Ethnicity and Social Housing Allocation in England: An Exploratory Analysis of CORE

Full text

Ethnicity and Social Housing
Allocation in England: An
Exploratory Analysis of CORE
March 2018
Helen Kowalewska – University of Southampton

2
Contents
Contents 2
List of Tables 4
List of Figures 6
1. Introduction 7
2. Ethnicity and Social Housing Allocation in England: A Review of the Literature 11
The Purpose of Social Housing 11
The Constrained Housing Choices of BME Households 13
Intersections between Ethnicity and Other Aspects of ‘Difference’ 15
Socioeconomic Factors 15
Household Size/Composition 16
Nationality 16
Age 16
Health 17
3. An Exploratory Analysis of the CORE 2016-17 Dataset 18
Regional Differences in New Social Lettings to BME Households 18
Two Step Clustering: Rationale and Procedure 23
Results 25
4. Disaggregating the BME Clusters 28
Key Characteristics of the BME Clusters 28
Cluster 14: BME Single Males 30
Cluster 15: Economically Inactive Black and Asian Females 32
Cluster 16: Economically Active Black and Asian Households 34
Cluster 17: Mixed, Chinese, and ‘Other’ Households 35
Cluster 18: BME Childless Couples and Single Females 36
5. Employment and Ethnicity Among Social Housing Tenants 37
Single Tenants 37
Lone Parents 42
Two-Adult Households (Under 60) 46
Understanding These Patterns 49

3
Location 49
Nationality 52
Age and Health 53
Additional Factors 54
6. Conclusion 56
Key Findings 56
Policy Implications 57
Suggestions for Future Research 59
References 61
Appendix 1 67

4
List of Tables
Table 1. Composition of CORE 2016-17 data by ethnicity and compared with general
population, %. ............................................................................................................ 8
Table 2. Top ten local authorities for new social lettings to Black households. ........ 20
Table 3. Top ten local authorities for new social lettings to Asian households. ........ 21
Table 4. Top ten local authorities for new social lettings to Mixed households. ....... 22
Table 5. Top ten local authorities for new social lettings to Chinese/Other
households. .............................................................................................................. 22
Table 6. Cluster solutions for new social housing tenants. ....................................... 26
Table 7. Taxonomy of social housing tenants based on Cluster Solution 2. ............ 27
Table 8. Composition of Clusters 14 to 18 by economic status, %. .......................... 29
Table 9. Characteristics of BME single males in Cluster 14 compared with White
single males in Clusters 1, 5, and 8.......................................................................... 31
Table 10. Characteristics of economically inactive, predominantly female Black and
Asian tenants in Cluster 15 compared with economically inactive, predominantly
female White tenants in Clusters 2, 3, and 10. ......................................................... 33
Table 11. Composition of Cluster 17 by ethnicity and household structure as a
percentage of total, %. ............................................................................................. 35
Table 12. Single tenants in CORE 2016/17. ............................................................. 37
Table 13. Single tenants by ethnicity and sex, %. .................................................... 38
Table 14. Results of the multinomial regression with White single tenants as the
baseline category. .................................................................................................... 42
Table 15. Economic status of lone parents within CORE by ethnicity, %. ................ 43
Table 16. Family characteristics of lone parents within CORE by ethnicity, %. ........ 44
Table 17. Results of the multinomial regression with White lone parent tenants as the
baseline category. .................................................................................................... 45
Table 18. Economic status of two-adult households with dependent children within
CORE by ethnicity, %. .............................................................................................. 46

5
Table 19. Results of the multinomial regression with two-adult households (aged
under 60) with a White head of household as the baseline category. ...................... 48
Table 20. Share of single tenants (new lettings and re-lets) from minority ethnic
backgrounds within local authorities with highest and lowest unemployment rates,
2016/17. ................................................................................................................... 51
Table 21. Table 13. Nationality of head of household among CORE households by
ethnicity, %. .............................................................................................................. 52
Table 22. Employment status of heads of households within CORE by nationality, %.
................................................................................................................................. 53
Table 23. Heads of household registered as sick/disabled or retired as a percentage
of ethnic group, %. ................................................................................................... 54
Table 24. Economic status of male single tenants, %. ............................................. 67
Table 25. Economic status of female single tenants, %. .......................................... 67

6
List of Figures
Figure 1. Distribution of CORE households and England population by ethnicity and
government region, %. ............................................................................................. 19
Figure 2. Distribution of new social lettings to BME households by Rural-Urban
classification, %. ....................................................................................................... 19
Figure 3. Ethnic composition of Clusters 14 to 18, %. .............................................. 28
Figure 4. Household types within Clusters 14 to 18, %. ........................................... 29
Figure 5. Economic status of single tenants in CORE by ethnicity, %. ..................... 38
Figure 6. Age profile of lone parents in CORE by ethnicity, %. ................................ 43
Figure 7. Age profile of heads of household by ethnicity, %. .................................... 53

7
1. Introduction
Social housing is housing provided by local authorities or private registered social
housing providers (also known as housing associations or Registered Social
Landlords) at below market rents. Hence, its central purpose is to provide a safety net
for the most vulnerable low-income households in society (e.g., Thornhill, 2010;
Fitzpatrick and Watts, 2017). Put simply, social housing is housing with a social
purpose (Oxley, 2001). Currently, around 17 per cent of all dwellings across England
are in the social rented sector (Department for Communities and Local Government,
2017). This represents a decline in recent decades. Right to Buy took around 1.8
million social rented homes out of the sector between 1979 and 2015 (Provan et al.,
2017). The problem of dwindling stock has been exacerbated by a slowdown in the
building of new houses since the the late-2000s recession (Thornhill et al., 2010).
Despite constrained supply, demand on the social rented sector remains steadfast
and will likely snowball in the coming decades, especially among Black and Minority
Ethnic (BME) communities. Data from the 2011 Census indicate that the White British
group comprises 80 per cent of the total population of England, down from 88 per cent
in 2001 (Office for National Statistics, 2012). The younger age structure of BME groups
means that the proportion of BME households within the population will only increase
in the coming years. Crucially, BME households’ lower average incomes, in a context
of rising house prices and a widening gap between rents in the private and social
sectors, mean that demand for social housing among BME communities - which is
already higher than for the ethnic majority - looks set to grow (Markkanen, 2009). The
greater numbers of extended and multi-family households among BME communities,
in addition to higher rates of overcrowding, also suggest the potential for increasing
numbers of households with unmet housing needs (Department for Communities and
Local Government, 2016).
Studies of ethnic residential segregation in England have highlighted ethnicity as an
important variable for understanding housing behaviour. Certain BME groups are
overrepresented in the social rented sector, while others are underrepresented. This
is illustrated in Table 1, which compares the ethnic composition of households in new
social tenancies and re-lets in 2016-17 with the total population in England and Wales.
It shows that while Black and most Mixed and ‘Other’ Minority Ethnic groups are
overrepresented in new social lettings, the Asian groups are underrepresented.
Yet, despite the overrepresentation of BME groups in the social rented sector,
previous studies indicate that BME households face additional barriers in entering this
sector and are more likely to be funnelled into the lowest quality and least desirable
properties due to various constraints on their housing choices (e.g., Henderson and
Karn, 1984; Harrison and Phillips, 2003; Reeve and Robinson, 2008). The apparent

8
ethnic penalty in accessing (quality) social housing is pressing, as it suggests that the
forecast growth in the size of the BME population could translate into greater
proportions of the population facing disadvantaged access to a welfare service
designed to meet the housing needs of the poorest in society. BME households’
disadvantages in accessing social housing also allude to the need for policies
designed to cope with the specific needs of BME communities and so enable equal
housing opportunities.
Table 1. Composition of CORE 2016-17 data by ethnicity and compared with
general population, %.
Ethnic Groupings
Composition of
households in new
social lettings and
re-lets, %
Composition of
total population
in England and
Wales, %
White
84.0
85.8
White: British
79.6
80.5
White: Irish
0.6
0.9
White: Other
3.8
4.4
Mixed
3.0
2.2
Mixed: White and Black Caribbean
1.4
0.8
Mixed: White and Black African
0.5
0.3
Mixed: White and Asian
0.4
0.6
Mixed Other
0.7
0.5
Asian
4.2
6.8
Asian/Asian British: Indian
0.7
2.5
Asian/Asian British: Pakistani
1.5
2.0
Asian/Asian British: Bangladeshi
0.8
0.8
Asian/Asian British: Other
1.2
1.5
Black
6.7
3.4
Black/Black British: Caribbean
2.3
1.1
Black/Black British: African
3.6
1.8
Black/Black British: Other
0.8
0.5
Chinese/Other
2.0
1.8
Other Ethnic Group: Chinese
0.1
0.7
Other Ethnic Group: Gypsy/Irish Traveller
0.1
0.1
Other Ethnic Group: Arab
0.5
0.4
Other Ethnic Group: Other
1.3
0.6
Source: 2016/17 Continuous Recording of Social Housing Lettings and Sales data; 2011 Census,
Office for National Statistics.
At the same time, there is potential for recent changes to social housing allocation
policies to impact negatively on BME communities. The Localism Act, which came

9
into force in June 2012 and gave greater controls to local councils over how to
allocate social housing, has limited access to social housing in many parts of the
country. Between June 2012 and April 2014, 126 local authorities had changed
their allocations policies, with many giving priority to households in employment or
seeking employment, or with a ‘local connection’ to the area under the guidance
of central government (Spurr, 2014). Because BME households are more likely
than White households to be out of work and be migrants, such changes to
allocations policies have the potential to indirectly impede BME groups’ access to
social housing (Douglas, 2014).
Against this background, this report maps out differences in access to social housing
by ethnicity and explores the characteristics of BME households in new social lettings
using the latest available COntinuous REcording of Lettings and Sales in Social
Housing in England (CORE) data. CORE is a census of new social lettings across
England, which covers both local authority and private registered social housing
providers, and provides detailed information on the demographic and economic
characteristics of households in new social lettings and re-lets. Accordingly, it is an
invaluable resource for monitoring which groups are currently accessing social
housing and which are not. Analysing the CORE data can reveal which BME groups
are currently underrepresented in new social lettings and so potentially face barriers
in accessing social housing. It can also tell us how BME groups in new lettings ended
up there, which can inform the direction of future policies towards helping BME groups
to access this welfare service on equal terms.
The report is divided into six chapters. The next chapter sets the background to the
analysis by outlining the purpose of social housing and reviewing extant research
evidence on the factors that underpin ethnic disparities in social housing. Chapter 3
then presents some initial descriptive analyses using CORE 2016-17 before carrying
out a cluster analysis of the data. The intention behind the cluster analysis is to
uncover the natural structure underlying the population of new social housing tenants,
which is currently lost within the sea of the approximately 375,000 households
contained in CORE 2016/17. Specifically, the cluster analysis reveals the groups or
‘types’ of people that are prevalent among new social lettings and how these differ by
ethnicity. Chapter 4 then builds on the third by unpacking further the socioeconomic
and other household characteristics of the BME groups identified in the cluster
analysis. This is to identify any unique characteristics of BME households in new social
lettings, thereby informing a more detailed understanding of the factors which can help
or hinder BME groups’ abilities to access social housing. Subsequently, Chapter 5
explores the associations between a tenant’s ethnicity and his or her employment
status. The purpose of this analysis is to paint a clearer picture of the barriers to
employment faced by BME social housing tenants, and the potential role of social

10
housing policies in reducing such inequalities. Chapter 6 concludes by summarising
the results, their implications for policy, and suggestions for future research.

11
2. Ethnicity and Social Housing Allocation in
England: A Review of the Literature
Studies of ethnic residential segregation in England since the 1960s have highlighted
disparities in the allocation of social housing by ethnicity. Evidence suggests that even
though some BME communities are overrepresented in the social rented sector, they
face greater difficulties in entering the social rented sector in the first place and are
more likely to be allocated less desirable properties in the poorest areas (e.g.,
Henderson and Karn, 1984; Harrison and Phillips, 2003; Reeve and Robinson, 2008).
This section reviews extant literature on how ethnicity potentially conditions a
household’s access to social housing.
1
To provide context for the discussion, the first
section briefly reviews the purpose of social housing. The second section then
discusses the prominent choice/constraints debate within the literature on BME
communities and housing. According to this literature, it is the interactions between
racial discrimination and cultural preferences that produce the disadvantages of BME
communities in accessing (quality) social housing. The third section then draws on
research evidence to consider how ethnicity might intersect with other markers of
social stratification (e.g., gender, household size, etc.) to produce ethnic disparities in
social housing.
The Purpose of Social Housing
The key purpose of social housing in England is to provide a safety net for the most
vulnerable low-income households in society (e.g., Thornhill, 2010; Fitzpatrick and
Watts, 2017). This was not the original intention, though. Initially, social housing was
too expensive for the poor (Bevan and Cowan, 2016). Hence, instead, it was allocated
based on ‘desert’, measured by waiting time and applicants’ housekeeping standards
(Fitzpatrick and Stephenson, 1999). It was only after the 1969 Cullingworth Report
that need displaced desert as the foremost allocation principle (Somerville, 2001). This
reflected the aims of moving away from the paternalism of allocation by desert, of
‘professionalising’ the sector by reducing the influence of elected officials on lettings
decisions, and of minimising unlawful discrimination by curtailing the scope for officer
discretion (Fitzpatrick and Pawson, 2007).
The Housing (Homeless Persons) Act 1977 obliges local authorities to provide social
housing to ‘unintentionally’ homeless households with a ‘local connection’, or who are
1
Unfortunately, a lot of the literature on this topic is quite dated, although this reiterates the need to
examine contemporary patterns of new lettings using the latest CORE data.

12
deemed to be in ‘priority need’. Under the 1996 Housing Act, priority need households
include families with a pregnant woman or dependent children, as well as ‘vulnerable’
single people and anyone made homeless or threatened with homelessness due to
an emergency, such as flood or a fire (Fitzpatrick and Stephenson, 1999). The scope
of priority need was expanded further by the 2002 Homelessness (Priority Need for
Accommodation) Order to include certain young people and other vulnerable
homeless applicants (e.g., victims of domestic violence) (Crisis, 2015). However, the
1996 Act had already reduced local authorities' maximum duty towards statutorily
homeless households, so that they are required to secure temporary accommodation
for a minimum of two years only (Fitzpatrick and Pawson, 2007). In addition, Section
167 (2A) of the Housing Act 1996 allows local authority to consider other factors for
determining the relative priority of applicants with similar housing needs. Examples of
such ‘other factors’ include the financial resources available to a person to meet their
housing costs and the behaviour of an applicant (Thornhill, 2010).
In addition to its role in addressing housing need, social housing serves broader social
and economic goals, including area regeneration, the creation of employment
opportunities, and social cohesion. For instance, housing association projects can
support a range of education and training opportunities, as well as jobs and enterprise.
In turn, higher levels of economic activity within a given area can lead to fewer rent
arrears and contribute to lower levels of crime, child poverty, drug abuse, and anti-
social behaviour (Mullins, 2010; Thornhill, 2010).
However, choice has also become an important principle in social housing allocation,
as in other public services across England. Traditionally, the process of social housing
allocation in England has been property-led. That is, councils matched vacant
properties with suitable applicants, with one household receiving a take-it-or-leave-it
offer of that property only (Pawson and Kintrea, 2002). Since the early-2000s,
however, there has been a shift towards an applicant-led ‘choice-based’ approach
(Pawson et al., 2006; Berry, 2007). Under the choice-based system, applicants for
social housing apply or ‘bid’ for any publicly advertised, available-to-let council and
housing association properties for which they are deemed eligible (Pawson et al.,
2006). Allocation is according to need and waiting time. The bidder with the highest
priority gets first refusal of a property, with the next bidder on the list getting second
refusal, and so on (Shelter, 2016). Strict rules limit the scope for housing staff, elected
councillors, and housing association committees to exercise unfair discretion in the
allocation process (Pawson and Kintrea, 2002). The precise rules of allocation vary by
area, and some councils supplement the choice-based lettings scheme with direct
offers of housing to applicants on council and housing association waiting lists
(Shelter, 2016a).

13
The Constrained Housing Choices of BME Households
Although allocation of social housing in England is intended to reflect need and BME
households are disproportionately from low-income backgrounds, extant research
suggests that BME households face various disadvantages in accessing social
housing. Some housing behaviour research suggests that the underrepresentation of
certain BME groups in social housing – most notably, South Asian groups – is rooted
in cultural preferences. For example, studies have shown a preference for owner
occupation among Pakistani communities, which reflects the prevalence and
desideratum of property and land ownership in Pakistan (e.g., Dahya, 1974; Ballard,
1994). It also helps to explain why home ownership is common among the Pakistani
ethnic group at 60 per cent (Barton, 2017). Other studies indicate that BME
communities may avoid social housing due to cultural perceptions of social housing
as a ‘welfare tenure’ or tenure of ‘last resort’, or as somewhere in which racial
harassment is more likely (e.g., Ratcliffe, 1981; Habeebullah and Slater, 1990; Chahal
and Julienne, 1999; Ratcliffe, 2001; Phillips et al., 2007).
However, there is evidence of growing demand for social housing among younger
generations of BME households (Markkanen, 2008). Hence, rather than being a
product of solely choice, most research on ethnic minority housing in Britain agrees
that the disadvantages of BME communities in accessing social housing are a product
of choice and structural constraints, with different studies placing varying emphasis on
either structure or agency (Beider, 2012). In other words, although BME individuals
may experience discrimination, they are not simply passive recipients of such
discrimination; rather, they exercise choices and preferences in relation to tenure,
property size, layout and location, albeit within a system of greater external constraints
and structural forces than for those of the ethnic majority (e.g., Peach, 1998; Robinson,
2002; Harrison and Phillips, 2003). In turn, the choices and actions of individuals can
influence the system at large. For example, in studying an Italian community in
Bedford, Sarre et al. (1989) found that private lenders often prevented Italians from
securing loans to buy homes, in part because of stereotypes of Italians as ‘unreliable’
and ‘feckless’. However, the subsequent loss of business, as it moved to more
progressive organisations, together with evidence that Italian homeowners were no
more likely to default on their payments than other groups, triggered lenders to reform
their practices.
According to the literature, broader structural inequalities can serve to constrain BME
households’ choices when it comes to applying for social housing and choosing which
properties to bid for in two main ways. First, social housing policies and processes,
which may otherwise appear fair and impartial, can inadvertently contribute to
restricting BME households’ housing choices by failing to address their specific needs

14
(Robinson, 2008). Policies often neglect that households lack perfect knowledge of
their entitlement to, and ways of accessing, social housing, which precludes the
exercise of rational choice (Robinson, 2008; Ratcliffe, 2009). For example, Pawson et
al. (2006) found that BME people who were not fluent in English faced difficulties in
accessing social housing under choice-based lettings, either because adequate help
and information was not available to them, or because they were unaware of the
existence of such support in the first place. Additionally, individuals without access to
computing facilities may struggle to make bids, especially on more desirable properties
that tend to be snatched up quickly (Beider and Netto, 2012).
What is more, because ethnic minority households are disproportionately represented
among the homeless and poorly housed, they are not necessarily as well placed as
(mostly White) transfer applicants to hold out for higher-quality housing. Consequently,
they may be forced to widen their ‘choice’ of areas and property types (Jeffers and
Hoggett, 1995; Cowan and Marsh, 2004). Statistics suggest that BME households are
around three times more likely to become statutorily homeless than the majority White
population, although rates of homelessness are highest among people of Black African
and Black Caribbean origins (Gervais and Rehman, 2005). The problem is that under
choice-based lettings, those in the greatest and most urgent need of housing may
have to bid ‘realistically’, i.e., for low demand properties that can be attained quickly
(Galbraith, 2017). Furthermore, the fear of having their priority status revoked may
lead applicants in the least secure circumstances to accept undesirable or unsuitable
properties (Dudleston and Harkins, 2007).
Failure to address racial harassment in the local area can also limit the housing
choices of BME households. Focus group research suggests that fears over racial
harassment can dictate many BME groups’ housing choices, with many actively
choosing to avoid certain areas perceived as racist (Markkanen, 2008). Furthermore,
where supposed ‘White’ or ‘no-go’ zones have a high concentration of social housing,
BME households may be deterred from applying for social housing altogether
(Robinson et al., 2007). Indeed, research suggests that even under choice-based
lettings, BME households are more likely than majority ethnic households to end up in
deprived areas and ethnic enclaves, thereby compounding ethnic segregation (Manley
and van Ham, 2011; van Ham and Manley, 2015; Gulliver, 2015). This is despite the
increasing openness of younger BME households to living in more ethnically diverse
areas (e.g., Phillips, 2008).
Second, broader inequalities and structures of state provision and regulation can lead
to or exacerbate discrimination against BME households in the allocation of social
housing (Robinson, 2002). For instance, the setting of ‘local connection’ rules to at
least two years under the encouragement of central government risks indirectly
discriminating against migrants, who are disproportionately BME (Douglas, 2014). The

15
discretion that local authorities have in defining local connection thresholds risks
subjecting the abilities of migrants to access social housing to a ‘postcode lottery’, as
local connection rules in some local authorities are lengthy. For instance, the London
Borough of Hillingdon gives preference to households that have lived in the borough
for at least ten years continuously.
Intersections between Ethnicity and Other Aspects of
‘Difference’
Of course, people have multiple identities rooted in other markers of social
stratification besides ethnicity, such as gender, age, religion, and class (Robinson et
al., 2005). Hence, ethnic distinctions in social housing allocation may be mediated, or
compounded, by other markers of social stratification (Platt, 2011).
Socioeconomic Factors
Phillips (1998) argues that socioeconomic class grouping also underpins the housing
patterns of BME groups in Britain. The socioeconomic disadvantages of certain ethnic
groups can increase their demand for social housing (Markkanen, 2008) and/or hinder
their competitive position within the social rented sector by potentially preventing them
for holding out for better-quality housing (Bowes et al., 2002). While people of Indian
and Mixed White/Asian heritage are slightly better off than White people, Pakistani,
Bangladeshi, and Caribbean ethnic groups are worse-off on average (Berthoud, 2005;
Clark and Drinkwater, 2007).
2
In addition, data indicate that Pakistani heads of
household are concentrated in low-paid employment and have high rates of
unemployment (Bowes et al., 2002; Markkanen, 2008), while rates of female
employment among Pakistani and Bangladeshi households are exceptionally low
(Peach, 1999; Clark and Drinkwater, 2007). Pakistani and Bangladeshi heads of
households also have lower levels of education compared with other ethnic groups
(Bowes et al., 2002). Changing educational profiles have, however, modified ethnic
differences in educational attainment, with British-born BME people achieving better
qualification levels than their migrant parents (Clark and Drinkwater, 2007). So, while
the average educational level of some BME groups is below that of the White British
group, young people from BME backgrounds are almost as likely to stay in full-time
education and enter higher education as those with a White British background
(Markkanen, 2008).
2
Although, as aforementioned, individuals of Pakistani origin are more likely to owner occupiers - as
are individuals of Indian origin - the housing quality and local area may be substandard (e.g., Darlington-
Pollock et al., 2017).

16
Household Size/Composition
Large families are prevalent among certain BME groups. For example, South Asian
elderly people commonly live with one of their sons, children often stay home longer,
and fertility rates among Pakistani and Bangladeshi households are higher than for
other ethnic groups. Yet, large properties (four or more bedrooms) comprise only a
tiny minority of social housing stock (Markkanen, 2008). Hence, demand for properties
to accommodate multi-family households cannot always be met within the social
rented sector, which may deter individuals from such communities from applying
(Bowes et al., 2002). Meanwhile, the high proportion of single person and single
mother households in the Caribbean community potentially contributes to the
historically higher-than-expected representation of Caribbean people in social
housing. Single-adult households may be less able to afford market rents or to buy a
property and are more likely to have their property size needs met in the social rented
sector.
Nationality
There is evidence that the perception that migrants are prioritised in the allocation of
social housing is unfounded (e.g., Robinson 2010). Data suggest that only a very small
proportion of new social housing lettings are to non-UK nationals, partly because most
are required to be a registered worker or to have secured habitual residence to be
eligible for social housing. However, available evidence also suggests that new
immigrants and migrants are often unaware of their social rights, the opportunities for
social housing, or how to access them. While refugees are eligible for social housing,
knowledge and understanding of the social rented sector can be patchy among
refugee communities, too (Robinson, 2008). Additionally, evidence shows that
refugees who do secure an allocation of social housing often end up in deprived
estates in low-demand areas that have been left behind by households who are able
to wait longer for better properties (Robinson, 2010).
Age
The younger age structure of BME groups also influences patterns of tenure among
the BME population. It underpins lower rates of home ownership among the BME
population overall, since saving for a deposit and earning enough for a mortgage take
time and have become increasingly hard for young people (Finney and Harries, 2013).
Furthermore, evidence suggests that younger BME social housing tenants face fewer
language barriers than their parents in accessing social housing (Wood, 2013). They
are also more likely to be more aware of, and able to access, welfare services
(Robinson et al., 2005). The abilities of BME young people to access social housing
may, however, vary by ethnicity. For example, research has shown that African

17
Caribbean young people, many of whom come from families with a tradition of social
renting, are more familiar with social renting options and how to apply than their South
Asian counterparts (Harrison et al., 2005; Simpson et al., 2007). At the same time,
Phillips (2008) argues that social housing is becoming increasingly important for BME
young people. She highlights how commentators have observed ‘hidden’ youth
homelessness among BME groups, which is particularly acute among Asian youth,
whereby young people are living in severely overcrowded households or are sharing
with friends. The greater vulnerability of BME youth to homelessness is rooted in the
disproportionate numbers facing debt, unemployment, and family disputes (see also
Watts et al., 2015).
Health
In terms of health, South Asian ethnic groups and Black Caribbean groups consistently
fare worse relative to the majority ethnic group. This contrasts with the relatively better
health of Black Africans and Chinese. Despite their relative health advantage, Whites
and White British are in poorer health than Bangladeshis and Black Caribbeans from
2001 when in social rentals, even after controlling for age. Hence, Darlington-Pollock
and Norman (2017) suggest that ethnic minorities in the poorest health may be less
able or willing to access social housing. They suggest that ethnic minority households
in the poorest health may instead be restricted to high-cost private rentals or poor-
quality owner occupier dwellings.

18
3. An Exploratory Analysis of the CORE 2016-17
Dataset
The COntinuous REcording of Lettings and Sales in Social Housing in England
(CORE) dataset contains around 375,000 cases for the year 2016/17. Hence, a cluster
analysis is the first logical step to uncovering the natural structure underlying this sea
of data, which is the aim of this chapter. Prior to the cluster analysis, the first section
maps out the geographical distribution of new social lettings to BME groups across
England. This is because rates of BME representation within new social lettings vary
by region and local authority; hence, the BME groups identified in the cluster analysis
are concentrated in certain areas. Accordingly, not all local authorities face the same
levels of demand from BME group or the challenges of addressing ethnic disparities in
new social lettings. The second section then explains the rationale and procedure of a
two-step cluster analysis. The third section presents the results of the cluster analysis.
Regional Differences in New Social Lettings to BME
Households
Housing is the most highly skewed of all welfare services; that is, there is a very strong
association between social housing investment and an area’s deprivation. In addition,
social housing is more prevalent in urban areas with dense populations and a relatively
high share of BME households (Cangiano, 2008).
Thus, Markkanen (2008) argues that the apparent overrepresentation of some BME
groups in social housing is at least partly explained by their geographical concentration
in regions in which social renting is more common. Analysing CORE 2006-07 data,
Markkanen found that new lettings to BME social tenants were concentrated in regions
in which social renting is more prevalent and BME groups comprise a larger share of
the total population. For instance, most BME households in social housing are in
London, and many London local authorities have disproportionately high percentages
of social housing out of total dwelling stock, as well as high BME populations.
These geographical patterns are confirmed in descriptive analyses of CORE 2016-17.
Figure 1 indicates that the proportions of social tenants from BME backgrounds are
higher in regions in which the BME population overall is larger, with London accounting
for around one-third of all new lettings to BME tenants and the West Midlands accounting
for 18 per cent of new social lettings to BME groups. Correspondingly, new social lettings
to BME households tend to cluster in urban, built-up areas (Figure 2).

19
Figure 1. Distribution of CORE households and England population by ethnicity
and government region, %.
Source: 2016/17 Continuous Recording of Social Housing Lettings and Sales data.
Figure 2. Distribution of new social lettings to BME households by Rural-Urban
classification, %.
Notes: Calculations are based on cases for which data on ethnicity are complete, i.e., were not refused
by the tenant; n = 303,633.
Source: 2016/17 Continuous Recording of Social Housing Lettings and Sales data.
0%
5%
10%
15%
20%
25%
30%
35%
40%
45%
North
East South
West East of
England East
Midlands South
East Yorkshire
and
Humber
North
West West
Midlands London
White (CORE) Minority ethnic (CORE)
White (Population) Minority ethnic (Population)
0%
10%
20%
30%
40%
50%
60%
70%
Mainly Rural Largely Rural Urban with
Significant
Rural
Urban with
City and
Town
Urban with
Minor
Conurbation
Urban with
Major
Conurbation
White Minority ethnic

20
What is more, within each region, BME social tenants tend to cluster into a small number
of local authorities only. The top ten local authorities for new social lettings to BME groups
in 2016/17 accounted for 31 per cent of all new social lettings to BME households.
Conversely, new lettings to the White British group are spread throughout the country,
with the ten top local authorities for new social lettings to White British households
absorbing only 15 per cent of new lettings to such households. Unsurprisingly, BME
households are underrepresented in local authorities with small BME populations. For
instance, 0.2 per cent and 0.3 per cent of new social lettings in North Norfolk and North
Devon went to BME households; in these local authorities, 98.6 per cent and 97.9 per
cent of the population respectively are White.
Breaking down new social lettings to BME households by ethnic grouping, Table 2
shows that the ten top local authorities for new social lettings to Black households
accounted for approximately 36 per cent of new lettings to such households. Among
the top ten districts are Birmingham, Manchester, and London boroughs with large
Black minorities and a high prevalence of social housing.
Table 2. Top ten local authorities for new social lettings to Black households.
Local Authority (LA)
Proportion of total
Black social
tenants in LA, %
Black population as
percentage of total
LA population, %
Social housing as
a percentage of all
dwellings in LA, %
Birmingham
9.6
8.9
24.8
Southwark
3.7
26.8
42.7
Manchester
3.6
8.6
31.0
Leeds
3.2
3.5
21.5
Lambeth
3.1
25.9
35.0
Croydon
2.8
20.2
17.2
Lewisham
2.7
27.2
30.1
Hackney
2.7
23.1
42.9
Waltham Forest
2.5
17.4
22.0
Sheffield
2.4
3.6
23.9
Average for all LAs
N/A
2.2
16.2
Sources: 2016/17 Continuous Recording of Social Housing Lettings and Sales data; Number of
Dwellings by Tenure and District: England 2016; 2011 Census: Ethnic Group, Local Authorities.
Similarly, the ten top local authorities for new social lettings to Asian households
accounted for 39 per cent of new lettings to such households. Again, the top ten
districts are Birmingham, Bradford, and London boroughs with large Asian minorities
and high levels of social housing (Table 3). Notably, 11 per cent of new lettings to
Indian households were in Leicester, 15 per cent to Pakistani were in Birmingham
while an additional 12 per cent were in Bradford, and 36 per cent of new lettings to
Bangladeshi households were in Tower Hamlets alone. This is despite the

21
underrepresentation of Asian groups in social housing – while comprising around 8
per cent of the whole population of England, Asian groups took only 4 per cent of new
social lettings in 2016-17 (Table 1).
Table 3. Top ten local authorities for new social lettings to Asian households.
Local Authority (LA)
Proportion of total
Asian social
tenants in LA, %
Asian population
as percentage of
total LA
population, %
Social housing as
a percentage of
all dwellings in
LA, %
Birmingham
9.8
25.4
24.8
Tower Hamlets
8.0
38.0
38.3
Bradford
5.3
26.4
15.1
Leicester
3.0
35.8
24.6
Manchester
2.8
14.4
31.0
Leeds
2.6
6.9
21.5
Sandwell
2.3
18.9
27.8
Sheffield
2.2
6.7
23.9
Newham
1.5
42.2
27.5
Newcastle upon Tyne
1.5
7.6
29.5
Average for all LAs
N/A
4.9
16.2
Sources: 2016/17 Continuous Recording of Social Housing Lettings and Sales data; Number of
Dwellings by Tenure and District: England 2016; 2011 Census: Ethnic Group, Local Authorities.
The underrepresentation of Asian groups in new social lettings in spite of their location
in areas dense with social housing indicates that either Asian households are less
likely to want to rent social housing, or face specific barriers in accessing social
housing. The literature review suggested that it is likely a combination of both choice
and barriers. As discussed, Asian groups are less likely to seek out social rented
housing because of more widespread negative perceptions of social housing and a
cultural preference for owner occupation (e.g., Ballard, 1994; Phillips et al., 2007).
However, the greater prevalence of properties that can accommodate multi-adult and
multi-family households in the private rented sector may also detract Asian households
from the social rented sector (Crofts, 2017). Indeed, rates of private renting increased
between 1991 and 2016 from 8 to 24 per cent among Indian groups, and from 11 to
23 per cent among Pakistani groups respectively (Finney and Harries, 2013; Barton,
2017). Similarly, rates of private renting among households with an ‘Other Asian’ head
are at 35 per cent compared with a UK average of 17 per cent. In addition, rates of
home ownership among the Indian group continue to be relatively high at 67 per cent
and 60 per cent respectively, compared with 65 per cent of all UK households (Barton,
2017).

22
Table 4. Top ten local authorities for new social lettings to Mixed households.
Local Authority (LA)
Proportion of total
Mixed social
tenants in LA, %
Mixed population
as percentage of
total LA
population, %
Social housing as
a percentage of all
dwellings in LA, %
Birmingham
7.0
4.4
24.8
Leeds
3.6
2.7
21.5
Sheffield
3.1
2.4
23.9
Manchester
3.1
4.7
31.0
Sandwell
2.5
3.4
27.8
Liverpool
2.3
2.5
26.9
Nottingham
2.3
6.7
26.9
Wolverhampton
1.8
5.1
26.6
Bristol, City of
1.8
3.6
20.4
Islington
1.7
6.5
40.8
Average for all LAs
N/A
1.8
16.2
Sources: 2016/17 Continuous Recording of Social Housing Lettings and Sales data; Number of
Dwellings by Tenure and District: England 2016; 2011 Census: Ethnic Group, Local Authorities.
Table 5. Top ten local authorities for new social lettings to Chinese/Other
households.
Local Authority (LA)
Proportion of total
Chinese/Other
social tenants in
LA, %
Chinese/Other
population as
percentage of total
LA population, %
Social housing as
a percentage of
all dwellings in
LA, %
Birmingham
9.5
3.2
24.8
Kirklees
5.1
1.0
15.5
Liverpool
4.4
3.5
26.9
Sheffield
4.0
3.5
23.9
Manchester
3.7
5.8
31.0
Leeds
3.6
1.9
21.5
Newcastle upon Tyne
2.8
3.6
29.5
Leicester
2.3
3.9
24.6
Hackney
1.9
6.7
42.9
Salford
1.7
2.2
27.9
Average for all LAs
N/A
1.3
16.2
Sources: 2016/17 Continuous Recording of Social Housing Lettings and Sales data; Number of
Dwellings by Tenure and District: England 2016; 2011 Census: Ethnic Group, Local Authorities.
Mixed groups are slightly less geographically concentrated, as the ten top local
authorities for new social lettings to Mixed households accounted for 27 per cent of
new lettings to such households. Again, the top ten districts include Birmingham and

23
Manchester; however, only one London borough (Islington) features in the top ten
(Table 4). Meanwhile, the ten top local authorities for new social lettings to
Chinese/Other BME households accounted for 39 per cent of new lettings to such
households (Table 5).
Two Step Clustering: Rationale and Procedure
Cluster analysis is a data reduction tool that involves organising cases into meaningful
groups. An important advantage of cluster analysis for the purposes of this study
concerns its robustness to violations of assumptions regarding the underlying
distributions of variables and their independence from one other (e.g., Chan, 2005;
Norušis, 2012). For example, a one-sample chi-square test revealed that the ethnicity
variable does not follow a multinomial distribution, as the observed number of counts
within each category differ from their expected values under a multinomial probability
model (p = 0.000). Furthermore, it is likely that many of the valuables that capturing
the different household characteristics are statistically associated with each another
(e.g., single-person households and age). Yet, because cluster analysis does not
involve hypothesis testing and calculations of observed significance levels, it still
performs well when continuous variables are not normally distributed, categorical
variables do not have a multinomial distribution, and the variables of analysis are
associated with one another (multicollinearity) (Norušis, 2012).
The choices of clustering algorithm and model parameters (i.e., the measure of
distance between cases and clusters and optimal number of clusters) depend partly
on the data that are being used (Suhr, 2014). Norušis (2012) advises using the IBM
SPSS Statistics two-step procedure when, as in this study, there is a large volume of
data, i.e. more than 1,000 cases. A two-step cluster analysis involves organising the
cases into an initial set of ‘pre-clusters’, from which the final clusters can be derived
(Norušis, 2012). By merging cases into a smaller number of pre-clusters and using
these pre-clusters in place of the raw data to develop the final clusters, a two-step
cluster approach provides for the determination of a more sensible and manageable
number of clusters than might otherwise be achievable with a large dataset. In turn, it
is easier to pick out the key patterns within the data.
An alternative approach to clustering, which can also cope effectively with a large
dataset, is the k-means cluster method. Yet, the IBM SPSS k-means procedure is
most appropriate when the variables used to cluster cases are strictly metric (Kent,
2015). This is because the k-means procedure uses a Euclidean measure of distance,
whereby (dis)similarities between objects are a function of the straight-line distance
between them. Accordingly, a shorter line indicates that objects are ‘closer’, i.e., more
similar, while a longer line indicates that they are less similar (Saitluanga, 2017). Such
a measure makes the k-means procedure unsuitable for the purposes of this study

24
given that we are interested in grouping social tenants by binary and nominal variables
that do not follow a natural order (e.g., gender and ethnicity). For such unordered
categorical variables, the absence of a natural metric precludes the derivation of an
arithmetically-based distance measure for capturing (dis)similarities between objects.
In contrast, the IBM SPSS Statistics two-step procedure allows for the selection of a
probability-based distance measure, whereby the distance between two observations
depends on the decrease in the log-likelihood of merging two clusters (SPSS, 2013).
As the name implies, the IBM SPSS two-step cluster analysis proceeds in two stages.
The first stage organises cases into a smaller number of pre-clusters using a
sequential clustering algorithm. This algorithm involves scanning cases, one-by-one,
to determine whether a given case should be merged into a previously formed pre-
cluster, or should instead form its own cluster according to the selected distance
criterion. The desired number of pre-clusters needs to be large enough to produce an
accurate result, but small enough to enable the efficient organisation of pre-clusters
into the final clusters (SPSS, 2013). The second stage then involves amalgamating
the pre-clusters into the final clusters using a standard hierarchical clustering
algorithm. This entails successively merging the pre-clusters into a larger number of
clusters until the desired number of cases is reached (agglomerative clustering). The
algorithm can determine the ‘optimal’ number of final clusters automatically using
either Schwarz’s Bayesian Criterion or Akaike’s Information Criterion. In addition, the
algorithm can standardise variables. It can also deal effectively with outliers - cases
that are very different from other cases but are not necessarily similar to one another
- by assigning them to an ‘outlier cluster’. The inclusion of an outlier cluster helps to
reduce the final number of clusters produced and increase the homogeneity of clusters
(Norušis, 2012; SPSS, 2013).
Cluster analysis always produces a solution, no matter how many variables or cases
used or the appropriateness of choices regarding these factors. Furthermore, in
hierarchical clustering, once a case has been assigned to a cluster, it cannot move to
a different cluster as the analysis proceeds (Cornish, 2007). Therefore, it is important
to determine quality of the cluster analysis and the conceptual meaningfulness of the
clusters.
There are various measures for quantifying the ‘goodness’ of the cluster solution. One
such measure is the silhouette coefficient. This measure is based on the average
distance between a given element within a cluster and all other elements in that
cluster, as well as between the given element and all elements in each of the other
clusters. The value of the silhouette coefficient can range from –1 to +1. A value closer
to -1 indicates a poor cluster solution. Conversely, a value closer to +1 indicates that
within-cluster distances are small (i.e. high in-group homogeneity), while between-

25
cluster differences are large (low between-group homogeneity), and therefore the
cluster solution is strong (Norušis, 2012).
Results
A series of cluster analyses were performed with the aim of achieving a silhouette
coefficient as close to +1 as possible and at a minimum 0.7, since this indicates a
strong structure to the data has been found (Kaufman and Rousseeuw, 2005). This
was very much a process of trial and error, which involved adding and subtracting
variables one-by-one until a silhouette coefficient of at least 0.7 emerged. Table 6
provides an overview of the strongest cluster solutions produced. To achieve a higher
silhouette coefficient, in certain instances the dataset was divided into different groups
initially (e.g., males versus females), with each group obtaining a separate cluster
solution.
Cluster Solution 2 offers the highest silhouette coefficient (0.8) even without the need
for an outlier cluster. This cluster solution groups households by the economic status
of the head of household, ethnicity of the head of household, and household structure.
The strength of the silhouette coefficient suggests that these variables are important
markers of stratification among new social housing tenants.
Table 7 presents the details of the clusters produced by Cluster Solution 2. Some
clusters are more homogenous than others. For instance, Cluster 1 contains strictly
White single males who are not available for work, whereas Cluster 16 encompasses
a variety of ethnicities (63 per cent are Black while 37 per cent are Asian), economic
statuses (83 per cent are employed while 17 per cent are unemployed), and household
structures (42 per cent are couples with children, 36 per cent are lone parents, and
the remainder are single males). Table 7 shows that single White people who are not
available for work (Clusters 1 and 2) comprise the largest group in the population of
new social housing tenants, accounting for around 31 per cent of the population. White
lone parents are also a significant population group (Clusters 3, 6, 14, and 15).
However, economic status varies widely within this subgroup. Moreover, Table 7
reveals that clusters can be divided into those containing strictly White households
(Clusters 1 to 13) and those containing strictly BME households (Clusters 14 to 18).
This suggests that ethnicity is an important dimension of variation among households
in new social lettings.

26
Table 6. Cluster solutions for new social housing tenants.
Clustering variables
Number of complete
cases1
Number of
clusters
Average silhouette
coefficient
‘Outlier’
cluster
Cluster Solution 1
Ethnicity of head
Household structure
Sources of income
Nationality of head
232,724
15
0.7
No
outlier
Cluster Solution 2
Economic status of head
Ethnicity of head
Household structure
286,724
18
0.8
No
outlier
Cluster Solutions 3-4, by
whether head of
household is White or not
Ethnicity of head
Household structure
Net weekly income
144,915 for White
29,632 for non-
White
8 for White
17 for non-White
0.8 for White
0.7 for non-White
No
outlier
Cluster Solutions 5-7, by
age categories
Ethnicity of head
Household structure
Net weekly income
40,580 for <25
108,491 for 25-59
25,038 for 60+
4 for <25
4 for 25-59
3 for 60+
0.7 for <25
0.7 for 25-59
0.7 for 60+
10% of
cases
Cluster Solutions 8-9, by
letting type: General
Needs (GN) and
Supported Housing (SH)
Ethnicity of head
Household type
Economic status of head
Net weekly income
126,669 for GN
45,858 for SH
13 for GN
8 for SH
0.8 for GN
0.7 For SH
10% of
cases
Notes: 1Complete cases are those for which data on the clustering variables are not missing. Any cases with missing data on these variables were excluded.
Sources: 2016/17 Continuous Recording of Social Housing Lettings and Sales data; own calculations.

27
Table 7. Taxonomy of social housing tenants based on Cluster Solution 2.
Cluster
Cluster Composition
Size
Economic Status
Ethnicity
Household Category
1
Economically inactive (100.0%)
White (100.0%)
Single male (100.0%)
15.9% (n = 45,538)
2
Economically inactive (100.0%)
White (100.0%)
Single female (100.0%)
14.9% (n = 42,727)
3
Economically inactive (100.0%)
White (100.0%)
Lone parent (100.0%)
8.8% (n =25,211)
4
Employed (100.0%)
White (100.0%)
Couple, dependent children (100.0%)
5.7% (n = 16,403)
5
In training/unemployed (100.0%)
White (100.0%)
Single male (100.0%)
5.5% (n = 15,889)
6
Employed (100.0%)
White (100.0%)
Lone parent (100.0%)
5.5% (n = 15,886)
7
Economically inactive (100.0%)
White (100.0%)
Couple, no children (100.0%)
5.2% (n = 14,840)
8
Employed (100.0%)
White (100.0%)
Single male (100.0%)
4.9% (n = 14,051)
9
Employed (100.0%)
White (100.0%)
Single female (100.0%)
4.5% (n = 12,851)
10
Economically inactive (100.0%)
White (100.0%)
Couple, dependent children (100.0%)
3.8% (n = 10,811)
11
Employed (100.0%)
White (100.0%)
Couple, no children (100.0%)
3.3% (n = 9,469)
12
In training/unemployed (100.0%)
White (100.0%)
Single female (100.0%)
2.8% (n = 7,932)
13
In training/unemployed (100.0%)
White (100.0%)
Lone parent (58.1%)
3.5% (n = 10,002)
14
Economically inactive (49.6%)
Black (39.4%)
Single male (100.0%)
4.2% (n = 12,003)
15
Economically inactive (100.0%)
Black (55.5%)
Lone parent (38.9%)
2.8% (n = 8,086)
16
Employed (83.4%)
Black (62.9%)
Couple, dependent children (41.6%)
3.7% (n = 10,589)
17
Economically inactive (43.8%)
Mixed (63.5%)
Single female (37.4%)
3.0% (n = 8,615)
18
Employed (48.5%)
Black (54.8%)
Single female (59.9%)
2.0% (n = 5,821)
Sources: 2016/17 Continuous Recording of Social Housing Lettings and Sales data.

28
4. Disaggregating the BME Clusters
The cluster analysis established that ethnicity is an important marker of differentiation
between households within new social lettings. This suggests that BME households
might face unique barriers in accessing social housing and require policies to cope
with their specific needs. Against this background, this chapter further disaggregates
the BME clusters (Clusters 14 to 18) and compares the characteristics of the more
homogeneous clusters with the characteristics and circumstances of White
households in CORE who are in similar situations.
Key Characteristics of the BME Clusters
Figures 1 and 2 and Table 3 give a breakdown of the ethnic, household, and
employment characteristics of Clusters 14 to 18.
Figure 3. Ethnic composition of Clusters 14 to 18, %.
Notes: Calculations are based on cases for which data on ethnicity, sex, and economic status are
complete, i.e., were not refused by the tenant; n = 45,114.
Source: 2016/17 Continuous Recording of Social Housing Lettings and Sales data.
Examining the clusters by other variables within CORE highlights certain similarities
between them. To begin with, the age profiles of these clusters are alike, in that most
are aged 25-39 years, with very few aged sixty and over. This reflects the young age
profile of BME households. Their profiles in terms of nationality are also very similar,
with around 69-75 per cent comprising UK nationals, 4-13 per cent comprising
individuals from other European Economic Area countries, and 17 to 26 consisting of
individuals from any other country. Among the population of White tenants in new
social lettings in 2016/17, just 92 per cent are non-UK nationals. Furthermore,
households contained within Clusters 14 to 18 are concentrated in London (32 per
cent) and the West Midlands (18 per cent), with one in ten living in Birmingham. Hence,
0% 20% 40% 60% 80% 100%
18
17
16
15
14
Black Asian Mixed Chinese/Other

29
households from these clusters are underrepresented among the South (11 per cent)
as well as within mainly or largely rural areas (4 per cent compared with 21 per cent
for White households in CORE 2016-17). Again, this reflects patterns of ethnic
segregation in the population more broadly. Most households in Clusters 14 to 18
obtained their current tenancy by applying directly (39 per cent) or through nomination
by a local authority (27 per cent). A significant minority (10 per cent) comprise internal
transfers. Additionally, 69 per cent have lived in the local authority area for more than
five years, with a minority (16 per cent) having lived there for under one year.
Figure 4. Household types within Clusters 14 to 18, %.
Notes: Calculations are based on cases for which data on ethnicity, sex, and economic status are
complete, i.e., were not refused by the tenant; n = 45,114.
Source: 2016/17 Continuous Recording of Social Housing Lettings and Sales data.
Table 8. Composition of Clusters 14 to 18 by economic status, %.
Cluster
14
15
16
17
18
Employed
9.4
0.0
83.3
38.8
48.9
Government training scheme
0.4
0.0
0.2
0.4
0.4
Unemployed
42.2
0.0
16.4
18.7
33.3
Retired
6.4
9.1
0.0
3.4
7.9
Not seeking work
12.1
59.4
0.0
23.8
3.6
Student
5.4
8.8
0.0
3.9
0.5
Sick or disabled
24.1
22.6
0.0
11.0
5.4
Total
100.0
100.0
100.0
100.0
100.0
Notes: Calculations are based on cases for which data on ethnicity, sex, and economic status are
complete, i.e. were not refused by the tenant; n =45,114. Columns may not sum to 100.0% exactly due
to rounding.
Source: 2016/17 Continuous Recording of Social Housing Lettings and Sales data.
0% 20% 40% 60% 80% 100%
18
17
16
15
14
Couple, no children Couple, children Lone parent Single male Single female

30
Cluster 14: BME Single Males
As Figures 1 and 2 and Table 3 show, Cluster 14 comprises a heterogeneous mix of
BME single male tenants in terms of ethnicity and employment status. Just over half
of BME single males in Cluster 14 are in Supported Housing tenancies. The
concentration of Cluster 14 tenants in Supported Housing likely reflects relatively high
rates of sickness and disability. In addition, 77 per cent of tenancies in Cluster 14 are
let by private registered social housing providers.
This cluster appears financially worse-off than a lot of other CORE households.
Notably, three-quarters of households in this cluster rely on state benefits for all their
income; this compares to 53 per cent of the whole CORE population. Furthermore, the
mean affordability rate
3
for this cluster is 0.53 in contrast to an average across all
CORE 2016-17 of 0.43.
In addition, 58 per cent of single males in Cluster 14 did not go through choice-based
lettings; this compares to 35 per cent for the whole CORE population. Still, most
tenants in Cluster 14 (53 per cent) applied directly for their current tenancy, although
a considerable proportion were nominated by their local housing authority (31 per
cent). This may because rates of prior homelessness are high within this group at 51
per cent, with many having previously been living with friends or family (21 per cent)
or in Supported Housing (14 per cent). Even so, only one-quarter of single males in
Cluster 14 were given reasonable preference. Within this cluster, the most common
reason given for leaving last settled home is overcrowding (18 per cent) compared
with 11 per cent for all CORE households.
An interesting point of comparison is between the BME single males contained in
Cluster 14 and the White single males found in Clusters 1 (economically inactive), 5
(unemployed), and 8 (employed). Comparing these groups can reveal if the
characteristics of BME single males in new social lettings, as well as they arrived at
their current tenancy, are significantly different from White single males. Hence, Table
9 details the key dimensions of difference between BME single males in Cluster 14
and White single males in Clusters 1, 5, and 8 combined.
3
The affordability rate is rent divided by income. Therefore, a high affordability rate reflects relatively
high rent and relatively low income, whereas a low affordability rate indicates low rent but high income.
In other words, the higher the value of the affordability rate statistic, the less affordable a property is for
that household.

31
Table 9. Characteristics of BME single males in Cluster 14 compared with White
single males in Clusters 1, 5, and 8.
BME Single
Males
White Single
Males
In employment
9.4
19.0
Unemployed
42.6
21.4
Economically inactive
48.0
59.6
In supported housing, %
53.0
46.8
In receipt of Housing Benefit/Universal Credit, %
90.7
81.1
Rates of sickness and disability, %
23.3
30.9
Dependent on state benefits for all income, %
76.2
66.0
Mean affordability rate
0.53
0.50
Obtained property through choice-based letting, %
41.6
54.5
Given reasonable preference, %
25.0
23.3
Previously homeless, %
50.6
32.9
Previously sleeping rough, %
6.7
6.1
Previously staying in Bed & Breakfast/other
temporary accommodation, %
11.0
6.2
Non-UK national, %
29.9
2.5
Notes: Calculations are based on cases for which data on the relevant variables are complete.
Source: 2016/17 Continuous Recording of Social Housing Lettings and Sales data
The table indicates several differences between the White single males and BME
single males. For a start, a greater proportion of BME single male tenants than White
single male tenants receive Housing Benefit or Universal Credit and are dependent on
state benefits for all their income. This may reflect the lower employment rate for BME
single males and slightly lower affordability rate. In addition, a greater proportion of
BME single male tenants than White single male tenants are in supported housing and
were previously homeless, as prior research has shown (Chapter 2), although the
absolute numbers of White single males who were previously homeless within the
CORE dataset are higher (24,261 compared with 7,709 BME single males).
Furthermore, and as previous research has also shown, rates of sickness and
disability are lower among BME single males than for White single males.
In addition, a slightly higher proportion of BME single males were given reasonable
preference, while lower proportions of BME single males obtained their current

32
tenancy through choice-based lettings. The lower use of choice-based lettings among
BME single males may be partly because of prior tenure. Although similar proportions
of BME as White single males were previously sleeping rough, a higher percentage of
BME single males were previously staying in temporary accommodation. As
highlighted in the literature review, the needs of those in the least secure housing
situations may be too urgent to participate in choice-based lettings or to hold out for
‘better’ choices. People who are homeless or staying in temporary accommodation
may be further disadvantaged in the bidding process by inadequate access to the
internet. The danger is that a lack of choice pushes such households into low demand
areas, thereby consolidating their marginalisation (Shelter, 2005). At the same time,
BME single males are more likely than White single males to be foreign nationals
(Table 9), which means they are more likely to face language barriers in accessing
welfare services or be less aware of such services.
Cluster 15: Economically Inactive Black and Asian Females
Cluster 15 contains Black and Asian households only (Figure 1). Within this cluster,
72 per cent of tenants are in General Needs housing, while 65 per cent rent from a
private registered landlord. Among tenants in Cluster 15, 31 per cent come from a
previous General Needs letting, while 13 per cent are from the private rented sector
and 17 per cent were living with friends and family. The mean affordability rate for this
cluster is 0.40, slightly better than the CORE average of 0.43.
Within Cluster 15, 92 per cent of households are headed by a female and all are
headed by an economically inactive individual (Table 3). Within this group, 22 per cent
of households are couples with children, 39 per cent are single females, and 39 per
cent are lone parent families. More broadly, one-quarter of all Black households in
CORE are lone parent families compared with 20 per cent of the whole CORE
population. This is in keeping with research that lone motherhood tends to be more
prevalent within Black groups, which may help to explain the overrepresentation of
Black groups in new social lettings. The high prevalence of lone parenthood in this
group also means that for many households in Cluster 15, economic inactivity likely
stems from childcare responsibilities. Indeed, 30 per cent of households in Cluster 15
contain a child aged under five, and in 80 per cent of such households there is only
one adult. In turn, high rates of economic inactivity mean that, as for the previous
cluster, tenants in Cluster 15 are disproportionately dependent on benefits for their
income, with 70 per cent deriving their entire income from benefits and a further 16
per cent relying on benefits for at least some of their income. The high prevalence of
lone mothers and single females in this cluster might also explain why domestic
violence is a prevalent reason for leaving the last settled home, with 18 per cent of
tenants in Cluster 15 giving this reason.

33
Again, it is instructive to compare economically inactive BME tenants in Cluster 15
with their White counterparts. These can be found in Clusters 10 (economically
inactive White couples with dependent children), 3 (economically inactive White single
females), and 2 (economically inactive White lone parents). Table 10 details the key
dimensions of difference between BME tenants in Cluster 15 and White tenants in
Clusters 2, 3, and 10 combined.
Table 10. Characteristics of economically inactive, predominantly female Black
and Asian tenants in Cluster 15 compared with economically inactive,
predominantly female White tenants in Clusters 2, 3, and 10.
Black and Asian
White
In supported housing, %
28.1
32.7
In receipt of Housing Benefit/Universal Credit, %
93.2
91.5
Rates of sickness and disability, %
20.7
26.1
Dependent on state benefits for all income, %
69.6
75.2
Mean affordability rate
0.40
0.39
Obtained property through choice-based letting, %
61.9
68.1
Left last settled home because of domestic
violence, %
17.9
9.3
Given reasonable preference, %
37.2
29.5
Previously homeless, %
40.1
22.8
Previously sleeping rough, %
0.5
1.0
Previously staying in Bed & Breakfast/other
temporary accommodation, %
16.1
6.4
Notes: Calculations are based on cases for which data on the relevant variables are complete.
Source: 2016/17 Continuous Recording of Social Housing Lettings and Sales data.
The table indicates several differences between the BME economically inactive,
predominantly female-headed households and White economically inactive female
households. Notably, although the shares of each group in Supported Housing and in
receipt of Housing Benefit or Universal Credit are roughly similar, rates of sickness
and disability are once again higher among the White group. In addition, although all
households are economically inactive, a higher proportion of White households
depend on state benefits for all of their incomes. Even so, rates of affordability are
similar across both groups.

34
Another important difference between the groups concerns how they came to be in
social housing. BME women within CORE are more likely to have been forced to leave
their previous home because of domestic violence. This is mostly explained by high
rates of domestic violence among South Asian women in CORE. Among the Indian
ethnic group, 26 per cent of all female tenants gave domestic violence as the reason
for fleeing their last home. The corresponding figrures for the Pakistani and
Bangladeshi groups are 32 per cent and 17 per cent respectively.
While domestic violence affects women of all ethnic groups in CORE, the specific
situations of South Asian and other BME women who experience domestic violence
can present them with unique disadvantages. Notably, BME women for whom English
is not their first language or who have recently arrived in the UK are at risk of not
knowing how to access social housing or being unable to access it altogether. In
addition, where mainstream services are not sufficiently culturally-sensitive, and
culturally-specific services lack the skills or expertise in such specialist areas as
domestic violence and housing, then BME women can be caught in a no-man’s-land
(e.g., Batsleer et al., 2002; Burman, 2003; Burman et al., 2004; Burman and Chantler,
2005). Accordingly, the evidence that a significant portion of South Asian women in
new social lettings are there because of domestic violence suggests that ensuring that
these women can continue to access social housing in the future should be a policy
priority.
In addition, BME women are more likely to have been previously homeless and to
have come from temporary accommodation. However, only a very small minority of
both BME and White female-headed economically inactive households were
previously sleeping rough. Even so, the fact that greater proportions of BME tenants
come from temporary accommodation may mean that they are forced to take the first
property they can get, which may indirectly exacerbate existing ethnic inequalities in
the quality of social housing (see Chapter 2).
Cluster 16: Economically Active Black and Asian
Households
Cluster 16 also comprises Black and Asian households only (Figure 1). Yet, unlike
Cluster 15, all households in Cluster 16 are economically active (Table 3), with 49 per
cent in full-time employment, 35 per cent in part-time employment, and 16 per cent
currently registered as unemployed. In addition, this cluster contains roughly equal
numbers of males and females and a mixture of household types (Figure 2). Because
they are economically active, tenants in Cluster 16 are, on average, financially better
off than those in Clusters 14 and 15. Within Cluster 16, around half of households do

35
not receive any income from benefits, and the mean affordability rate is 0.37.
Furthermore, only 6 per cent are in Supported Housing. Around three-quarters
obtained their current tenancy through choice-based lettings, a similar proportion as
for all economically active White households within CORE. In terms of previous tenure,
32 per cent came from a General Needs tenancy, 20 per cent from the private rented
sector, and 18 per cent from living with family and friends. The most common reason
for leaving the last settled home was overcrowding (26 per cent); this compares with
10 per cent of economically active White households in CORE, again reflecting the
higher risk of overcrowding among BME households. In addition, 32 per cent of
households in Cluster 16 were previously homeless, while 38 per cent were given
reasonable preference. The corresponding figures for White economically active
households are 23 per cent and 25 per cent. Hence, as for other household types,
economically active Black and Asian households in Cluster 16 have higher proportions
of prior homelessness and reasonable preference than White households in similar
circumstances.
Cluster 17: Mixed, Chinese, and ‘Other’ Households
Cluster 17 is a highly heterogenous cluster, which is broken down in more detail Table
11. The most prevalent groups within this cluster are Mixed lone parents and Mixed
single females. As for the Black ethnic group, lone parents are overrepresented among
all Mixed households in CORE. This is particularly the case for the White and Black
Caribbean group, of which 28.1 per cent of CORE households are lone parents, and
the White and Black African group, of which 24 per cent of CORE households are lone
parents.
Table 11. Composition of Cluster 17 by ethnicity and household structure as a
percentage of total, %.
Couple with children
Lone parent
Single female
Mixed
11.3
25.6
26.5
Chinese/Other
16.1
9.5
10.8
Source: 2016/17 Continuous Recording of Social Housing Lettings and Sales data.
The prevalence of lone parents and single females within this cluster means that the
majority of heads of households (81 per cent) of this cluster are women. The majority
(80 per cent) of tenants in this cluster are also in General Needs housing, and 68 per
cent are in private registered provider tenancies. Two-thirds of this groups acquired
their current tenancy through choice-based lettings, with 30 per cent originating from
another General Needs tenancy, 15 per cent from a private tenancy, and 19 per cent
from living with family and friends. The main reason for seeking social housing was

36
overcrowding (17 per cent). Within this cluster, 36 per cent of tenants were previously
homeless. The same percentage of tenants was given reasonable preference. In
addition, mean affordability is 0.40, and 74 per cent of households derive some or all
of their incomes from benefits.
Cluster 18: BME Childless Couples and Single Females
The final cluster contains mostly Black and Asian households as well as a small
proportion of Mixed and Chinese/other BME households (Figure 1). All households in
this cluster are childless. Around 40 per cent are couples, while the remaining 60 per
cent are single females. Hence, the majority of heads of household within this cluster
are, again, women (78 per cent, Figure 3). Most tenants in this category are in full-
time employment (32 per cent) or seeking employment (33 per cent). A further 17 per
cent are in part-time employment, while the remaining 18 per cent are economically
inactive (Table 3). Within this cluster, 60 per cent acquired their current tenancy
through choice-based lettings. Many were previously living with friends or family (29
per cent), while sizeable proportions come from another social sector tenancy (24 per
cent) or a private sector tenancy (15 per cent). Average affordability in this cluster is
the same as for single males in Cluster 14 at 0.53. But while 40 per cent of tenants in
Cluster 18 depend on benefits for all of their income, 41 per cent do not receive
benefits at all. The majority (71 per cent) are in General Needs lettings, and 72 per
cent are with private registered providers. Around one-third of tenants in this cluster
were previously homeless, but only 27 per cent were given reasonable preference.
The reasons for leaving the last settled home are varied within this group, but among
the most prominent are domestic violence (10 per cent), overcrowding (11 per cent),
being asked to leave by family or friends (13 per cent), or needing to move to
independent accommodation (13 per cent).

37
5. Employment and Ethnicity Among Social Housing
Tenants
The emergence of ethnicity and employment as key variables in the cluster analysis
suggests that there is potentially a statistically significant relationship between a social
tenant’s ethnicity and his or her economic status. This would be in keeping with prior
research that suggests that BME people face additional employment disadvantages
compared with White people. Accordingly, the next three sections of this chapter
explore the nature of the association between a tenant’s ethnicity and employment
status for: (i) single tenants; (ii) lone parents; and (iii) working-age couples.
Subsequently, the chapter tries to identify the factors which may be responsible for
explaining these patterns. In turn, the findings can inform future social housing
allocation policies, in particular: whether working-related conditions will impact
negatively on BME groups; and for which groups employment initiatives targeted at
social housing tenants would be most beneficial.
Single Tenants
Single tenants are by far the largest group of new social housing tenants, as they
account for over half of the total population contained in CORE for 2016/17. It is not
surprising, then, that single people are the most prevalent household type within each
ethnic group (Table 12, Column 3).
Table 12. Single tenants in CORE 2016/17.
The majority of single tenants (54.9 per cent) are men (Table 13). Across most ethnic
groups, the share of single tenants is not much larger than the share of female tenants.
The most notable exception is the Chinese/Other group, among which 70.2 per cent
Count
[1]
As a percentage of all tenants
in ethnic group, %
[3]
As a percentage of all
tenants in CORE, %
[2]
White
142,987
56.1
47.2
Black
11,073
54.1
3.7
Asian
5,561
43.8
1.8
Mixed
5,053
56.7
1.7
Chinese/Other
3,236
52.9
1.1
Total
167,910
56.9
56.9
Notes: Calculations are based on cases for which data on ethnicity are complete, i.e. were not refused
by the tenant.
Source: 2016/17 Continuous Recording of Social Housing Lettings and Sales data.

38
are males (n = 2,273) and 29.8 per cent are females (n = 963).
Table 13. Single tenants by ethnicity and sex, %.
Figure 5. Economic status of single tenants in CORE by ethnicity, %.
Notes: Calculations are based on cases for which data on sex, ethnicity, and economic status are
complete, i.e. were not refused by the tenant; n = 163,194.
Source: 2016/17 Continuous Recording of Social Housing Lettings and Sales data.
Like the cluster analysis, Figure 5 indicates a potential relationship between ethnicity
and economic activity. Notably, White single tenants have higher rates of economic
inactivity compared with single tenants of all Minority Ethnic groups. Additionally, BME
single tenants have slightly higher rates of employment than White single tenants.
These patterns hold when the subpopulation of single tenants is broken down by sex
(Appendix 1). Across each ethnic group, slightly higher proportions of single female
tenants are in employment (excluding the Asian group) or economically inactive
compared with single male tenants. Accordingly, a smaller proportion of single females
are registered as unemployed compared with single males in their ethnic group.
To ascertain whether the relationship between a single tenant’s ethnicity and his or
her employment status is statistically significant, or whether the patterns we observe
0% 20% 40% 60% 80% 100%
Chinese or other
Mixed
Asian
Black
White
Employed Unemployed/in training Economically inactive
White
Black
Asian
Mixed
Chinese/Other
Total
Male
54.1
60.0
58.2
53.5
70.2
54.9
Female
45.9
40.0
41.8
46.5
29.8
45.1
Total
100.0
100.0
100.0
100.0
100.0
100.0
Notes: Calculations are based on cases for which data on ethnicity and sex are complete, i.e. were
not refused by the tenant; n = 167,910.
Source: 2016/17 Continuous Recording of Social Housing Lettings and Sales data.

39
in the cluster analysis and Table 6 are meaningless and due merely to chance, we can
use Pearson’s chi-square test for independence (Pearson, 1900; Fisher, 1922). The
chi-square test is the most appropriate measure of the relationship between variables
when our variables are categorical, as in this instance.
4
Pearson’s chi-square, , is
given by:
󰇛  󰇜

Where i is ethnicity, j is economic status, n is the data or frequency observed in each
category, and µ is the frequency that we might expect to observe in each category due
to chance. For instance, if we have an overall employment rate of 50 per cent and a
sample of ten White tenants, then we would expect half of the tenants in our sample
– i.e., five – to be employed. However, if the observed frequency differs significantly
from five, then we can surmise that the observed frequency may not be due to chance
but may instead reveal that ethnicity and employment are associated in some way.
Note that a chi-square test cannot inform us of the direction of any identified relationship;
rather, it simply tells us that there is some form of a relationship between variables.
A chi-square test involves testing a null hypothesis, H0, and an alternative hypothesis,
H1. In this case, H0 states that among the population of single tenants in CORE,
ethnicity is not associated with economic status. Conversely, H1 states that a single
tenant’s ethnicity is associated with his or her economic status. Whether we reject or
fail to reject H0 is determined through comparing the test statistic, χ2, against a ‘critical
value’. The critical value is a product of: (i) the degrees of freedom, df, in the test; and
(2) the probability value, p, of the test. The number of degrees of freedom is
determined using the following formula:
df = (number of rows – 1) * (number of columns – 1)
In this case, we have five rows, corresponding to five ethnicity groups, and three
columns, since tenants can be employed, unemployed, or economically inactive.
Hence: df = (5 – 1) * (3 – 1) = 8.
4
A categorical variable is one that comprises categories of objects or entities. Hence, employment
status - classified as employed, unemployed, or economically inactive - is a categorical variable, since
individuals can fall into only one of these categories. Similarly, ethnicity is a categorical variable (Field,
2009).

40
If  is larger than the test statistic at 8 degrees of freedom, then we can reject H0 and
conclude there is a relationship between ethnicity and economic status. For 8 degrees
of freedom, the critical value at the 5 per cent significance level is 15.5073 and at the
1 per cent significance level is 20.0902.
Running the analysis produced a test statistic of 3889.206 with a p-value of 0.000.
Such a small p-value means that the probability of obtaining this test statistic value by
chance is extremely unlikely. In addition, none of the expected counts were below 5,
thereby ensuring the accuracy of the test statistic. Because  is larger than our critical
values and is highly significant (p = 0.00), we can reject H0. Hence, a single tenant’s
ethnicity is associated with his or her employment status.
5
The chi-square test tells us that a single tenant’s ethnicity is associated with his or her
economic status. Yet, it does not tell us about the nature of this association. For this
purpose, we require a multinomial logistic regression analysis. Multinomial or
polychotomous logistic regression allows us to predict to which of three or more
categories a person is likely to belong given information on another variable (Field,
2009). Hence, it can tell us the likelihood that a single tenant will be employed rather
than unemployed, or employed rather than economically inactive, based on his or her
ethnicity.
Table 14 presents the results of the multinomial regression analysis. The most important
column in Table 14 is Column 3, which gives the odds ratios. The odds ratios give the
odds of a given event (e.g., being unemployed rather employed) occurring in the
comparison group (e.g., Black single tenants) compared to the baseline group, i.e., White
single tenants. An odds ratio of 1 indicates that the odds of a particular outcome are
equal in both groups. An odds ratio greater than 1 indicates that the odds of the given
event occurring are greater for the comparison group than for White single tenants.
Conversely, an odds ratio of below 1 indicates that the odds of the event occurring are
lower for the comparison group than for White single tenants.
Table 14 tells us that:
• Black single tenants are more likely than White single tenants to be unemployed
or in training rather than employed. The odds ratio indicates that as ethnicity
changes from White to Black, the change in the odds of being unemployed
compared to being employed is 1.21. In other words, the odds of a Black single
5
The chi-square test was repeated separately for each single male tenants and single female tenants;
however, the results for males and females suggest the same conclusion, i.e., that ethnicity is
associated with employment status.

41
tenant being unemployed compared to being employed is 1.21 times greater
than the odds of a White single tenant being unemployed compared to being
employed. However, Black single tenants are also less likely than White single
tenants to be economically inactive rather than employed or registered as
unemployed. The odds ratio tells us that as ethnicity changes from White to
Black, the change in the odds of being employed compared to being
economically inactive is 2.43. Put another way, the odds of a Black single tenant
being economically inactive compared to being in employment are about two-
and-a-half those of a White single tenant. Similarly, Black single tenants are
almost three times as likely as White single tenants to be registered as
unemployed rather than economically inactive.
• Asian single tenants are also more likely than White single tenants to be employed
rather than unemployed. The odds of an Asian single tenant being unemployed
compared to being unemployed are 1.65 times greater more than for a White
single tenant. However, Asian single tenants are less likely than White single
tenants to be economically inactive rather than in work or looking for work.
Specifically, the odds of an Asian single tenant being employed rather than
economically inactive are 1.35 times greater than for a White single tenant; and
Asian single tenants are more than twice as likely as White single tenants to be
job seeking or in training rather than economically inactive.
• Single tenants of Mixed ethnic origins are more likely than White single tenants to
be unemployed rather than employed. The odds that a single tenant of Mixed
ethnic origins is unemployed rather than in work are 1.26 times greater more
than for a White single tenant. Yet, compared with the White group, single
tenants of Mixed ethnic origins are 1.68 times more likely to be employed rather
than economically inactive, and more than twice as likely to be jobseekers
rather than economically inactive.
• Chinese/Other single tenants are also more likely than White single tenants to be
unemployed rather than employed. The odds of a Chinese/Other single tenant
being unemployed compared to being employed are 1.66 times greater than for
a White single tenant. Still, Chinese/Other single tenants are almost 1.76 times
more likely than White single tenants to be employed rather than economically
inactive, and almost three times as likely as White single tenants to be looking
for employment rather than economically inactive.
In summary, Table 14 tells us that BME single tenants are more likely than their White
counterparts to be unemployed rather than in employment. Compared to the White
group, Asian and Chinese/Other single tenants have the highest odds of being
unemployed rather than employed. However, Table 6 also tells us that, compared with
White single tenants, BME single tenants are less likely to be economically inactive rather
than in work or looking for work. Black single tenants have the greatest odds of being

42
employed rather than economically inactive, while Black and Chinese/Other single
tenants have the greatest odds of being economically active but unemployed rather than
economically inactive.
Table 14. Results of the multinomial regression with White single tenants as
the baseline category.
So, while economically active BME single tenants have lower odds of being in
employment than White single tenants, BME single tenants are less likely to be
registered as economically inactive than White single tenants.
Lone Parents
After single people, the next most common household type within CORE comprises
lone parents, who account for 19.8 per cent of CORE households. Of these, 93.5 per
cent are lone mothers. The ethnic group with the greatest share of lone parents is
95% CI for Odds Ratio
B (SE)
[1]
Lower
[2]
Odds Ratio
[3]
Upper
[4]
Unemployed/in training rather than employed
Intercept
-0.12 (0.01)***
Black single tenants
0.18 (0.03)***
1.15
1.21
1.27
Asian single tenants
0.50 (0.04)***
1.52
1.65
1.78
Mixed single tenants
0.23 (0.04)***
1.16
1.26
1.36
Chinese/Other single tenants
0.51 (0.05)***
1.51
1.66
1.83
Employed rather than economically inactive
Intercept
-1.19 (0.01)***
Black single tenants
0.89 (0.03)***
2.32
2.43
2.55
Asian single tenants
0.30 (0.04)***
1.25
1.35
1.45
Mixed single tenants
0.52 (0.04)***
1.57
1.68
1.80
Chinese/Other single tenants
0.56 (0.05)***
1.61
1.76
1.93
Unemployed/in training rather than economically inactive
Intercept
-1.31 (0.01)***
Black single tenants
1.08 (0.02)***
2.80
2.94
3.08
Asian single tenants
0.80 (0.03)***
2.08
2.22
2.36
Mixed single tenants
0.75 (0.04)***
1.98
2.12
2.27
Chinese/Other single tenants
1.07 (0.04)***
2.70
2.93
3.17
Notes: R2 = 0.023 (Cox & Snell), 0.027 (Nagelkerke). Model χ2(8) = 3718.002 p < 0.001. * p < 0.05, **
p < 0.01, *** p < 0.001.
Sources: 2016/17 Continuous Recording of Social Housing Lettings and Sales data; own calculations.

43
Black (24.9 per cent), while the ethnic group with the smallest share of lone parents is
Chinese/other (14.0 per cent); however, 82.3 per cent of lone parents are White.
Among the population of lone parents contained in CORE, 13.2 per cent are in full-
time employment while 21.4 per cent are in part-time employment. Just over half are
economically inactive. As for single people, economic status among lone parents
seems to vary by ethnicity (Table 15). Black single parents have the highest
employment rates, while Asian and Chinese/Other lone parents have the lowest.
Rates of economic inactivity are highest for White, Asian, and Chinese/Other lone
parents, but lowest for lone parents with Black and Mixed ethnic origins. This is despite
narrow inter-ethnic differences among lone parents in terms of age (Figure 6) and
family characteristics (Table 16).
Table 15. Economic status of lone parents within CORE by ethnicity, %.
Figure 6. Age profile of lone parents in CORE by ethnicity, %.
Notes: Calculations are based on cases for which data on sex, ethnicity, and age are complete, i.e.
were not refused by the tenant; n = 56,887.
Source: 2016/17 Continuous Recording of Social Housing Lettings and Sales data.
0% 20% 40% 60% 80% 100%
Chinese…
Mixed
Asian
Black
White
24 years and under 25-39 years 40-59 years
White
Black
Asian
Mixed
Chinese/Other
Employed
34.7
47.5
24.6
39.3
29.7
Government training scheme
0.1
0.2
0.3
0.1
0.4
Unemployed
12.6
14.9
18.2
14.0
19.3
Not seeking work
43.1
28.6
49.5
37.9
39.0
Student
1.3
4.1
1.2
3.2
1.5
Sick or disabled
8.2
4.7
6.3
5.5
10.1
Total
100.0
100.0
100.0
100.0
100.0
Notes: Calculations are based on cases for which data on ethnicity, sex, and economic status are
complete, i.e. were not refused by the tenant; n = 55,368. Columns may not sum to 100.0% exactly
due to rounding.
Source: 2016/17 Continuous Recording of Social Housing Lettings and Sales data.

44
Table 16. Family characteristics of lone parents within CORE by ethnicity, %.
To identify whether there is a statistically significant relationship between a lone parent
tenant’s ethnicity and employment status, we can again use Pearson’s chi-square test.
H0 states that among the population of lone parents in new social lettings, ethnicity is
not associated with economic status. H1 states that a lone parent’s ethnicity is
associated with economic status. Running the chi-square test gave a test statistic of
515.349, p = 0.000. Because χ2 > 20.0902, the critical value for 8 degrees of freedom
at the 1 per cent level, we can reject H0. Hence, a lone parent’s ethnicity is associated
with employment status.
Again, a multinomial regression analysis can tell us the likelihood that a single parent
will be employed rather than unemployed, or employed rather than economically
inactive, based on his or her ethnicity. The results (Table 17) tell us that:
• Black lone parents are less likely than White lone parents to be unemployed or in
training rather than employed. The odds ratio indicates that as ethnicity changes
from White to Black, the change in the odds of being unemployed compared to
being employed is 0.87. In other words, the odds of a Black lone parent being
employed compared to being unemployed is 1/0.87 = 1.15 times greater than
the odds of a White lone parent. In addition, the odds of a Black lone parent
being employed rather than economically inactive are about double the odds
for a White single parent. Similarly, Black single parents are about one-and-a-
half times more likely than White lone parents to be registered as unemployed
rather than as economically inactive.
• Asian lone parents are about twice as likely as White single parents to be
unemployed rather than employed. Asian single parents are also less likely than
White single tenants to be employed rather than economically inactive. The
White
Black
Asian
Mixed
Chinese/Other
Number of
dependent
children
One
56.4
49.0
43.8
56.0
51.2
Two
28.5
29.7
32.8
28.6
31.4
Three or more
15.1
21.3
23.4
15.4
17.4
Total
100.0
100.0
100.0
100.0
100.0
Age of
first child
0-2 years
33.3
24.6
24.3
32.1
24.4
3-4 years
14.2
12.9
14.5
16.0
14.8
5-10 years
29.8
32.6
35.8
30.2
33.4
11-16 years
20.5
25.3
21.2
18.5
22.5
17 years or over
3.5
4.5
4.2
3.3
4.9
Total
100.0
100.0
100.0
100.0
100.0
Notes: Calculations are based on cases for which data on ethnicity and sex are complete, i.e. were
not refused by the tenant; n = 58,576. Columns may not sum to 100.0% exactly due to rounding.
Source: 2016/17 Continuous Recording of Social Housing Lettings and Sales data.

45
odds ratio indicates that the odds of an Asian single parent being economically
inactive rather than employed are 1/0.65 = 1.54 times those of a White single
parent. However, Asian single parents are 1.34 times more likely than White
single parents to be looking for work rather than economically inactive.
• Whether a lone parent is of Mixed or White ethnic origin does not significantly
predict his or her chances of being unemployed rather than employed, since p >
0.05 (Column 1). Still, the odds ratios indicate that Mixed lone parents are
somewhat more likely than White lone parents to be employed or registered as
unemployed rather than as economically inactive.
• Compared with White lone parents, Chinese/Other single parents are almost
twice as likely to be unemployed rather than employed and one-and-a-half times
as likely to be unemployed rather than economically inactive. Whether a lone
parent is Chinese/Other or White does not significantly predict his or her chances
of being employed rather than economically inactive, since p > 0.05.
Table 17. Results of the multinomial regression with White lone parent tenants
as the baseline category.
95% CI for Odds Ratio
B (SE)
[1]
Lower
[2]
Odds Ratio
[3]
Upper
[4]
Unemployed/in training rather than employed
Intercept
-1.01 (0.02)***
Black lone parents
-0.14 (0.05)**
0.80
0.87
0.95
Asian lone parents
0.72 (0.07)***
1.79
2.05
2.36
Mixed lone parents
-0.02 (0.07)
0.86
0.98
1.12
Chinese/Other lone parents
0.60 (0.11)***
1.49
1.83
2.25
Employed rather than economically inactive
Intercept
-0.46 (0.01)***
Black lone parents
0.59 (0.03)***
1.69
1.80
1.92
Asian lone parents
-0.43 (0.05)***
0.59
0.65
0.73
Mixed lone parents
0.22 (0.05)***
1.13
1.24
1.36
Chinese/Other lone parents
-0.15 (0.08)
0.73
0.86
1.01
Unemployed/in training rather than economically inactive
Intercept
-1.47 (0.02)***
Black lone parents
0.45 (0.05)***
1.43
1.57
1.72
Asian lone parents
0.29 (0.06)***
1.19
1.34
1.51
Mixed lone parents
0.20 (0.07)**
1.07
1.22
1.39
Chinese/Other lone parents
0.45 (0.10)***
1.30
1.57
1.89
Notes: R2 = 0.009 (Cox & Snell), 0.011 (Nagelkerke). Model χ2(8) = 513.734 p < 0.001. * p < 0.05, ** p
< 0.01, *** p < 0.001.
Sources: 2016/17 Continuous Recording of Social Housing Lettings and Sales data; own calculations.

46
In summary, Table 17 tells us that Black single parents are more likely than their White
counterparts to be in employment and be economically active. However, lone parents
from Asian and Chinese/Other BME backgrounds are more likely than White lone
parents to be unemployed rather than in paid work. At the same time, BME lone parents
are less likely than White single parents to be economically inactive. The exception is the
Asian group, who are more likely to be economically inactive; around half of Asian lone
parents are not seeking paid employment.
Two-Adult Households (Under 60)
Two-adult households in which both adults are under sixty years of age comprise 18.0
per cent of all households contained in CORE. Of these, more than two-thirds also
contain children. The majority of two-adult households are White (81.9 per cent).
6
As for households with one adult only, economic status among two-adult households
seems to vary by ethnicity (Table 18). Those with a Chinese/Other head appear least
likely to contain any earners, since 41.9 per cent of such households have no one in
employment. In addition, lower proportions of two-adult households with a BME head
comprise two full-time earners or one full-time and one part-time earner than two-adult
households with a White head; however, minority ethnic households are more likely
than White households to contain a part-time earner.
Table 18. Economic status of two-adult households with dependent children
within CORE by ethnicity, %.
White
Black
Asian
Mixed
Chinese/Other
Sole full-time earner
27.2
32.6
32.5
29.9
23.2
No earners
31.0
27.9
27.3
25.3
41.5
One-and-a-half full-time earners
13.1
9.9
5.0
10.8
4.6
Dual full-time earners
18.1
12.2
5.3
15.8
6.9
Sole part-time earner
8.6
15.3
27.0
15.2
21.8
Dual part-time earners
1.9
2.3
2.9
3.0
1.9
Total
100.0
100.0
100.0
100.0
100.0
Notes: Calculations are based on cases for which data on ethnicity, sex of person 1, age of person 1
and of person 2, and economic status of person 1 and person 2 are complete, i.e. were not refused
by the tenant; n = 50,627. Columns may not sum to 100.0% exactly due to rounding.
Source: 2016/17 Continuous Recording of Social Housing Lettings and Sales data.
6
Data on the ethnicity of the second adult are not available through CORE.

47
Using Pearson’s chi-square test, H0 states that among two-adult households, ethnicity
is not associated with economic status, while H1 states that ethnicity is associated with
economic status for such households. Running the analysis gave a test statistic of
2162.736, p = 0.000. The critical value at 20 degrees of freedom
7
is 37.57. Because
χ2 > 37.57, we can reject H0. Hence, ethnicity is associated with employment status
among two adult households.
Building on this, the results of a multinomial regression analysis (Table 19) tell us that:
• Two-adult households with a Black head of household are less likely than those
with a White head of household to have two full-time earners or one-and-a-half
full-time earners, i.e., one full-time earner and one part-time earner, rather than
no earners. The odds ratios indicate that the Black ethnic group is 1/0.75 = 1.33
times more likely than the White ethnic group to have no earners rather than
two earners, and 1/0.84 = 1.19 times more likely to have no earners rather than
one-and-a-half full-time earners. However, couples with a Black head of
household are 1.34 times more likely to have one full-time earner and 1.33 times
more likely to have one part-time earner, and almost twice as likely to have a part-
time earner rather than no earner at all.
• The odds of a two-adult household with an Asian head having two full-time
earners instead of no earners are one-third of those of a couple with a White head.
The Asian group is also less than half as likely to have one-and-half full-time
earners than no earners. Yet, two-adult households with an Asian head are 1.36
times as likely as those with a White head to have one full-time earner rather than
no earner. They are also 1.71 more likely to have two part-time earners and three-
and-a-half times more likely to have one part-time earner rather than no earners.
• Whether a head of household in a two-adult family is Mixed rather than White
does not significantly predict the likelihood of having two or one-and-a-half rather
than no earners, since p > 0.05. Still, the odds ratios indicate that the Mixed
group is 1.35 times more likely to have one full-time earner and around twice
as likely to have at least one part-time earner rather than no one in employment.
• The odds of a two-adult household with a Chinese/Other minority ethnic head of
household having two full-time earners or one-and-a-half full-time earners rather
than none are about one-quarter the odds for a two-adult household with a White
head. In addition, couple households with a Chinese/Other head are 1/0.64 =1.56
times more likely than their White counterparts to have no earners rather than one
full-time earner; nevertheless, they are almost twice as likely to have one part-
7
As per Table 10, the number of economic statuses is six, while the number of ethnic groups is five.
Therefore, the number of degrees of freedom is: (6-1) * (5-1) = 20.

48
time earner rather than no earners. However, whether the head of a couple
household is Chinese/Other or White does not significantly predict the odds that
the household contains two part-time workers rather than no workers.
Table 19. Results of the multinomial regression with two-adult households (aged
under 60) with a White head of household as the baseline category.
95% CI for Odds Ratio
B (SE)
Lower
Odds Ratio
Upper
Two full-time earners rather than no earners
Intercept
-0.54 (0.01)***
Black head of household
-0.29 (0.07)***
0.66
0.75
0.85
Asian head of household
-1.10 (0.08)***
0.29
0.33
0.39
Mixed head of household
0.07 (0.11)
0.89
1.07
1.29
Chinese/Other head of household
-1.26 (0.11)***
0.23
0.28
0.35
One-and-a-half earners rather than no earners
Intercept
-0.86 (0.02)***
Black head of household
-0.18 (0.07)*
0.73
0.84
0.96
Asian head of household
-0.85 (0.08)***
0.36
0.43
0.51
Mixed head of household
0.00 (0.11)
0.81
1.00
1.24
Chinese/Other head of household
-1.34 (0.13)***
0.20
0.26
0.34
One full-time earner rather than no earners
Intercept
-0.13 (0.01)***
Black head of household
0.29 (0.05)***
1.21
1.34
1.47
Asian head of household
0.31 (0.05)***
1.24
1.36
1.49
Mixed head of household
0.30 (0.08)***
1.15
1.35
1.58
Chinese/Other head of household
-0.45 (0.07)***
0.56
0.64
0.73
Two part-time earners rather than no earners
Intercept
-2.79 (0.04)***
Black head of household
0.28 (0.13)*
1.02
1.33
1.72
Asian head of household
0.54 (0.11)***
1.38
1.71
2.13
Mixed head of household
0.65 (0.19)***
1.33
1.91
2.75
Chinese/Other head of household
-0.28 (0.20)
0.52
0.76
1.12
One part-time earner rather than no earners
Intercept
-1.28 (0.02)***
Black head of household
0.68 (0.06)***
1.75
1.98
2.23
Asian head of household
1.27 (0.05)***
3.23
3.57
3.93
Mixed head of household
0.77 (0.10)***
1.79
2.16
2.62
Chinese/Other head of household
0.64 (0.07)***
1.64
1.89
2.18
Notes: R2 = 0.040 (Cox & Snell), 0.041 (Nagelkerke). Model χ2(20) = 2052.906 p < 0.001. * p < 0.05, **
p < 0.01, *** p < 0.001.
Sources: 2016/17 Continuous Recording of Social Housing Lettings and Sales data; own calculations.

49
Overall, Table 19 tells us that, excluding the Mixed group, two-adult households with a
BME head are less likely than those with a White head to have two-full time earners or
one-and-a-half full-time earners than no earners at all. At the same time, couple
households with a BME head (excluding the Chinese/Other group) are more likely than
those with a White head to contain one full-time earner or two part-time earners rather
than no earners at all. The single breadwinner model is particularly prevalent among the
Mixed: Other group (36.4 per cent), Pakistani group (35.8 per cent), African group (35.2
per cent), and Indian group (34.9 per cent). In addition, two-adult households with a
minority ethnic head are more likely than the White group to have one part-time earner
rather than no earners. This model is common among the Bangladeshi group (35.0 per
cent).
So, while the dual full-time earner and one-and-a-half full-time earner models of
employment are more common among two-adult households with a White head of
household, the sole-earner and part-time models of employment are more prevalent
among two-adult households with a BME head of household.
Understanding These Patterns
On the whole, BME tenants in new social lettings are more likely to be economically
active than White tenants. However, with certain exceptions, they are also more likely to
be unemployed rather than in employment. Based on the findings of the literature review
(Chapter 2), as well as descriptive statistics generated from the 2016/17 CORE dataset,
it is possible to identify certain household characteristics which might contribute to
explaining why employment status varies by ethnicity among social housing tenants.
These characteristics, which are explored in turn in the following, are: (1) location; (2)
nationality; and (3) age and health.
Location
It is plausible that inter-ethnic variations in the odds of being employed are at least partly
a function of differences in the health of the local labour markets. In other words, BME
tenants may be concentrated in areas with fewer available labour market opportunities,
which in turn underpins the lower odds of employment among economically active BME
tenants.
We can use Pearson’s correlation coefficient to test whether there is an association
between a tenant’s location and employment status. Pearson’s correlation coefficient is
a summary statistic that captures the association or strength of relationship between the

50
values of two continuous variables.
8
Pearson’s correlation coefficient can take any value
between -1 and +1. A value of zero indicates that there is no linear relationship between
the two variables. Meanwhile, a value of -1 indicates a perfect negative (linear)
relationship; i.e., as one variable increases in value, the other variable decreases in
value. A value of +1 suggests that there is a perfect positive (linear) relationship; that is,
as the value of one variable goes up, the value of the other variable goes down.
9
However, in the social sciences, correlation coefficients rarely approach -1 or +1, since
we are using data that were not obtained under perfect, laboratory-type conditions.
Hence a lot of noise or random error inevitably attenuates the correlations. Accordingly,
we can consider a correlation coefficient greater than ±0.5 as signifying a fairly strong
association, and a value greater than ±0.3 as indicating a moderate correlation (Fiske,
2010). The formula for Pearson’s correlation coefficient, , is given by:
  󰇟󰇛 
󰇜󰇛
󰇜󰇠
󰇟󰇛 
󰇜󰇛
󰇜󰇠
The two variables included in the correlation analysis are: (1) share of tenants from BME
backgrounds within a local authority area; and (2) unemployment rates across 324 local
authorities in England.
10
Running the correlation analysis gave a value of 0.44 significant
at the p = 0.00 level. The value of the correlation coefficient indicates a moderate, positive
(linear) relationship between the share of BME social housing tenants within a local
authority and the local authority’s overall unemployment rate.
Table 20 contrasts the share of BME tenants in local authorities with the highest and
lowest unemployment rates and levels of deprivation. Notably, five of the ten local
authorities with the highest unemployment rates (Column 2) have above-average shares
of BME households in new social lettings (Column 1). These are Tower Hamlets,
Birmingham, Nottingham, Sandwell, and Newcastle upon Tyne. These local authorities,
with the exception of Newcastle upon Tyne, are also within the top 4 per cent most
deprived local authorities (Column 3). In contrast, across the ten local authorities with the
lowest unemployment rates, around 4 per cent of all households within CORE are BME.
Five of these local authorities – Uttlesford, South Cambridgeshire, West Oxfordshire,
Waverley, and Hart - are in the 10 per cent of least deprived local authorities.
8
A continuous variable is one that can be measured to any level of precision. Thus, the percentage of
people registered as unemployed is a continuous variable, since it can take on any value between 0
and 100 and can, in principle, be measured at an infinite level of decimal places and precision (Field,
2009).
9
Note that neither a value of -1 nor +1 indicates why the two variables are associated.
10
Unemployment data were unavailable for the Isles of Scilly or City of London.

51
Table 20. Share of single tenants (new lettings and re-lets) from minority ethnic
backgrounds within local authorities with highest and lowest unemployment
rates, 2016/17.
Local Authority
All households in
CORE within LA
who are BME, %
[1]
LA unemployment
rate, %
[2]
Ranking by
Multiple
Deprivation Index1
[3]
Ten Local Authorities with Highest Unemployment Rates
Hartlepool
2.2
10.1
32
Tower Hamlets
69.4
9.1
6
Middlesbrough
7.4
8.9
16
South Tyneside
2.7
8.9
31
Birmingham
48.3
8.5
11
Sunderland
2.5
8.4
38
Sandwell
29.8
8.1
12
Redcar and Cleveland
2.4
7.9
78
Newcastle upon Tyne
18.1
7.8
92
Nottingham
28.4
7.7
10
Average
21.1
8.5
33
Ten Local Authorities with Lowest Unemployment Rates
Uttlesford
5.5
2.1
297
South Cambridgeshire
6.9
2.2
314
Eden
1.9
2.2
182
South Lakeland
2.6
2.2
251
Ribble Valley
1.4
2.3
290
West Oxfordshire
4.2
2.3
316
Waverley
5.0
2.3
323
Derbyshire Dales
3.7
2.4
258
North Dorset
1.2
2.4
210
Hart
6.3
2.4
326
Average
3.9
2.3
277
Average for all LAs
10.6
4.4
N/A
Notes: 1The Multiple Deprivation Index (IMD) is a relative measure based on seven different domains
of deprivation. The IMD covers Lower-layer Super Output Areas; hence, the rank provided in the
table is a higher-level geography summary measure. A rank of 1 indicates that the LA has the highest
average level of deprivation across all its smaller areas; a rank of 326 indicates that the LA has the
lowest level of deprivation across all its smaller areas. In other words, a lower ranking indicates higher
levels of deprivation.
Sources: 2016/17 Continuous Recording of Social Housing Lettings and Sales data; 2016/17 Local
labour market indicators by unitary and local authority July 2017, available from the Office for National
Statistics; English Indices of Deprivation 2015, available from the Office for National Statistics.

52
The reasons why BME tenants are clustered in areas with higher unemployment and
deprivation are undoubtedly numerous and complex. Nevertheless, this observation
suggests that more perhaps should be done to ensure people from BME backgrounds
are allocated social housing in less deprived areas and in areas in which there are
more labour market opportunities. There is a danger that placing economically inactive
BME new social tenants in high unemployment areas will only compound their
marginalisation. Of course, such a policy initiative is limited by the concentration of
social housing properties in deprived areas. Hence, tackling the wider problem of
unemployment in the most deprived areas is arguably equally important for improving
BME social tenants’ employment odds.
Nationality
BME tenants are disproportionately likely to be non-UK nationals, especially those in
the Chinese/Other group (Table 21). Hence, an additional factor which might help to
explain the different employment profiles of White and BME tenants is nationality. This
is because non-UK nationals
11
face more stringent conditions in accessing social
housing and are typically required to be a registered worker or jobseeker unless they
have passed a ‘habitual residence’ test.
Table 21. Table 13. Nationality of head of household among CORE households
by ethnicity, %.
Thus, as Table 22 shows, a far larger proportion of non-UK national single tenants
compared with UK national single tenants are either in employment or registered
jobseekers. Tenants from the European Economic Area, who can work in the UK
without restriction, are most likely to be employed. Non-UK tenants from outside of
Europe, who face greater restrictions in accessing employment, are most likely to be
looking for employment. Accordingly, we can surmise that because BME tenants are
11
This excludes refugees; yet, refugees are a minority, comprising 0.7 per cent of CORE households.
White
Black
Asian
Mixed
Chinese/Other
UK national
95.6
70.2
75.3
82.2
45.9
European Economic Area national
3.8
6.1
4.5
10.7
12.6
Any other nationality
0.6
23.7
20.2
7.0
41.5
Total
100.0
100.0
100.0
100.0
100.0
Notes: Calculations are based on cases for which data on ethnicity and nationality of head of household
are complete, i.e., were not refused by the tenant; n = 298,701. Columns may not sum to 100.0%
exactly due to rounding.
Source: 2016/17 Continuous Recording of Social Housing Lettings and Sales data.

53
more likely to be foreign nationals than White tenants, this partly explains higher rates
of economic activity among BME tenants.
Table 22. Employment status of heads of households within CORE by
nationality, %.
UK national
European Economic Area
national
Any other
nationality
Employed
28.1
67.6
31.2
Unemployed/in training
14.9
7.9
30.3
Economically inactive
57.0
24.6
38.5
Total
100.0
100.0
100.0
Notes: Calculations are based on cases for which data on sex, employment status, and nationality are
complete, i.e. were not refused by the tenant; n = 162,201.
Source: 2016/17 Continuous Recording of Social Housing Lettings and Sales data.
Age and Health
Another potentially important factor conditioning the relationship between ethnicity and
employment status is the older age profile of White households within CORE compared
with BME households. Figure 7 shows that while 27.5 per cent of White heads of
household within CORE are aged sixty years and over, between 7.2 per cent (Mixed)
and 9.2 per cent (Chinese/Other) of BME heads of household fall into this age bracket.
Figure 7. Age profile of heads of household by ethnicity, %.
Notes: Calculations are based on cases for which data on ethnicity and age of household head are
complete, i.e., were not refused by the tenant; n = 301,785.
Source: 2016/17 Continuous Recording of Social Housing Lettings and Sales data.
Accordingly, it is plausible that a greater proportion of White tenants are economically
inactive due to sickness or disability or retirement because of their older age profile. Table
0% 20% 40% 60% 80% 100%
Chinese/Other
Mixed
Asian
Black
White
24 years and under 25-39 years 40-59 years 60 years and over

54
23 suggests that this is the case, as around half of White heads of household are
sick/disabled or retired.
Table 23. Heads of household registered as sick/disabled or retired as a
percentage of ethnic group, %.
Additional Factors
Of course, factors other than location, nationality, age, and health and which are
beyond the scope of the CORE data contribute to explaining BME tenants’ lower odds
of employment. For example, labour market discrimination, ethnic group differences
in such socio-demographic indicators as educational levels or time of entry into the UK
for those born overseas, the effects of ethnic group preferences, traditions, and
‘norms’, and employer discrimination (see e.g., Catney and Sabater, 2015).
Additional potential factors that explain inter-ethnic variations in rates of employment and
economic activity within the population of CORE are propensity and ability to make use
of childcare (e.g., Datta et al., 2007; Huskinson et al., 2016). Research finds that BME
parents are less likely to use childcare than White families, partly because of less
awareness of such facilities, and partly because of costs, inadequate supply of local
childcare, a deficit of BME staff in childcare, and language barriers (e.g., Box et al., 2001;
Daycare Trust, 2004). An additional barrier to employment for BME households may be
a mismatch between work schedules and the operational hours of childcare facilities. For
instance, a study conducted by Barnardo’s found that long and atypical working hours
were a factor in the childcare choices of the Chinese community in Northern Ireland
(Webb et al., 2014). Similarly, a report by the Daycare Trust (2004) mentions the
difficulties faced by some Muslim parents whose children go to the mosque after
school. Indeed, the prevalence of sole-earner and part-time models of employment
among two-adult households with a BME head of household may reflect the use of
alternate shift patterns between parents so that children can be cared for by one while
the other works (Datta et al., 2007). Furthermore, BME parents may be less likely to
make use of informal childcare networks, since migration to the UK may split up childcare
support networks, making it harder for BME parents to rely on their families for childcare
White
Black
Asian
Mixed
Chinese/Other
Registered as sick/disabled
20.3
11.5
14.1
14.8
13.3
Retired
15.7
5.0
6.2
3.7
6.4
Total
36.0
16.5
20.3
18.5
19.7
Notes: Calculations are based on cases for which data on ethnicity and economic status of head of
household are complete, i.e., were not refused by the tenant; n = 288,418. Columns may not sum to
100.0% exactly due to rounding.
Source: 2016/17 Continuous Recording of Social Housing Lettings and Sales data.

55
provision. Indeed, just over half of Chinese/Other lone parents in CORE 2016/17 are
from outside the UK.
Yet, the multinomial regression analyses revealed that not all BME households are at
greater risk of unemployment than their White counterparts. Notably, Black lone parents
have higher odds of being in employment than White lone parents. In explaining this,
cultural factors are likely important. As Peach (1998) highlights, there is a strongly
developed tradition of female independence within the Afro-Caribbean and Caribbean
populations in Britain. In addition, Black lone mothers are more likely to consider
employment as an important form of provision and way of enacting care for their children
(Box et al., 2001). In addition, unlike other BME groups, Black Caribbean families exhibit
similar levels of childcare usage to White families, with 74 per cent of Black Caribbean
children in childcare compared with 73 per cent of White British children (Huskinson et
al., 2016). Even so, Black lone parents’ higher employment odds do not necessarily
translate into higher incomes or welfare. There is evidence that despite their higher
employment rates, Black Caribbean women are underrepresented in senior level jobs
(e.g., Buckner et al., 2007).
Still, on the whole, economically active BME households in new social lettings last
year were less likely to be in employment than their White counterparts. This
unemployment rate gap reflects the ‘ethnic penalty’ in employment observed in the UK
population at large; in the period from July to September 2017, the unemployment rate
among BME groups was 7.8 per cent compared with 4.0 per cent for the White ethnic
group (Office for National Statistics, 2017). BME tenants’ lower odds of employment
suggest that the shift towards prioritising working households in social housing
allocation across many councils could shut many BME people out of the social rented
sector.

56
6. Conclusion
This report has sought to explore the composition of households in new social lettings
across England. This reflects the background of a growing BME population in the UK,
a dwindling supply of social housing, and research evidence suggesting that BME
households are disadvantaged in accessing social housing, despite their low average
incomes. Accordingly, identifying who is taking up new social lettings and how they
get there, as well as how these factors potentially vary across ethnic groups, can
inform the direction of future policies towards helping BME groups to access social
housing on equal terms. To this aim, the report made use of the COntinuous
REcording of Lettings and Sales in Social Housing in England (CORE) dataset to
reveal the structure of the population in new social housing lettings through a cluster
analysis and other statistical and descriptive analyses. This chapter summarises the
key findings of the report, their policy implications, and suggestions for future research.
Key Findings
The analysis confirmed the uneven geographical spread of new social lettings to BME
households. According to the 2016-17 CORE data, Black and most Mixed, Chinese
and ‘Other’ BME groups are overrepresented in new social lettings. The
overrepresentation of some BME groups in social housing is at least partly explained
by the geographical concentration of BME communities in regions in which social
renting is more common, namely London and the West Midlands. What is more, BME
social tenants within these regions tend to cluster into a small number of local authorities
only (e.g., Birmingham, Hackney, Leeds, Manchester). Yet, even though Asian ethnic
groups also cluster into local authorities with high levels of social housing compared with
the national average, Asian households remain underrepresented in new social lettings.
Existing literature suggests that this is potentially because of cultural perceptions of
social housing as a less desirable tenure, especially compared with owner occupation,
as well as the greater prevalence of properties that can accommodate multi-family
households in the private rented sector. Indeed, while rates of owner occupation
among Asian groups have declined over time, rates of private renting have increased
significantly since the early-1990s.
The cluster analysis revealed eighteen groups or ‘types’ of typical households in the
CORE 2016/17 data. Ethnicity emerged as an important dimension of the cluster
analysis, in that organising households by ethnicity, employment status, and
household type revealed a meaningful structure to the data, i.e., maximised between-
group differences while minimising within-group differences. This suggests that BME
tenants in new social lettings may have unique needs compared with the ethnic

57
majority. More specifically, through dissecting and comparing the characteristics of the
clusters further, the report highlighted that:
• While BME households in new social lettings are, on the whole, more likely to
be economically active than White households, they are also less likely to be in
employment rather than unemployed. Higher rates of economic inactivity
among White households in new social lettings are partly explained by the older
age structure of the White group and corresponding higher rates of
sickness/disability and retirement. BME households’ lower odds of employment
are potentially linked to the geographical concentration of BME households in
areas with higher unemployment rates, as well as more stringent employment-
related conditions on access to social housing for migrants, who are
disproportionately BME. Prior research also suggests that BME households
with children are, on average, less likely to make use of formal childcare
services than White households with children.
• BME groups in new social lettings experience higher rates of prior
homelessness. The problem is especially acute among BME single males, of
which around half were previously homeless. This is in keeping with evidence
that BME groups are disproportionately at risk of homelessness.
• Rates of lone parenthood are high among the Black and Mixed ethnic groups
within new social lettings, which may contribute to explaining their
overrepresentation in new social lettings.
• BME tenants are less likely to have acquired their current social tenancy
through the choice-based letting system. This fits with evidence suggesting that
migrants, who are overwhelmingly from BME backgrounds, may be less likely
to have knowledge of their social rights or be able to access them due to
language barriers. The lower use of choice-based lettings among BME
households is also potentially associated with higher rates of homelessness
and temporary accommodation among BME groups.
• BME groups are slightly more likely to have been given reasonable preference.
• Single female tenants with or without children from BME backgrounds are more
likely to have left their previous settled home because of domestic violence than
White female tenants. The ethnic group in new social lettings that is most likely
to have previously experienced domestic violence is the South Asian group.
Policy Implications
Geographical Dispersion of Lettings to BME Groups. Research suggests a waning
preference among younger BME groups for living in ethnic enclaves, with the quality
of housing, schools, and neighbourhood taking greater priority. At the same time, focus
group research suggests that fears over racial harassment continue to dictate many

58
BME groups’ housing choices, with many actively choosing to avoid certain areas
perceived as racist (Markkanen, 2008). Hence, the concentration of new lettings to
BME groups in areas with high BME populations may be less a consequence of real
choice, and more out of a lack of adequate properties in different areas or concerns
about harassment. Accordingly, policies to support BME households that wish to enter
into more ethnically diverse areas to do so alongside policies that address racial
harassment in certain areas may contribute to enabling more BME households to take
up social lettings in a wider diversity of areas. In turn, this can contribute to lessening
ethnic residential segregation.
Employment. Close attention must be paid to the impact of the current policy trend
towards strengthening the relationship between priority for social housing and
employment status. Given higher unemployment among BME communities, such
a policy change has the potential to impede BME households’ access to social
housing. In turn, any move towards requiring current tenants to find work as a condition
of keeping their tenancy could impact disproportionately negatively on BME tenants.
The higher prevalence of unemployment among certain BME groups within CORE
also suggests that policies that use housing providers as an access point for reducing
unemployment among social housing tenants could be fruitful in reducing ethnic
inequalities in labour market participation. For example, the Kush Housing
Association’s Akaba project in Hackney, which helps vulnerable African Caribbean
people with mental health problems, helped over fifty participants gain employment
and job placements over two years (Shelter, 2008). Similarly, childcare initiatives could
help more social tenants to enter into employment. Integrating employment and
housing policies may be especially beneficial for migrant and refugee households in
new social lettings, since they are less likely to have the social connections for finding
work, knowledge of the support available to unemployed people, or the language skills
to access such support.
Homelessness. It is well known that BME households are more likely to become
statutorily homeless than White households, and the CORE data revealed that a
higher proportion of BME households in new social lettings were previously homeless
compared with White households. Therefore, preventative measures aimed at
reducing homelessness among BME populations by addressing their specific needs
(e.g., ‘hidden’ homelessness and overcrowding) are important for managing future
demand on the social housing sector and reducing the share of the population with
unmet housing needs. This also requires more research on the causes of BME
homelessness, which remains inadequate.
Choice-Based Lettings. BME households in new social lettings are less likely to have
acquired their current tenancy through the choice-based lettings system. This is

59
potentially due to higher rates of homelessness and temporary accommodation among
the BME community, which can inhibit participation in choice-based lettings because
of the pressing nature of the housing need or disadvantaged access to the means of
bidding for properties. Therefore, reducing ethnic inequalities in access to and the
quality of social housing arguably requires action to ensure that households which are
homeless or in temporary accommodation do not face undue barriers in accessing
choice-based lettings.
Domestic violence. Social housing has an important role to play in the autonomy and
survival of women who are experiencing domestic abuse. While domestic violence
affects all ethnic groups, the CORE data suggest that social housing is especially
important for certain South Asian women, as one-quarter of South Asian women in
new social lettings last year were forced out of their previous home because of
domestic abuse. BME victims of domestic violence can face additional disadvantages
in accessing social housing if English is not their first language, their immigration
status presents a barrier to accessing welfare services, or services are not adequately
responsive to BME women’s needs and circumstances (e.g., Burman and Chantler,
2005). Therefore, ensuring that women from South Asian and other Minority Ethnic
groups can access social housing on equal terms as White women, such as through
services that are sensitive to language and culture or via exceptions to immigration
stipulations, should be a priority.
Suggestions for Future Research
Quality of Housing and Choice. The higher rates of homelessness and concentration
in temporary accommodation among BME groups, coupled with the lower usage of
the choice-based lettings system, suggest that BME groups may be less well-placed
than White households to hold out for the best properties. However, this cannot be
tested using the CORE dataset in its current form. Hence, future research could benefit
from the inclusion of variables which capture a property’s quality (e.g., dampness), as
well as the quality of the local area (e.g., pollution levels). Additional potential ways of
operationalising choice within CORE could involve variables that capture length of time
on the waiting list for social housing or the number of properties refused after being
invited to view that property. These variables could be useful for assessing whether
BME households are potentially pushed into taking up properties quickly. Alternatively,
the English Housing Survey provides data on housing quality.
Comparing Tenures. Fully understanding the relationship between ethnicity and
access to social housing requires comparing BME households’ representation within
and mobility between other tenures. This could involve predicting the probability of a

60
household entering into the social rented sector, as opposed to the private rented
sector or owner occupation, by ethnicity while controlling for other characteristics.
Focus Groups and Qualitative Research. To really disentangle the relationship
between ethnicity and social housing access and quality, qualitative research that
engages end-users and gives voice to vulnerable BME households is imperative.
Previous literature suggests that culture and agency are important for understanding
the relationship between ethnicity and housing. Yet, statistical analyses of quantitative
data are not well-suited to analysing the role of culture and choice in conditioning BME
households’ housing outcomes, as well as how much agency these households feel
they are able to exercise. Rather, data drawn from the subjective experiences of
individuals are necessary.

61
References
Ballard, R. (1994) Introduction: The Emergence of Desh Pardesh IN Ballard, R. (ed.)
Desh Pardesh: The South Asian Presence in Britain. London: C. Hurst and
Co, 1-34.
Barton, C. (2017) Home Ownership and Renting: Demographics. Briefing Paper No.
CBP 7706. London: House of Commons Library.
Batsleer, J., Burman, E., Chantler, K., McIntosh, S.H., Pantling, K., Smailes, S. and
Warner, S. (2002) Domestic Violence and Minoritisation: Supporting Women
to Independence. Manchester: Women’s Studies Research Centre,
Manchester Metropolitan University.
Beider, H. (2012) Race, Housing & Community: Perspectives on Policy and Practice.
Chichester: John Wiley and Sons.
Berthoud, R. (2005) Family Formation in Multi-Cultural Britain: Three Patterns of
Disadvantage. Cambridge: Cambridge University Press.
Bevan, C. and Cowan, D. (2016) Uses of Macro Social Theory: A Social Housing
Case Study. Modern Law Review, 79 (1), 76-101.
Bowes, A.M., Dar, N.S. and Sim, D.F. (2002) Differentiation in Housing Careers: The
Case of Pakistanis in the UK. Housing Studies, 17 (3), 381-99.
Box, L., Bignall, T. and Butt, J. (2001) Supporting Parents through Provision of
Childcare. Black and Minority Ethnic Families Policy Forum Discussion Paper
No. 2. London: Race Equality Unit.
Buckner, L., Yeandle, S. and Botcherby, S. (2007) Moving On Up? Ethnic Minority
Women and Work. Ethnic Minority Women and Local Labour Markets.
London: Equal Opportunities Commission.
Burman, E. (2003) From Difference to Intersectionality: Challenges and Resources.
European Journal of Psychotherapy and Counselling, 6 (4), 293-308.
Burman, E. and Chantler, K. (2005) Domestic Violence and Minoritisation: Legal and
Policy Barriers Facing Minoritized Women Leaving Violent Relationships.
International Journal of Law and Psychiatry, 28 (1), 59-74.
Burman, E, Smailes, S and Chantler, K (2004) Culture as a Barrier to Service
Provision and Delivery: Domestic Violence Services for Minoritised Women.
Critical Social Policy, 24 (3), 332-57.
Cambridge Centre for Housing and Planning Research (2006) The Impact of a Pan-
London Choice-Based Lettings Scheme on Homelessness and Temporary
Accommodation. London: Association of London Government.
Cangiano, A. (2008) Mapping of Race and Poverty in Birmingham. Oxford: ESRC
Centre on Migration, Policy and Society (COMPAS).

62
Catney, G. and Sabater, A. (2015) Ethnic Minority Disadvantage in the Labour
Market Participation, Skills and Geographical Inequalities. York: Joseph
Rowntree Foundation.
Chahal, K. and Julienne, L. (1999) We Can't All Be White!: Racist Victimisation in the
UK. York: Joseph Rowntree Foundation.
Cheesman, D. (2007) The Margins of Public Space: Muslims and Social Housing in
England. People, Place and Policy, 1 (1), 39-46.
Clark, K. and Drinkwater, S. (2007) Ethnic Minorities in the Labour Market. Bristol:
Policy Press.
Cowan, D. and Marsh, A. (2004) From Need to Choice: R(a) v. Lambeth LBC;
R(Lindsay) v. Lambeth LBC. Modern Law Review, 67 (3), 478-490.
Crisis (2015) The Homelessness Legislation: An Independent Review of the Legal
Duties Owed to Homeless People. London: Crisis.
Crofts, S. (2017) International Migration and the Changing Nature of Housing in
England - What Does the Available Evidence Show? Newport: Office for
National Statistics.
Dahya, B. (1974) The Nature of Pakistani Ethnicity in Industrial Cities in Britain IN
Cohen, A. (ed.) Urban Ethnicity. London: Tavistock, 77-118.
Datta, K., McIlwaine, C., Evans, Y., Herbert, J., May, J. and Wills, J. (2007) From
Coping Strategies to Tactics: London's Low-Pay Economy and Migrant
Labour. British Journal of Industrial Relations, 45 (2), 404–32.
Daycare Trust (2004) Parents Eye: Building a Vision of Equality and Inclusion in
Childcare Services. London: Daycare Trust.
Darlington-Pollock, F. and Norman, P. (2017) Examining Ethnic Inequalities in Health
and Tenure in England: A Repeated Cross-Sectional Analysis. Health and
Place, 46 (2017), 82-90.
Department for Communities and Local Government (2015) Household Projections
2014 to 2039. London: Department for Communities and Local Government.
Department for Communities and Local Government (2017) English Housing Survey:
Social Rented Sector, 2015-16. London: Department for Communities and
Local Government.
Douglas, D. (2014) Ethnicity and New Social Housing Lets, Inside Housing, Available
from: https://www.insidehousing.co.uk/home/home/ethnicity-and-new-social-
housing-lets-40289 [Accessed 18 October 2017].
Dunnell, K. (2007) The Changing Demographic Picture of the UK: National
Statistician’s Annual Article on the Population. Population Trends, 130
(2007): 9-21.
Field, A. (2009) Discovering Statistics Using Ibm Spss Statistics. London: Sage.
Finney, N. and Harries, B. (2013) How has the Rise in Private Renting
Disproportionately Affected Some Ethnic Groups?: Ethnic Differences in

63
Housing Tenure 1991-2001-2011. Manchester: ESRC Centre on Dynamics
of Ethnicity.
Finney, N., Lymperopoulou, K., Kapoor, N., Marshall, A., Sabater, A. and Simpson,
L. (2014) Local Ethnic Inequalities: Ethnic Differences in Education,
Employment, Health and Housing in Districts of England and Wales,
2001−2011. London: Runnymede.
Fisher, R.A. (1922) On the Interpretation of Chi Square from Contingency Tables,
and the Calculation of P. Journal of the Royal Statistical Society, 85 (1922),
87-94.
Fiske, S.T. (2010) Social Beings: Core Motives in Social Psychology. Hoboken, New
Jersey: John Wiley and Sons.
Fitzpatrick, S., Pawson, H., Bramley, G., Wilcox, S. and Watts, B. (2013) The
Homelessness Monitor: England 2013. London: Crisis.
Fitzpatrick, S. and Stephenson, M. (1999) Homelessness, Need and Desert in the
Allocation of Council Housing. Housing Studies, 14 (4), 413-431.
Fitzpatrick, S. and Watts, B. (2017) Competing Visions: Security of Tenure and the
Welfarisation of English Social Housing. Housing Studies, 32 (8), 1021-1038.
Gervais, M.C. and Rehman, H. (2005) Causes of Homelessness Amongst Ethnic
Minority Populations. London: Office of the Deputy Prime Minister.
Habeebullah, M. and Slater, D. (1990) Equal Access to Council Housing in
Rochdale. London: Community Development Foundation Publications.
Harne, L. and Radford, J. (2008) Tackling Domestic Violence: Theories, Policies and
Practice. Maidenhead: Open University Press.
Harrison, M. and Phillips, D. (2003) Housing and Black and BME Communities:
Review of the Evidence Base. London: Office of the Deputy Prime Minister.
Harrison, M., Phillips, D., Chahal, K., Hunt, L. and Perry, J. (2005) Housing, ‘Race’
and Community Cohesion. Coventry: Chartered Institute of Housing.
Hemmerman, L., Law, I., Simms, J. and Sirriyeh, A. (2007) Situating Racist Hostility
and Understanding the Impact of Racist Victimisation in Leeds. Leeds:
Centre for Ethnicity and Racism Studies.
Henderson, J. and Karn, V. (1984) Race, Class and the Allocation of Public Housing
in Britain. Urban Studies, 21 (2), 115-128.
Huskinson, T., Hobden, S., Oliver, D., Keyes, J., Littlewood, M., Pye, J. and Tipping
S. (2016) Childcare and Early Years Survey of Parents 2014 to 2015.
London: Department for Education.
Jeffers, S. and Hoggett, P. (1995) Like Counting Deckchairs on the Titanic: A Study
of Institutional Racism and Housing Allocations in Haringey and Lambeth.
Housing Studies, 10 (3), 325-344.
Manley, D. and van Ham, M. (2011) Choice-Based Letting, Ethnicity and
Segregation in England. Urban Studies, 48 (14), 3125–3143.

64
Markkanen, S. (2008) BME Housing Needs and Aspirations. Cambridge: Cambridge
Centre for Housing and Planning Research.
Markkanen, S. (2009) Looking to the Future: Changing Black and BME Housing
Needs and Aspirations. Better Housing Briefing Paper No. 11, London: Race
Equality Foundation.
Markkanen, S. and Harrison, M. (2013) ‘Race’, Deprivation and the Research
Agenda: Revisiting Housing, Ethnicity and Neighbourhoods. Housing
Studies, 28 (3), 409-428.
Mullins, D. (2010) Housing Associations. Third Sector Research Centre Working
Paper No. 16.
Murad, F. and Saeed, T. (2004) Understanding the Housing Needs and Aspirations
of the Muslim Communities: A Case Study of the Royal Borough of
Kensington and Chelsea. London: North London Muslim Housing
Association.
Office for National Statistics (2012) Ethnicity and National Identity in England and
Wales: 2011. Available from:
https://www.ons.gov.uk/peoplepopulationandcommunity/culturalidentity/ethni
city/articles/ethnicityandnationalidentityinenglandandwales/2012-12-11
[Accessed 20 March 2018].
Office for National Statistics (2017) Labour Market Status by Ethnicity. Available
from:
https://www.ons.gov.uk/employmentandlabourmarket/peopleinwork/employm
entandemployeetypes/datasets/labourmarketstatusbyethnicgroupa09
[Accessed 19 February 2018].
Pawson, H., Jones, C., Donohoe, T., Netto, G., Fancy, C., Clegg, S. and Thomas, A.
(2006) Monitoring the Longer Term Impact of Choice Based Lettings.
London: Department for Communities and Local Government.
Peach, C. (1998) South Asian and Caribbean Ethnic Minority Housing Choice in
Britain. Urban Studies, 35 (10), 1657-1680.
Peach, C. (1999) London and New York: Contrasts in British and American Models
of Segregation. International Journal of Population Geography, 5 (1999),
319-351.
Peach, C. (2006) Muslims in the 2001 Census of England and Wales: Gender and
Economic Disadvantage. Ethnic and Racial Studies, 29 (4), 629-655.
Pearson, K. (1900) On the Criterion That a Given System of Deviations from the
Probable in the Case of a Correlated System of Variables Is Such That It
Can Be Reasonably Supposed to Have Arisen from Random Sampling.
Philosophical Magazine, 50 (5), 157-175.
Phillips, D. (1998) Black BME Concentration, Segregation and Dispersal in Britain.
Urban Studies, 35 (10), 1681-1702.

65
Phillips, D. (2008) Black and BME Young People and Housing. Better Housing Brief
No. 8, London: Race Equality Foundation.
Phillips, D., Davis, C. and Ratcliffe, P. (2007) British Asian Narratives of Urban
Space. Transactions of the Institute of British Geographers, 32 (2), 217-234.
Platt, L. (2011) Inequality within Ethnic Groups. York: Joseph Rowntree Foundation.
Provan, B., Belotti, A., Lane, L. and Power, A. (2017) Low Cost Home Ownership
Schemes. London: Social Mobility Commission.
Ratcliffe, P. (1981) Racism and Reaction: A Profile of Handsworth. London:
Routledge and Kegan Paul.
Ratcliffe, P. (2001) Breaking Down the Barriers: Improving Asian Access to Social
Rented Housing. Coventry: Chartered Institute of Housing.
Ratcliffe, P. (2009) Re-Evaluating the Links between ‘Race’ and Residence. Housing
Studies, 24 (4), 433-450.
Reeve, K. and Robinson, D. (2008) Beyond the Multi-Ethnic Metropolis: BME
Housing Experiences in Small Town England. Housing Studies, 22 (4), 547-
571.
Rex, J. and Moore, R. (1967) Race, Community and Conflict. Oxford: Oxford
University Press.
Robinson, D. (2002) Missing the Target? Discrimination and Exclusion in the
Allocation of Social Housing IN Somerville, P. and Steele, A. (eds.) 'Race',
Housing and Social Exclusion. London: Jessica Kingsley, 94-113.
Robinson, D. (2008) New Immigrants and Migrants in Social Housing in England.
Sheffield: Sheffield Hallam University, Centre for Regional Economic and
Social Research.
Robinson, D. (2010) New Immigrants and Migrants in Social Housing in Britain:
Discursive Themes and Lived Realities. Policy and Politics, 38 (1), 57-77.
Robinson, D., Iqbal, B. and Harrison, M. (2000) Bids to Black and BME Housing
Opportunities. Research Report to the Housing Corporation, Sheffield:
Sheffield Hallam University, Centre for Regional Economic and Social
Research.
Robinson, D., Reeve, K. and Casey, R. (2007) The Housing Pathways of New
Immigrants. York: Joseph Rowntree Foundation.
Robinson, D., Reeve, K., Coward, S., Bennington, J. and Buckner, L. (2005) BME
Housing Experiences in North Lincolnshire. Sheffield: Centre for Regional
Economic and Social Research, Sheffield Hallam University.
Rutter, J. and Latorre, M. (2009) Social Housing Allocation and Immigrant
Communities. Manchester: Equality and Human Rights Commission.
Sarre, P., Phillips, D. and Skellington, R. (1989) Ethnic Minority Housing:
Explanations and Policies. Aldershot: Avebury.
Sellik, P. (2004) Muslim Housing Experiences. London: Housing Corporation.
Shelter (2005) Briefing: Fair Allocations and Choice Based Lettings. London: Shelter.

66
Shelter (2008) Policy Briefing: Worklessness and Social Housing. London: Shelter.
Shelter (2016) Choice-Based Lettings. Available from:
https://england.shelter.org.uk/housing_advice/council_housing_association/ch
oice-based_lettings [Accessed 18 October 2017].
Simpson, L., Ahmed, S. and Phillips, D. (2007) Oldham and Rochdale: Race,
Housing and Community Cohesion. Cathie Marsh Centre for Census and
Survey Research (CCSR) Working Paper No. 200815, Manchester: CCSR,
University of Manchester.
Somerville, P. (2001) Allocating Housing—or ‘Letting’ People Choose? IN Cowan, D.
and Marsh, A. (eds.) Two Steps Forward: Housing Policy into the New
Millennium. Bristol: Policy Press.
Spurr, H. (2014) Councils Bar 113,000 from Waiting Lists. Inside Housing, Available
from: https://www.insidehousing.co.uk/news/news/councils-bar-113000-from-
waiting-lists-39676 [Accessed 26 March 2018].
Thornhill, J. (2010) Allocating Social Housing: Opportunities and Challenges.
Coventry: Chartered Institute of Housing.
van Ham, M. and Manley, D. (2014) Segregation, Choice-Based Letting and Social
Housing: How Housing Policy Can Affect the Segregation Process IN Lloyd,
C., Shuttleworth, I. and Wong, D.W. (eds.) Social-Spatial Segregation:
Concepts, Processes and Outcomes. Bristol: Policy Press, 247-268.
Watts, B., Johnsen, S. and Sosenko, F. (2015) Youth Homelessness in the UK: A
Review for the OVO Foundation. Edinburgh: Heriot-Watt University.
Webb, M.A., Kernaghan, D. and Caffrey, M. (2014) Believe in Childcare? The
Childcare Needs of Ethnic Minority Communities in Northern Ireland. Policy
and Practice Briefing No. 17. Belfast: Barnado's Northern Ireland.
Wood, S. (2013) Developing an Evidence-Base for Housing and Related Services for
Older People. Oldham: Aska Homes.

67
Appendix 1
Table 24. Economic status of male single tenants, %.
Table 25. Economic status of female single tenants, %.
White
Black
Asian
Mixed
Chinese/Other
Employed
18.6
26.3
23.2
22.9
22.0
Unemployed/in training
21.1
36.6
30.0
30.9
38.5
Economically inactive
60.3
37.0
46.9
46.2
39.6
Total
100.0
100.0
100.0
100.0
100.0
Notes: Calculations are based on cases for which data on ethnicity, sex, and economic status are
complete, i.e. were not refused by the tenant; n = 89,904. Columns may not sum to 100.0% exactly
due to rounding.
Source: 2016/17 Continuous Recording of Social Housing Lettings and Sales data.
White
Black
Asian
Mixed
Chinese/Other
Employed
20.2
33.7
16.6
26.5
25.6
Unemployed/in training
12.5
23.2
29.6
23.4
23.3
Economically inactive
67.3
43.1
53.8
50.1
51.1
Total
100.0
100.0
100.0
100.0
100.0
Notes: Calculations are based on cases for which data on ethnicity, sex, and economic status are
complete, i.e. were not refused by the tenant; n = 73,290. Columns may not sum to 100.0% exactly
due to rounding.
Source: 2016/17 Continuous Recording of Social Housing Lettings and Sales data.