equality and SocialSecurity Service Delivery, University ofLeeds

Marsh, A. and McKay, S. (1993) Families,Work and Benefits, Policy Studies Institute

McClements, L. (1997) ‘Equivalence scalesfor children’, Journal of Public Economics,8

McKay, S. and Marsh, A. (1994) Lone Parentsand Work, HMSO

Middleton, S., Ashworth, K. andBraithwaite, I. (1997) Small Fortunes:spending on children, childhood povertyand parental sacrifice, Joseph RowntreeFoundation

Modood, T. (1994) ‘Political blackness andBritish Asians’, Sociology, vol 28, no 4

Modood, T., Berthoud, R., Lakey, J., Nazroo,J., Smith, P., Virdee, S. and Beishon, S.(1997) Ethnic Minorities in Britain:diversity and disadvantage, Policy StudiesInstitute

NACAB (1991) Barriers to Benefits: blackclaimants and social security, NationalAssociation of Citizens Advice Bureaux

Pahl, J. (1989) Money and Marriage,Macmillan

Peach, C. (1996) ‘Does Britain have ghettos?’Transactions of the Institute of BritishGeographers 21

Ratcliffe P. (ed, 1996) Social Geography andEthnicity in Britain: geographical spread,spatial concentration and internalmigration, TSO

Smith, D. (1977) Racial Disadvantage inBritain, Penguin

Smith, P. and Prior, G. (1997) The FourthNational Survey of Ethnic Minorities:technical report, SCPR

Wilson, G. (1987) Money in the Family,Avebury

64

ISER Report 98-1

Wilson, W. J. (1987) The TrulyDisadvantaged: the inner city, theunderclass and public policy, University ofChicago Press

Some ethnic minorities have prospered inBritain, but others remain severely dis-advantaged. Until recently there has been noreliable information on the incomes availableto minority households — from earnings,social security benefits and other sources.Richard Berthoud has been analysing the newFamily Resources Survey. He shows that thereis wide diversity between minority groups.Some are in serious poverty.

This ISER Report points to three importantpolicy issues. The first, and most striking, isthe extent of poverty among Pakistanis andBangladeshis. Second, the new analysisdemonstrates the importance of the socialsecurity system, and of means-tested benefits,to minority groups. Third, the findings raiseissues for discussion about diversity anddisadvantage in multi-cultural Britain.

INSTITUTE FOR SOCIAL AND ECONOMIC RESEARCH

University of EssexWivenhoe ParkColchesterCO4 3SQ

Telephone: 01206 872938Facsimile: 01206 873151

ISSN: 1464-8350ISBN: 1 85871 2009

Richard Berthoud - iser.essex.ac.uk · Author: Richard Berthoud, Institute for Social and Economic...

Documents

Transcript of Richard Berthoud - iser.essex.ac.uk · Author: Richard Berthoud, Institute for Social and Economic...