Ethnic minority poverty and disadvantage in the UK Lucinda Platt Institute for Social & Economic Research University of Essex Institut d Anàlisi Econòmica, CSIC, Barcelona
2 Focus on child poverty Scope National coverage (Great Britain / UK) Ethnic minority children Wider than immigrant children: not all are the children of immigrants but majority are and most have recent immigrant background Narrower than immigrant children: focus on specific groups, measured in surveys as ethnic categories Specific (larger, and coherent) groups only NB: main UK survey used for measurement of poverty has ethnic but not country of birth information. group
3 Non UK born, by ethnic group (%) Ethnic Group WOMEN MEN CHILDREN White British 3 3 1 Other White 61 59 32 White and Black Caribbean 13 16 2 White and Black African 58 56 12 White and Asian 29 37 6 Other Mixed 56 49 8 Indian 79 81 14 Pakistani 74 78 8 Bangladeshi 86 91 6 Other Asian 93 93 37 Black Caribbean 51 54 5 Black African 89 88 27 Other Black 58 52 10 Chinese 89 90 22 Source: Pooled Household Labour Force Survey datasets 2004-2008, own analysis
Distribution of ethnic groups 4 Ethnic group All Men Women Children < 16 White British 84.6 85.6 86.1 80.9 Other White 5.2 5.9 5.8 3.5 Mixed White & Black Caribbean 0.4 0.1 0.2 1.1 Mixed White & Black African 0.2 0.1 0.1 0.4 Mixed White & Asian 0.3 0.1 0.1 0.8 Other Mixed 0.2 0.1 0.2 0.5 Indian 2.1 2.0 1.7 2.3 Pakistani 1.6 1.2 1.1 2.8 Bangladeshi 0.7 0.4 0.4 1.3 Other Asian 0.7 0.7 0.7 0.8 Black Caribbean 1.0 0.8 0.9 1.2 Black African 1.3 1.0 1.0 2.2 Other Black 0.1 0.1 0.1 0.2 Chinese 0.5 0.5 0.5 0.3 Other 1.4 1.4 1.3 1.7 All groups 100 100 100 100 (397,668) Source: Platt (2009), Table 1 (from Household Labour Force Survey 2004-2008)
5 Background and Context Child poverty a general concern across European nations In UK specific child policy targets Extensive research on child poverty in UK and internationally Recognition not just of the significance of poverty but of persistent or long-term poverty for children s contemporary and later life outcomes However, little on ethnicity and child poverty: Analysis of ethnic minority disadvantage does not tend ot focus on poverty Analysis of child poverty typically not ethnically differentiated
6 But ethnicity matters 1. Ethnic differences in child poverty rates can be very large 2. If we are concerned about welfare, then poverty matters 3. Understanding ethnic differences in child poverty is informative about future trajectories of groups as well as current welfare 4. Differences across different groups can potentially refine, challenge or inform existing accounts of causal processes, impacts of poverty, or effectiveness of interventions 5. Understanding of the differences in dynamics of poverty can be especially pertinent in refining long-term policy agendas
7 Overview of paper How much child poverty is there in the UK? How does this vary across ethnic groups? Can this variation be understood in terms of acknowledged risk factors? How do poverty transitions of young children vary between ethnic groups? Can these be understood in terms of differences in risk factors? What are the key triggers for poverty entries and exits? Some conclusions
Data for measuring 8 (and monitoring) child poverty Family Resources Survey (FRS) and derived data set, Households Below Average Income (HBAI) Annual, cross-sectional survey, of c. 28,000 households, covering 32,000 benefit units ( families ) and around detailed Information on income sources and amounts of all household members. Also information on housing costs. And other demographic and supplementary information, including ethnic group of all individuals. HBAI calculates equivalent household income (net income adjusted for household size using modified OECD), and weights adjusted for response and lack of representation of top incomes. From this, poverty measure derived
9 Poverty Equivalent household income allocated to each member of household. Low income threshold: 60% of median equivalent income of all individuals Hence those poor are those with equivalent household incomes below this Before Housing Costs and After Housing Costs measures. Here BHC used
10 Child poverty in the UK (%) Year All Children 2002/3 18 23 2003/4 18 22 2004/5 17 21 2005/6 18 22 2006/7 18 22 2007/8 18 23 Child poverty greater than adult poverty Largely stable over this period Source: HBAI 2009 edition
11 Features of UK child poverty Child poverty rates higher in lone parent families families with no-one in paid work large families (4+ or 3+ children) families where a member of the household is longterm sick or disabled ethnic minority families
Child poverty in the UK, 12 2005/6-2007/8, by ethnic group Poverty rate % % of all children % of poor children White British 20 83.0 74.0 Indian 28 2.5 3.2 Pakistani 56 2.9 7.5 Bangladeshi 65 0.8 2.4 Black Caribbean 25 1.5 1.7 Black African 34 2.1 3.3 Source: Household Below Average Income [HBAI] data, 2005/06, 2006/07 and 2007/08, pooled, own analysis.
Children by type of family, 13 Source: Household Below Average Income [HBAI] data, 2005/06, 2006/07 and 2007/08, pooled, own analysis. by ethnic group (cell %) Lone parent 3+ children no worker Sick or disabled member 3+ adults White British 26 29 18 24 8 Indian 10 30 13 19 21 Pakistani 16 64 30 28 25 Bangladeshi 16 69 42 29 23 Black Caribbean 60 31 32 29 6 Black African 45 55 41 16 7
Decomposing poverty gaps Divides the poverty gap between two groups into that accounted for by differences in characteristics and an unexplained residual. The contribution of each characteristic to the gap is calculated by replacing the distribution in minority group with that in the reference group. The residual can be attributed to different chances of poverty associated with a given characteristic (differences in coefficients), holding other characteristics constant, or to unobservable differences between the groups. The explanatory variables were those associated with higher poverty risks, namely: family type, family and household size; disability within the household; work status of the household. Plus age of child; housing tenure; and region. Acknowledgement: Using Fairlie s approach for decomposition of binary outcomes, implemented using stata program fairlie developed by Ben Jann
Source: Household Below Average Income [HBAI] data, 2005/06, 2006/07 and 2007/08, pooled, own analysis. Share of poverty gap explained Indian Pakistani Bangladeshi Black Caribbean Black African Poverty gap (% points) 8.2 36.8 45.3 5.5 14.2 Explained (% points) -2.2 9.5 15.4 2.2 7.2 % explained -27 26 34 39 50 % unexplained: 127 74 66 61 50 N 896 1089 304 554 873
Contribution of selected characteristics Source: Household Below Average Income [HBAI] data, 2005/06, 2006/07 and 2007/08, pooled, own analysis. Explained % pts Indian Pakistani Bangladeshi Black Caribbean Black African -2.2 9.5 15.4 2.2 7.2 Lone parent 1.6 1.2 1.4-2.6-1.3 Workless hh -2.2 5.4 10.1 6.6 9.1 Housing -1.6-1.0 1.6 2.2 3.5 Region -0.4-0.2-2.2-3.3-4.1 3 plus children 0.1 1.8 2.2 0.0 1.0 3 plus adults -0.3-0.4-0.5 0.0 0.0 Sick adult or child 0.1-0.1-0.1 0.0 0.3
And the unexplained? Differences in returns to characteristics possible for region, housing tenure and work status. Among those in work, possibilities are: differences in earnings; differences in full-time workers in multiple adult households; among those not in work, differences in benefits (take-up and type); among both, possibilities are differences in savings and other sources of income.
18 The Millennium Cohort Study Cohort of c. 18,000 children sampled from all live births born between September 2000 and December 2001 from across UK, who will be followed over time Data collection (so far): when child was around 9 months; around three years old ; around 5 years old, and around 7 years old. Sample selected from a random sample of electoral wards, disproportionately stratified to ensure adequate representation of all four UK countries, deprived areas and areas with high concentrations of families from Black and South Asian minority ethnic groups Main respondents are predominantly mothers, but partners are also interviewed at each wave Detailed information on child health and development; parental pre- and post-natal behaviour; relationships within the family. It also has some information on income and some deprivation measures Country of birth not collected initially but subsequently added for those living in England (majority of sample). All cohort children UK born by definition. Almost all (over 90%) minority group mothers non-uk born immigrants.
Child poverty rate by (mother s) ethnic group Wave 1: % FRS equivalent Wave 2: % FRS equivalent Wave 3: % FRS equivalent Wave 4: % FRS equivalent 19 White 26 22 26 20 28 20 24 18 Indian 27 26 26 22 31 24 24 29 Pakistani 63 58 68 52 71 47 64 59 Bangladeshi 74 68 71 60 71 59 68 63 Black Caribbean 49 41 47 32 49 33 49 26 Black African 49 39 50 32 46 35 44 30 Source: MCS Waves 1-4, estimates adjusted for design effects and non-response, FRS and HBAI 2001/2-
Poverty rates wave1 20 Source: MCS Waves 1-4, estimates adjusted for design effects and nonresponse, FRS and HBAI 2001/2-2007/8, weighted. Own analysis conditional on later response % Poor wave1 crosssection % Poor wave1 waves1-2 responding % Poor wave1 waves1-3 responding % Poor wave1 waves1-4 responding Equivalent FRS White 26 26 24 24 22 Indian 27 21 24 21 26 Pakistani 63 61 59 60 58 Bangladeshi 74 69 64 61 68 Black Caribbean 49 45 40 44 41 Black African 49 47 42 39 39
21 Poverty patterns over 4 waves White Indian Pakistani Bangladeshi Black Caribbean Black African 0 20 40 60 80 100 Percentage share Never poor Poor one to three times Always poor Source: MCS Waves 1-4, estimates adjusted for design effects and non-response, own analysis
Focusing on waves 1 and 2
23 Poverty transition rates (row %) White Indian Pakistani Bangladeshi Black Caribbean Black African 0 20 40 60 80 100 Percentage share Not poor w1 or w2 Poverty exit Poverty entry Poor w1 and w2 Source: MCS Waves 1-2, estimates adjusted for design effects and non-response, own analysis
24 Modelling poverty transitions Probit regressions of probability of entry into poverty at T1 (given not poor at T0); and of exit from poverty (given poor at T0); and unconditional probability of being poor T1 and T2 Controls similar to FRS analysis (though family rather than household focus): work status of parents, 3+ adults, 3+children, lone or couple parents, age of mother (and square), disabled parent, housing tenure, region
Predicted probabilities of exit, by ethnic group 25 Other characteristics: age 30, in owner occupied housing, in London, fewer than 3 children, fewer than 3 adults, no sick parent. Exit probabilities significantly different from White for Pakistani and Bangladeshi parents. Couple 1+ in ft work Lone, not in paid work White Indian Pakistani Bangladeshi Black Caribbean Black African White Indian Pakistani Bangladeshi Black Caribbean Black African 0.2.4.6.8 Exit Probability Source: MCS Waves 1-2, estimates adjusted for design effects and non-response, own analysis
Predicted probabilities of entry, by ethnic group 26 Other characteristics: age 30, in owner occupied housing, in London, fewer than 3 children, fewer than 3 adults, no sick parent. Exit probabilities significantly different from White for Indian, Pakistani, Bangladeshi and Black Caribbean parents. Couple 1+ in ft work Lone, not in paid work White Indian Pakistani Bangladeshi Black Caribbean Black African White Indian Pakistani Bangladeshi Black Caribbean Black African 0.2.4.6 Entry Probability Source: MCS Waves 1-2, estimates adjusted for design effects and non-response, own analysis
Predicted probabilities of persistently poor by ethnic group 27 Other characteristics: age 30, in owner occupied housing, in London, fewer than 3 children, fewer than 3 adults, no sick parent. Exit probabilities significantly different from White for Pakistani and Bangladeshi parents. Couple 1+ in ft work Lone, not in paid work White Indian Pakistani Bangladeshi Black Caribbean Black African White Indian Pakistani Bangladeshi Black Caribbean Black African 0.1.2.3.4.5 Probability Poor waves 1 and 2 Source: MCS Waves 1-, estimates adjusted for design effects and non-response, own analysis
Direction of travel.
29 Source: Household Below Average Income [HBAI] data, 2001/2-2007/08, pooled, weighted, own analysis. Poverty trends over period 2001/2-2003/4 2002/3-2004/5 2003/4-2005/6 2004/5-2006/7 2005/6-2007/8 White British 20 20 19 20 20 Indian 28 28 30 27 28 Pakistani 59 56 53 54 56 Bangladeshi 72 66 64 58 65 Black Caribbean Black African 31 27 30 26 25 38 38 37 35 34
30 Conclusions (1) Children from minority groups face higher risks of poverty than majority; but there is great diversity between groups There is great variation across groups in factors that place them at higher risk of poverty But variation in poverty rates across children by ethnic groups cannot be fully explained by such variation in risk factors. Family circumstances and household work status are nevertheless strongly implicated in the poverty rates of some groups. Differences in poverty transitions are marked for Pakistani and Bangladeshi minority group children. And these also cannot be fully explained by family characteristics Events associated with entries and exits are much more likely to be demographic for minority ethnic groups Moves into work are less strongly associated with poverty exit among minority groups, but moves of the mother out of work are more strongly associated with poverty entry among minorities compared to the majority
31 Conclusions 2 Policy appears to have had some impact on those most disadvantaged, but large differentials remain Addressing differences in workless households on its own will not altogether remove ethnic differences in poverty Ethnic groups are defined by both composition and opportunity structures and these need further disentangling Children from minority groups will form a higher proportion of the future population. Therefore there are practical policy as well as equity reasons to address their differential poverty. This is relevant for European countries beyond the UK. Data (and will) are lacking adequate fully to monitor the poverty of subgroups of children and thence the futures of the second and subsequent generations.
32 Thank you! lplatt@essex.ac.uk
33 Data acknowledgements(1) Labour Force Survey Office for National Statistics. Social and Vital Statistics Division and Northern Ireland Statistics and Research Agency. Central Survey Unit, Quarterly Labour Force Survey Household Datasets: April - June, 2004 [computer file]. 2nd Edition. Colchester, Essex: UK Data Archive [distributor], October 2008. SN: 5464. October - December, 2004 [computer file]. Colchester, Essex: UK Data Archive [distributor], October 2008. SN: 6037. April - June, 2005 [computer file]. 2nd Edition. Colchester, Essex: UK Data Archive [distributor], October 2008. SN: 5465. October - December, 2005 [computer file]. Colchester, Essex: UK Data Archive [distributor], October 2008. SN: 6038. April - June, 2006 [computer file]. 3rd Edition. Colchester, Essex: UK Data Archive [distributor], October 2008. SN: 5500. October - December, 2006 [computer file]. 2nd Edition. Colchester, Essex: UK Data Archive [distributor], October 2008. SN: 5616. April - June, 2007 [computer file]. 2nd Edition. Colchester, Essex: UK Data Archive [distributor], October 2008. SN: 5716. October - December, 2007 [computer file]. 2nd Edition. Colchester, Essex: UK Data Archive [distributor], October 2008. SN: 5802. April - June, 2008 [computer file]. Colchester, Essex: UK Data Archive [distributor], October 2008. SN: 6034.
34 Data Acknowledgements (2) Family Resources Survey Department for Work and Pensions, Office for National Statistics. Social Survey Division and National Centre for Social Research, Family Resources Survey, 2005-2006 [computer file]. Colchester, Essex: UK Data Archive [distributor], November 2007. SN: 5742. Department for Work and Pensions, National Centre for Social Research and Office for National Statistics. Social and Vital Statistics Division, Family Resources Survey, 2006-2007 [computer file]. 2nd Edition. Colchester, Essex: UK Data Archive [distributor], July 2009. SN: 6079. Department for Work and Pensions, National Centre for Social Research and Office for National Statistics. Social and Vital Statistics Division, Family Resources Survey, 2007-2008 [computer file]. Colchester, Essex: UK Data Archive [distributor], July 2009. SN: 6252 Households Below Average Income Department for Work and Pensions, Households Below Average Income, 1994/95-2007/08 [computer file]. 3rd Edition. Colchester, Essex: UK Data Archive [distributor], July 2009. SN: 5828.
35 Data Acknowledgements (3) Millennium Cohort Study University of London. Institute of Education. Centre for Longitudinal Studies, Millennium Cohort Study: First Survey, 2001-2003 [computer file]. 9th Edition. Colchester, Essex: UK Data Archive [distributor], April 2010. SN: 4683. University of London. Institute of Education. Centre for Longitudinal Studies, Millennium Cohort Study: Second Survey, 2003-2005 [computer file]. 6th Edition. Colchester, Essex: UK Data Archive [distributor], April 2010. SN: 5350. University of London. Institute of Education. Centre for Longitudinal Studies, Millennium Cohort Study: Third Survey, 2006 [computer file]. 4th Edition. Colchester, Essex: UK Data Archive [distributor], April 2010. SN: 5795. University of London. Institute of Education. Centre for Longitudinal Studies, Millennium Cohort Study: Fourth Survey, 2008 [computer file]. Colchester, Essex: UK Data Archive [distributor], April 2010. SN: 6411. I am grateful to the Department for Work and Pensions, the Office for National Statistics and to the Centre for Longitudinal Studies for the use of these data, and to the UK Data Archive for making them available. None of these bear any responsibility for their further analysis or interpretation.
36 References Department for Work and Pensions (2009) Households Below Average Income: An analysis of the income distribution 1994/95 2007/08. London: DWP. Fairlie, R. W. (2006) An Extension of the Blinder-Oaxaca Decomposition Technique to Logit and Probit Models. IZA Discussion Paper No. 1917. Bonn: IZA. Jann, B. (2006) 'fairlie: Stata module to generate nonlinear decomposition of binary outcome differentials Platt, L (2009) Platt, L. Ethnicity and family - Relationships within and between ethnic groups: An analysis using the Labour Force Survey. London: Equality and Human Rights Commission.