How Neighbourhood Social Mix Shapes Access to Resources from Social Networks and from Services

Abstract

Social mix policies have become controversial. Claims about the harms caused by neighbourhood effects have been challenged while counter-claims have been made about the potential benefits for low-income households from living in poor communities. This paper examines two aspects of this debate: whether deprived communities provide greater access to social networks and hence resources in the form of gifts, and whether they provide worse access to resources in the form of services. Data come from the largest survey of poverty ever conducted in the UK—the Poverty and Social Exclusion UK Survey 2012. Results do not support either position in the debate. They do not suggest that access to services is worse in deprived neighbourhoods for all services, but only for a minority. While people in deprived neighbourhoods report marginally greater contact with family and slightly higher levels of social support, there is no evidence of greater levels of exchange of gifts or reciprocity through social networks.

Keywords::

• Social mix
• neighbourhood effects
• poverty
• social network
• public services
• private services

Introduction

An extensive neighbourhood effects literature explores the consequences of neighbourhood context at one point in time (often childhood) for future welfare outcomes or opportunities. This is usually through a focus on the negative impacts of living in more deprived neighbourhoods, particularly for those on low incomes. The results have frequently been used to support calls for policies to limit segregation and to promote social mix. Such policies have become widespread in developed countries. In the UK, for example, they have included planning policies requiring some level of social mix in new housing developments as well as neighbourhood policies to introduce greater social mix within more deprived locations as part of broader regeneration strategies.

In recent years, such policies have been subjected to a number of criticisms. First, policy is said to have got ahead of the evidence, which is seen as suffering from a number of methodological problems, not least a failure to deal adequately with selection effects (van Ham et al., 2012). Many people remain to be convinced about the true scale—and even direction—of neighbourhood effects. Second, the interventions to promote social mix are criticised for doing more damage than good. The neighbourhood effects research is said to have played a role in legitimising interventions by governments and private developers which harm communities which are already marginalised and disempowered (Darcy & Gwyther, 2012; Slater, 2006). These harms are said to arise from the disruption and dispersal of communities, which can be a source of positive identity, solidarity and reciprocity.

This paper contributes to that debate in two ways. First, it examines the idea that deprived neighbourhoods may be positive environments for people on low incomes by virtue of the greater access they provide to supportive social networks and hence material resources in the form of gifts from family and friends. Second, we explore whether deprived neighbourhoods may be more damaging environments by virtue of the weaker access they provide to a wide range of services, both public and private. Public services are a source of ‘income-in-kind’, so poorer quality of services represents a loss of resources. Poor access to all kinds of services, public or private, can lead to greater costs, another drain on resources.

The paper examines these issues through an analysis of data from the Poverty and Social Exclusion UK2012 (PSE-UK) survey. The PSE-UK survey is the largest and most comprehensive survey of poverty and social exclusion ever conducted in the UK. It aims to provide both breadth by covering economic resources, living standards and circumstances across a wide range of domains of exclusion, as well as depth by virtue of sample size and structure.

The paper begins with Townsend's conceptualisation of relative poverty and how this arises from the lack of command of resources over time. It then explores how neighbourhood context may affect access to resources through its impacts on the resource intensity of social networks, and on access to services. In the Data and Methods section, the paper provides details of the PSE-UK survey and the approach to analysis. In the Findings section, we report results from the different elements of the analysis. Finally, we summarise the findings and discuss possible implications for policy.

The Neighbourhood and Access to Resources

Poverty, Deprivation and Resources

The starting point for our research is Townsend's writings on poverty. Townsend's work (1979, 1987, 1993) is important for several reasons, not least for the extensive influence he had on debates about poverty, both in the UK and internationally. He insisted on the relative nature of poverty—that it had to be defined in relation to the contemporary standards of the society the individual came from. He stressed that it was relational, involving the ability of an individual to function in a social and civic sense, not merely their ability to sustain biological functioning. And he emphasised the direct observation of living standards or deprivation as the means by which to gauge poverty, rather than relying on indirect measures such as those based on income (Ringen, 1988).

For Townsend, poverty is identified through deprivations caused by a lack of command of resources over time:

People are relatively deprived if they cannot obtain, at all or sufficiently, the conditions of life—that is, the diets, amenities, standards and services—which allow them to play the roles, participate in the relationships and follow the customary behaviour which is expected of them by virtue of their membership of society. If they lack or are denied resources to obtain access to these conditions of life and so fulfil membership of society they may be said to be in poverty. (Townsend, 1993, p. 36, emphasis added)

Resources come from diverse sources (Townsend 1979). Cash income comes from employment (formal or informal), from unearned income (e.g. interest on savings or private pensions) and from state transfers or benefits. Assets (including housing) and savings may be a source of resources, while debts are a negative resource. Public services which are free or subsidised at the point of consumption contribute ‘income in-kind’. Access to adequate private services including shops and banking may be important for general functioning, and a lack of such services, low quality or high costs may be financially draining. Finally, resources may come through transfers from family or friends (gifts of cash or goods) or from charitable sources. In this paper, we focus on the last two—services and gifts—and consider the possible influence which neighbourhood social mix may have on these.

Social Networks, Gifts and the Neighbourhood

Within the neighbourhoods literature, there has been an extensive discussion of the value of local social networks or social capital for those on low incomes. Much of the focus of this has been on the impacts of this social capital for ‘getting on’—for accessing opportunities, notably in the labour market—rather than for ‘getting by’. Those on low incomes tend to have social networks which are more limited in size but also more geographically constrained to their local neighbourhood (Forrest & Kearns, 1999; Galster, 2012). Where people live in low-income neighbourhoods, there is therefore a concern that the potential of their networks is further limited by the effects of social mix: they will be more likely to live next to other households with low levels of employment and limited informational networks. Such networks lack ‘bridging’ capital, to use Putnam's (2000) terminology, and, while they may be high in ‘bonding’ capital, this may act to limit aspirations (Forrest & Kearns, 1999; Pinkster, 2007).

In this paper, we want to pay more attention to the value of social networks for ‘getting by’ (Briggs, 1998) and specifically for providing access to economic or material resources. Managing on a low income is a constant struggle for most, requiring ingenuity and effort (Lister, 2004). The ability to draw on material or financial support from family and friends may make a substantial difference, both in routine coping but also in dealing with unforeseen events (Curley, 2010). Family tends to play a greater role than friends in this regard (Lin, 1999). Estimates of the financial value of gifts vary widely. In a US survey, Henly et al. (2005) suggest that such gifts are minor compared to welfare payments but that those in most need had least access to such gifts. In a UK study, however, Taylor & Brown (2011) find that, for some low income families at least, the value of such gifts in cash and in kind forms a significant part of household income—in some cases, as much as half the total.

The distinction between support for ‘getting by’ and that for ‘getting on’ should not be over-stated. The personal networks of low-income households may not be a good source of information about labour market opportunities but they may help households take up paid work if they provide access to cheap, flexible childcare, particularly where work is part-time or sporadic (Howard, 2006). Likewise, we should not measure the value of supportive networks solely by the level of material resources exchanged. Support can be emotional as well as material, although in practice the two are often highly correlated (see Findings section). In this paper, however, it is that aspect of personal networks on which we focus.

In relation to ‘getting by’, it has been suggested that more disadvantaged neighbourhoods may be advantageous places for lower-income groups precisely because they offer higher levels of ‘bonding’ rather than ‘bridging’ capital (Putnam, 2000). Livingston et al. (2010) found that local networks may be an important source of support or even protection in more deprived neighbourhoods. It is this conception of working class communities as sources of reciprocity and solidarity which underpins some of the objections to urban renewal policies noted above (Slater, 2006).

At the same time, it should be stressed that social networks involve reciprocal relationships—they may be a source of support but also of draining obligations. Briggs et al. (2010) follow low-income families as they are provided with assistance to move out of poor neighbourhoods in American cities through the Moving to Opportunity programme. They note that, for many of those who leave, there are reductions in social contact with family and friends from the neighbourhood. In some instances, these reductions are a source of regret and a motivation for moving back. In other cases, these reductions were a positive outcome because draining or damaging ties with wider family networks or friends were broken or at least reduced. Curley (2010) and Howard (2006) also find that networks can be stressful and demanding as well as supportive. In judging the value of local social networks, we therefore need to look both at any material support received from them but also at the reciprocal demands they place on an individual.

Access to and Quality of Local Services

The second way in which neighbourhood social mix may shape resources is through its influence on access to local services. The classic literature on equity in urban services focuses mainly on geographical accessibility and on income/class divides (Davies, 1968; Pinch, 1985; Smith, 1977; Troy, 1982). One strand in this literature argues that the middle classes and middle-class areas have diverse ways of influencing service provision in their favour (Goodin & Le Grand, 1987; Le Grand, 1982). Policy initiatives focused on regenerating deprived areas have also tended to argue that poor local services are part of the problem to be addressed, including problems with ‘private’ services like retailing and finance as well as public services (Social Exclusion Unit, 1998, 2001). Despite these concerns with spatial equity, public services are, at least in the UK and Europe, predominantly a mechanism for redistribution from general taxation to the general population as a whole and to lower income groups specifically (Sefton, 1997), and this is reflected in the picture of spatial distribution of public spending (Bramley & Evans, 2000).

To properly understand how access to services varies by neighbourhood, it is necessary to consider the different types of local services and their logic of operation. For example, local ‘public goods’, universal, demand-driven and needs-rationed services would each be expected to show different patterns of usage across socio-economic groups (Bramley, 1996). Usage at neighbourhood level would, as a first approximation, tend to represent a mapping of the relevant socio-demographic groups. However, a second-order effect might still be observed, whereby people in poorer neighbourhoods used services less than expected, or were more dissatisfied with their quality, because of constraints on access, cultural preferences or poorer quality ‘residualised’ provision. In order to detect such neighbourhood effects, it would therefore be necessary to control for the expected effects of socio-demographics within any modelling.

The spatial economics of provision of some services, like buses, makes these denser urban locations easier to provide for and may therefore enable a more intensive service for a given level of resource. At the other end of the spectrum are rural areas, where it may be difficult to make some services viable at all. There is clear evidence in the PSE-UK and predecessor surveys of greater service constraints in rural areas for a range of mainly private services including dentists, opticians, post offices, chemists, supermarkets, buses and trains (Bramley, 1997). This underlines the need to also control for urban–rural situation in any modelling.

There is a greater expectation of adverse differences in service quality for poorer neighbourhoods, and some evidence to support this although also some doubts have been raised about the consistency of subjective reported quality (Bashir, 2011; Duffy, 2000). Ideally, we would seek more objective ways of evidencing quality, particularly through generating and analysing measures of outcomes, but this gets into significant issues about ‘co-production’ going beyond the scope of this paper. The data-set used for the analysis presented further uses a combined indicator of usage, access and one aspect of quality (‘adequacy’).

The conventional expectation, based on this literature, is that poor households, for whom public and locally based services are particularly important, will typically receive a poorer quality of service. By extension, the quality may be expected to be particularly low in poor neighbourhoods. This is referred to in some of the ‘neighbourhood effects’ literature as the ‘de-institutionalization theory’, which stresses that poor neighbourhoods will lack the middle-class social capital or leadership to support or improve local service organisations (Small et al., 2008; Wilson 1987). In the USA and some other countries, with less systematic policies to counter social exclusion and urban decline, this process of institutional disinvestment may be more advanced, as part of a wider process of ‘territorial stigmatization’ (Wacquant, 2008; Wacquant et al., 2014). Interestingly, some recent work questions this, pointing to evidence that institutions providing services to poor neighbourhoods are either better resourced or better connected, or are simply providing more relevant services for poor households than those available in more affluent areas (Curley, 2010; Pinkster, 2007; Small et al., 2008).

A further reason why services may play a more positive role in poor areas is that service institutions and their settings provide local meeting places for residents from poor areas, at which they may gain more social network connections in general and those of a supportive nature in particular (Curley, 2010; Small, 2009). These factors may lie behind some of the empirical results reported further.

Summary and Research Questions

The aim of this paper was to explore a number of potential relationships between neighbourhood and access to resources, with neighbourhood seen as a contextual factor potentially influencing access to resources in the form of gifts and access to important local services. The key questions can be summarised as follows:

• Is living in a more deprived neighbourhood associated with greater receipt of gifts?

• Is living in a more deprived neighbourhood also associated with greater giving of gifts?

• Is living in a more deprived neighbourhood associated with worse access to services?

We explore these relationships for the population as a whole but also for those in poverty, since this is the group in whose name ‘social mix’ policies have been pursued. While our ultimate interest is in the impact of neighbourhood context on access to resources through gifts, the analysis also looks at the possible influence of important intervening factors, notably the scale and quality of personal networks: are personal networks larger and/or more supportive for those in more deprived neighbourhoods and, if so, is it this which helps explain relationships between neighbourhood context and the receiving or giving of gifts?

Data and Methods

The PSE-UK Survey

The PSE-UK survey is based on re-interviewing a sample of households previously interviewed for a large Government survey, the Family Resources Survey (FRS) 2010/2011. This has the advantage of permitting the targeting of the sample but also of providing a wealth of additional information on household incomes and resources, including information on the giving and receiving of gifts. The PSE-UK survey captured a wealth of detail on living standards and access to public services, amongst many other matters.

Fieldwork for the PSE-UK was conducted between February and October 2012, 12–18 months after the FRS survey. The PSE-UK survey checks whether characteristics such as income have changed in the intervening period. The sample included several boosts to improve data for key groups, notably for low-income households and people from minority ethnic groups. Weights allowed for unequal chances of selection and adjusted the sample to match the age/sex population structure for region/country established by the 2011 Census. Interviews were achieved with 5193 households (59 per cent response rate). Within those households, full interviews were achieved with 7511 adults (83 per cent of those present).

Individual Deprivation and Poverty

The PSE-UK constructs a direct measure of deprivation using a consensual or democratic methodology. This builds on Townsend's (1979) initial work and subsequent developments, particularly by Mack & Lansley (1985). First, a public opinion survey identifies a set of items or activities which the public regard as ‘necessities’: things which the majority of the population believe everyone should be able to afford and which no one should have to go without. The aim is not to identify a comprehensive basket as is the case in work on ‘minimum income standards’ (Davis, 2012), but rather an indicative set of items and activities from across the range of areas of consumption or social life: housing, food or household goods, as well as leisure, recreation and social activities. Separate lists are produced for adults and children.

Second, a survey of household living standards (the PSE-UK main survey) is used to identify the proportion of the population lacking each necessity because they cannot afford it. Extensive statistical checks are used to ensure that the set of necessities works as a valid, reliable and additive instrument identifying one latent variable, deprivation (Gordon, 2006; Guio et al., 2012). The deprivation measure is a count of the number of necessities items which people lack. In the analyses here, the scale is capped at 12 or more items. The average is 2.4 but the median is 1. The correlation with income (equivalised, after housing costs, natural log) is moderate ( − 0.42), partly because the measures are designed to discriminate at different points in the distribution and partly because income is a single snapshot, whereas living standards and deprivation are determined by command of resources over time.

The deprivation measure is combined with income (equivalised, after housing costs) to identify individuals regarded as ‘poor’. First, a threshold level of deprivation is selected by looking at the relationship between deprivation and income, selecting the point on the deprivation scale where between-group income differences are maximised (Gordon, 2006). For the PSE-UK survey, the threshold is three deprivations. Second, this deprivation threshold is used to identify an equivalent low-income poverty line—the median (equivalised) income for those on the deprivation threshold. The ‘poor’ are those who meet both deprivation and low-income tests (22 per cent of all adults).

Neighbourhood Context

To the household survey data, we have attached two variables to measure neighbourhood context: neighbourhood deprivation and urban–rural location. The neighbourhood deprivation measure is taken from the various national neighbourhood deprivation indices which have been constructed separately in England, Wales, Scotland and Northern Ireland. These use similar approaches but with differences in detail. The most comparable element within them is a measure of ‘income deprivation’ based on the proportion of people in each neighbourhood in receipt of a low-income (means-tested) benefit. While there are still some subtle differences between the countries in the list of benefits included and in timing (Payne & Abel, 2012), this measure is quite comparable across the four nations of the UK. The urban–rural measure is taken from various government classifications and provides a fourfold classification from ‘large urban’ to ‘village/rural’.

Measures are constructed for neighbourhood units which have a population between 1000 and 2000 in England and Wales and between 500 and 1000 in Scotland. To protect respondent identity and preserve confidentiality, both locational variables have been ‘blurred’ by adding a small amount of random noise. With income deprivation, the percentages are averaged within deciles, i.e. the models use the average proportion of people ‘income deprived’ in each decile as a continuous or scale measure.

There has been a great deal of debate within the ‘neighbourhood effects’ literature about the nature of the relationship between neighbourhood context and various outcomes. Several studies have provided evidence that relationships may be nonlinear and that there may be threshold effects or ‘tipping points’; the latter usually implies a level of neighbourhood poverty beyond which damaging processes accelerate or individual outcomes ‘deteriorate’ more rapidly (Galster, 2012; Galster et al., 2000). We had no a priori expectations here, but explored this systematically using a variety of model specifications, including comparing models with a simple linear term to those which included additional dummy terms for the more deprived neighbourhoods. Overall, we found that the addition of these terms did not add to the explanatory power of the models (in no case did the proportion of the variance explained rise by even 1 per cent). Such effects as were apparent suggested a slight reduction or attenuation of the underlying relationship with neighbourhood deprivation, not the kind of acceleration usually associated with ‘threshold effects’. In this paper, we therefore present results only for the simple linear relationships.

Other Variables

We construct several dependent variables for our analyses, including measures of: the scale and quality of social networks; the giving and receiving of material help; and the experience of service constraints. Details of these dependent variables are provided further in the relevant analytical sections, along with details of independent control variables included in the models.

Analysis

For outcomes related to social networks, support and help given or received, we construct linear regression models. The ultimate interest is in the relationship between neighbourhood deprivation and gifts received or given, but the relationships between deprivation and both networks and support are examined as important intervening factors or causal pathways. These models use an identical set of explanatory variables, so results can be summarised in a single path analysis. In some cases, the distribution of the errors is not normal, violating one of the assumptions of these models. The findings are therefore checked using logistic regression models with a binary version of the outcome. The relationships identified in the logistic models are consistent with those from the linear models in every case: in the same direction, with the same relative magnitude between models and usually with the same significance level. We therefore report only the linear models.

The models relating to access to services are logistic regressions. They use the same explanatory variables including the continuous measure of neighbourhood deprivation but also include the measure of urban–rural location (three dummies, with the default ‘large urban’). The impact of rurality on access to services is a major theme in the literature, but there is also a relationship between deprivation and urban–rural location, so it was important to control for the latter here.

At each stage, we construct separate models for (i) all adults and (ii) poor adults only. The latter enable us to identify whether the relationships between neighbourhood context and the various outcomes are the same for poor adults as for the population as a whole (interaction effects). These models include controls for individual deprivation and income in order to allow for variations within the ‘poor’ population.

Findings

Resources from Social Networks or Social Capital

We are interested in how neighbourhood shapes help given or received directly, but also how it shapes the scale and quality of social networks which may in turn affect help given or received. We start with the last of these.

Scale of networks

The size of personal social networks and the frequency of contacts are captured by four questions on: the number of (i) relatives and (ii) friends with whom respondents have contact at least once a month, and the frequency of contact in each case. Answers to all four questions are captured on banded scales. Correlations and factor analysis suggest that rather different patterns of contact exist for family and for friends, but that the measures for the number and frequency of contacts are strongly related in each case. We therefore aggregate the questions by summing to give two measures (rescaled from 0 to 100): contact with family and contact with friends. The maximum score indicates people who have contact with nine or more family members or friends at least once a month, and have daily contact with at least one of them. The average score is 63 for family and 66 for friends (standard deviations 26 and 27 respectively). The correlation between the scales is quite modest (0.18).

Women report more contact than men with family but there are similar levels of contact with friends. Young adults (18–24) are more oriented to friends, but otherwise there is little variation with age for either type of contact. People with children have more contact with family, particularly lone parents and those with very young children. People from the White majority have slightly higher levels of contact with family and friends than those from minority ethnic backgrounds with some exceptions; most notably, people of Asian ethnicity tend to have the highest levels of contact with family. People with a longstanding illness or disability have less contact with friends but no less contact with family. People who work full time have less contact with family or friends than those working part-time, or unemployed or inactive.

Our main interest is the potential impact of individual and neighbourhood deprivation on levels of contact. Contact with family varies very little with either, although, at the extreme, those with the very highest incomes and the very lowest report less contact. With friends, there is a simpler picture of increasing levels of contact as both individual and neighbourhood deprivation fall. The result is that those on low incomes or in poor neighbourhoods have similar levels of contact with family and friends, while those on high incomes or in more affluent neighbourhoods report much more contact with friends than with family.

To separate out the influence of different factors on levels of contact, we construct linear regression models for the whole population and those poor only (Table 1). Overall, the characteristics included are fairly poor predictors of levels of contact, but almost all of the differences summarised above persist. In addition, the length of time at a particular address is associated with greater levels of contact with both family and friends, either because ties develop over time or because good contacts act as a tie to an area. With individual resources, deprivation is associated with lower levels of contact with family and with friends. In contrast, higher income is also associated with lower contact with family, although the relationship is weaker (not significant in some models). This is unlikely to be a problem of multicolinearity given the moderate correlation between income and deprivation noted already. It is more likely a reflection of the fact that deprivation discriminates between poorer households in a way that income does not. Controlling for these other factors, people in more deprived neighbourhoods report slightly greater contact with family and less with friends.

Table 1 Linear regression models for contact with family and friends

	All		Poor only
	Family	Friends	Family	Friends
	Beta	Beta	Beta	Beta
Age (years)	− 0.51^*^*^*	− 0.78^*^*^*	− 0.60^*^*^*	− 0.60^*^*^*
Age (sqd/100)	0.50^*^*^*	0.68^*^*^*	0.54^*^*^*	0.41^*^*
Female	0.17^*^*^*	0.03^*	0.20^*^*^*	− 0.02
Ethnic minority	− 0.01	− 0.05^*^*^*	− 0.02	− 0.04
Limiting health prob./disability	− 0.01	− 0.05^*^*^*	− 0.08^*^*	− 0.10^*^*^*
Lone parent	0.06^*^*^*	0.01	0.06	0.04
Couple	0.08^*^*^*	− 0.03	− 0.03	0.04
3+Adult	0.00	− 0.02	0.01	0.03
Child 0–4	0.03^*	− 0.04^*	0.05	− 0.14^*^*^*
Child 5–10	− 0.02	0.02	0.02	− 0.03
Child 11+	0.01	0.00	− 0.03	0.00
Years lived at address	0.04^*^*	0.03^*	0.08^*^*	− 0.03
In employment	− 0.01	− 0.03	− 0.09^*^*	− 0.03
Student	− 0.05^*^*^*	0.07^*^*^*	− 0.13^*^*^*	0.03
Income (AHC, log)	− 0.03^*	− 0.01	0.02	− 0.01
Individual deprivation	− 0.06^*^*^*	− 0.12^*^*^*	− 0.08^*^*	− 0.13^*^*^*
Neighbourhood deprivation	0.04^*^*	− 0.03^*	0.04	0.02
R^2 (per cent)	5	5	11	8
N	7194	7181	1669	1668

Notes: Significance: ^*5 per cent; ^**1 per cent; ^***0.1 per cent.

For poor individuals, the relationships are almost identical, although neither is significant. The explanatory power of the models is slightly greater as other individual characteristics have a greater influence. For example, having a limiting health problem or disability reduces contact with both family and friends for poor adults.

Functioning: perceived support

As well as assessing the scale of networks, the PSE-UK survey gives an assessment of the perceived quality of those networks through a set of seven questions on the level of support which respondents believe they would get in a range of hypothetical situations (Table 2). Respondents are directed to think about support from family or friends, or from other sources, so we cannot distinguish support received solely from the former, nor can we distinguish the relative role of local versus more distant social contacts. The questions cover instrumental as well as emotional support. While these may be conceptually distinct, they frequently correlate highly in practice, reflecting a single latent factor (Thoits, 1995). That is certainly the case with our data where factor analysis reveals one underlying factor which explains around 60 per cent of the variance. A Cronbach's alpha test on a scale using all seven items gives a result of 0.89 (generally seen as ‘good’ and close to ‘excellent’); omitting any item reduced this score. We therefore treat them as identifying a single latent variable (‘sense of support’).

Table 2 PSE-UK survey questions on sense of support

How much support would you get if…

• you were ill in bed and needed help at home? [HomeIll]

• you needed practical help around the home e.g. moving heavy furniture, DIY jobs? [Homejob]

• you needed advice about an important change in your life, e.g. changing jobs, moving to another area? [Advice]

• you were upset because of relationship problems or were feeling a bit depressed and needed someone to talk to? [RelProb]

• you needed someone to look after your home or possessions when away? [Things]

• you had a serious personal crisis and needed someone to turn to for comfort and support? [Crisis]

• you needed a lift somewhere in an emergency? [Lift]

Responses: a lot; some; not much; none at all.

Scores on the seven questions are combined in a simple additive manner and rescaled from 0 to 100, giving an average of 82 and a standard deviation of 19. One limitation with this index is that a significant number of people score the highest possible value, suggesting that the underlying questions do not do enough to distinguish people at this end of the scale. This creates some slight problems with the linear regression models (a non-normal distribution of residuals), so we use logistic regression to check findings. Results are very similar so, for simplicity, we present only the linear models below.

Older people tend to express a slightly higher sense of support. People with children, but particularly lone parents, feel they have rather less support as do people from minority ethnic groups and those with a long-standing illness or disability. People who have lived in an area for longer feel they have more support. Some of the biggest variations are with income, where the higher-income groups and the least deprived express significantly higher levels of support. These differences are also reflected at the neighbourhood level.

The question for this paper is whether differences between neighbourhoods are simply compositional, or whether they also suggest an additional contextual effect after controlling for individual characteristics (Table 3). As previously, we present two models for the whole population and two for the poor only. In each case, the first model uses the same set of socio-demographic factors as previously while the second adds the measures of contact with family and friends to identify their impact as possible intervening variables.

Table 3 Linear regression models for sense of support

	All		Poor only
	1	2	1	2
	Beta	Beta	Beta	Beta
Age (years)	− 0.32^*^*^*	− 0.15^*	− 0.03	0.18
Age (sqd/100)	0.34^*^*^*	0.18^*	0.04	− 0.13
Female	0.06^*^*^*	0.02	0.04	0.00
Ethnic minority	− 0.08^*^*^*	− 0.07^*^*^*	− 0.10^*^*^*	− 0.09^*^*^*
Limiting health prob./disability	− 0.04^*^*	− 0.03^*^*	− 0.10^*^*^*	− 0.07^*^*
Lone parent	− 0.01	− 0.03^*	− 0.07^*	− 0.09^*^*
Couple	0.24^*^*^*	0.23^*^*^*	0.13^*^*^*	0.13^*^*^*
3+Adult	0.22^*^*^*	0.22^*^*^*	0.15^*^*^*	0.14^*^*^*
Child 0–4	− 0.03^*	− 0.03^*	0.00	0.01
Child 5–10	− 0.02	− 0.02	− 0.01	− 0.01
Child 11+	0.01	0.01	0.00	0.01
Years lived at address	0.03^*	0.02	0.06^*	0.04
In employment	0.05^*^*	0.05^*^*^*	0.06^*	0.08^*^*
Student	0.07^*^*^*	0.07^*^*^*	0.13^*^*^*	0.15^*^*^*
Income (AHC, log)	0.01	0.02	0.04	0.04
Individual deprivation	− 0.27^*^*^*	− 0.24^*^*^*	− 0.19^*^*^*	− 0.15^*^*^*
Neighbourhood deprivation	0.02	0.02	0.05^*	0.03
Family contact		0.21^*^*^*		0.23^*^*^*
Friends contact		0.09^*^*^*		0.13^*^*^*
R^2 (per cent)	18	23	15	23
N	6883		1587

Notes: Significance: ^*5 per cent; ^**1 per cent; ^***0.1 per cent.

Overall, the models have rather better fit than those for contact with family and friends, although they are still not strong. In terms of individual resources, income and individual deprivation measures now show the same effect, although only deprivation is significant: people who are more deprived feel that they have markedly less support. When we include measures of the scale of networks in the models, both have the expected sign but it is contact with family which is much more predictive of sense of support.

Once we have controlled for these individual characteristics, people in more deprived neighbourhoods report very slightly more support, although the difference is not significant. For those who are poor, the effect is slightly greater but still dwarfed by the negative effect of individual deprivation. In both cases, the relationship is unchanged when we control for contact with friends and family. Overall, we would conclude that there is some evidence that people, particularly poor people, feel a marginally greater sense of support in more deprived neighbourhoods, but this is not due to having stronger family networks.

Help received

The next stage is to look at levels of help received. Both help received and help given were captured in the FRS survey which preceded the PSE-UK survey. There is a gap in timing of around 12–18 months between the two surveys which will introduce some additional noise into the data but which is unlikely to be a source of bias. The kinds of help captured are mostly about direct financial or material transfers (receiving or giving money or goods, or having things paid for or paying for them). They also include more practical help with managing money or benefits, or with household chores (Table 4). There is no direct recording of the scale of help given or received, or of its approximate financial value. Respondents are given a simple score (count) for the number of items which they report having received or given over the previous year. One-third had received some help with an average for the whole population of 0.9 items of help received.

Table 4 FRS survey questions on help received or given

[HelpRec] Some people receive financial or other types of help from their family or friends. Over the past 12 months [since date], have your family or friends helped you by…(READ OUT).

1. …buying or bringing you food or meals?

2. …paying towards bills (such as utility bills, rent or grocery bills (excluding food))?

3. …helping you to manage your money or deal with your benefits?

4. …helping with home repairs or decoration whether by paying for it or doing it for you?

5. …helping with household chores (such as cleaning, gardening) whether by paying for it or doing it for you?

6. …giving you lifts to places or paying for travel costs (such as taxi, train or bus fares)?

7. …paying for trips/holidays?

8. …buying or giving you clothes?

9. …buying clothes, toys or other equipment for your child(ren)

10. …buying a big electrical item like a cooker, boiler, fridge or washing machine?

11. Other help received

12. None of these

13. SPONTANEOUS—Not applicable has no family or friends

[HelpGvn] Some people give financial or other types of help to their family or friends. Over the past 12 months [since date], have you helped your family or friends by…(READ OUT)

[responses as above]

In spite of the limitations of the measure, the models are moderately successful at explaining variations in help received (Table 5). Individual deprivation is a strong predictor of help received which is not surprising since it indicates need. Beyond that, scale of family networks and sense of support are both important, but not friendship networks. People with young children, students and those with long-term health problems or disability receive more help. Those in employment receive less. Length of time at an address is associated with lower levels of help received; perhaps significant proportions of help are associated with moving house or the early years of establishing a new home.

Table 5 Linear regression models for help received

	All		Poor only
	1	2	1	2
	Beta	Beta	Beta	Beta
Age (years)	− 0.43^*^*^*	− 0.37^*^*^*	− 0.51^*^*	− 0.46^*^*
Age (sqd/100)	0.29^*^*^*	0.22^*^*	0.36^*	0.32^*
Female (male)	0.01	− 0.01	0.02	0.00
Ethnic minority	− 0.09^*^*^*	− 0.09^*^*^*	− 0.12^*^*^*	− 0.11^*^*^*
Limiting health prob./disability	0.04^*^*^*	0.05^*^*^*	0.02	0.04
Lone parent	0.02	0.01	− 0.04	− 0.03
Couple	− 0.07^*^*^*	− 0.10^*^*^*	− 0.10^*^*	− 0.12^*^*
3+ Adult	− 0.06^*^*^*	− 0.08^*^*^*	− 0.01	− 0.03
Child 0–4	0.12^*^*^*	0.12^*^*^*	0.13^*^*^*	0.12^*^*^*
Child 5–10	0.03^*	0.03^*	0.06^*	0.06^*
Child 11+	0.01	0.01	0.05	0.06
Years lived at address	− 0.07^*^*^*	− 0.08^*^*^*	− 0.09^*^*	− 0.10^*^*^*
In employment	− 0.04^*^*	− 0.04^*^*	− 0.11^*^*^*	− 0.11^*^*^*
Student	0.06^*^*^*	0.06^*^*^*	0.05	0.04
Income (AHC, log)	− 0.04^*^*	− 0.04^*^*	0.00	− 0.01
Individual deprivation	0.19^*^*^*	0.22^*^*^*	0.06^*	0.09^*^*^*
Neighbourhood deprivation	0.02	0.02	− 0.02	− 0.03
Family contact		0.07^*^*^*		0.09^*^*
Friends contact		0.00		− 0.01
Sense of support		0.09^*^*^*		0.11^*^*^*
R^2 (per cent)	18	19	11	13
N	6883		1587

Notes: Significance: ^*5 per cent; ^**1 per cent; ^***0.1 per cent.

Once individual factors are taken into account, people in more deprived neighbourhoods received slightly more support although differences are not significant. This does not change when we include measures of contact with family or friends, or of support. This suggests that, as with support, any effect is not due to stronger social capital but is a direct effect of neighbourhood deprivation. For those who are poor, the picture is essentially the same although the main effect appears to be of lower help received in more deprived neighbourhoods.

The results can be drawn together in a path diagram which shows how neighbourhood deprivation is related to help received, both directly and through its influence on social networks and support (Figure 1). For comparison, the effects of individual deprivation are also shown. The overall effects of both neighbourhood and individual deprivation (direct and indirect) are shown by the respective coefficients in the first model in Table 5 (without controls for social networks and support). Three points come out of this path analysis. First, individual deprivation is a much greater influence on levels of help received than where someone lives and that is true for the poor as much as for the population as a whole. Second, the overall effect of neighbourhood deprivation appears negligible once other characteristics have been controlled for. Even with a large sample, the coefficients are not significant while, for the poor only, the estimate is actually negative. Third, the effect of living in a more deprived neighbourhood (such as it is) does not appear to stem from the greater contact with family or the higher levels of support reported by people in more deprived neighbourhoods. Rather, it is a direct effect, i.e. as yet unexplained.

Figure 1 Path diagram for contact, support and help received. Notes: Figures on paths are standardised regression coefficients for whole population and, in square brackets, for ‘poor’ only. Solid arrow—positive relationship; dashed arrow—negative relationship. Thickness of arrow indicates approximate strength of relationship. ‘Individual deprivation’ is shown twice to reduce the number of crossing arrows.

Help given

The final step in this part of the analysis is to look at help given and, in particular, whether people in more deprived neighbourhoods appear to be involved in patterns of reciprocal exchange and hence, in material terms at least, in a zero-sum game. This is not to imply that such exchange has no resource value—it can be important in smoothing ups and downs of income. Nor is it to downplay the potential wider benefits (e.g. psychosocial) which might arise from the reassurance such solidaristic networks can provide.

To explore this, we use the same set of factors from the model of help received to model help given. Nearly half the sample said they had given help with an average of 1.2 types of help given. This is 40 per cent higher than the number of people who said they had received help, which suggests some misreporting by givers or receivers, or both. The models for help given have a very poor fit, indicating that the factors which determine whether one gives help are quite different to those which determine whether one receives it. Higher income is associated with more giving, mirroring the effect of low income on likelihood of receiving (Table 6).

Table 6 Linear regression models for help given

	All		Poor only
	1	2	1	2
	Beta	Beta	Beta	Beta
Age (years)	0.77^*^*^*	0.86^*^*^*	− 0.03	0.03
Age (sqd/100)	− 0.74^*^*^*	− 0.82^*^*^*	0.03	− 0.02
Female (male)	0.00	− 0.02^*	− 0.01	− 0.02
Ethnic min (not)	− 0.06^*^*^*	− 0.06^*^*^*	− 0.15^*^*^*	− 0.14^*^*^*
Health prob./disab. (no)	− 0.01	− 0.01	− 0.05	− 0.04
Lone parent (not)	0.02	0.01	− 0.06	− 0.06
Couple (single)	0.01	0.00	− 0.10^*	− 0.10^*
3+ Adult (single)	0.02	0.02	0.05	0.04
Child 0–4 (none)	0.01	0.00	0.06	0.06
Child 5–10 (none)	− 0.03^*	− 0.03^*	0.07^*	0.07^*
Child 11+ (none)	− 0.04^*^*	− 0.04^*^*	− 0.01	0.00
Years lived at addr.	− 0.03	− 0.03^*	− 0.05	− 0.05
In employment (not)	0.02	0.03	0.01	0.02
Student (not)	0.00	0.00	− 0.05	− 0.04
Income—log	0.11^*^*^*	0.11^*^*^*	− 0.01	− 0.01
Deprivation score	0.00	0.01	− 0.06^*	− 0.05
Nhd depvn score	− 0.07^*^*^*	− 0.07^*^*^*	− 0.06^*	− 0.07^*^*
Family network scale		0.11^*^*^*		0.06^*
Friends network scale		0.04^*^*^*		0.06^*
Sense of support		0.00		0.00
R^2 (per cent)	5	6	4	5
N	6883		1587

Notes: Significance: ^*5 per cent; ^**1 per cent; ^***0.1 per cent.

The models include scale and quality of networks. We might reasonably expect people with larger networks and more frequent contact to give help more often. It is less clear that more supportive networks would lead to more giving—highly ‘draining’ networks might be perceived as unsupportive—but we include support here for comparability. Greater contact with both family and friends, but particularly the former, are also associated with more giving, for all people and for the poor. This reinforces the idea that networks can be a ‘drain’ as well as a source of support (Briggs et al., 2010; Howard, 2006). There is no relationship with support.

Controlling for all these factors, people who live in more deprived neighbourhoods gave less rather than more, in spite of the fact that people in the same areas were more likely to be recipients of help. The negative relationship was significant and was there for poor people as well as all people. The implication of this is that there is a net inflow of help through giving into deprived neighbourhoods from other less-deprived places.

Access to and Quality of Local Services

We move on now to examine the situation for services. Households are asked, in relation to a range of services, whether they use the service or not, reasons for not using it, and whether they regard it as adequate. The survey covers 17 general services, six more targeted at children or young people, and a further five for elderly or disabled people. Alternative responses are: ‘use—adequate’, ‘use—inadequate’, ‘don't use—inadequate or unavailable’, ‘don't use—can't afford’, or ‘don't use—don't want to or not relevant’. Different indicators can be derived from this (Bramley & Besemer, 2015). The broadest, which we use here, is that of ‘service constraints’ which combines the second, third and fourth responses, i.e. services are viewed as inadequate, unavailable or not accessible on grounds of cost or affordability.

In simple descriptive terms, people in more deprived areas report lower adequacy for half of the services (14 out of 28), but better adequacy for 10—a very mixed picture. To remove the effect of other confounding demographic factors, notably individual deprivation, we use logistic regression models (Table 7); control variables as previously, but with the addition of dummies for urban–rural location. Cell values show the estimated effect of neighbourhood deprivation rate on the odds ratio (the exponent of the regression coefficient, B) of finding a service inadequate; a value above one indicates that people in more deprived neighbourhoods experience greater constraints on access to services. Following the discussion above, general services are shown in three groups: universal (demand-driven) local government services; health-related and other regulated services; and commercial retail and utility services.

Table 7 Effects of neighbourhood deprivation on local service usage constraints

	All	Poor only
General services
Libraries	1.02^*^*^*	1.03^*^*^*
Public sports	1.00	0.99
Museums and galleries	0.99^*^*^*	0.98^*^*^*
Evening classes	1.00	1.00
Community hall	0.99^*^*^*	0.98^*^*
Doctor	0.99^*	0.99
Dentist	0.99^*^*	0.98^*^*
Optician	1.02^*^*^*	1.00
Post Office	0.99^*	0.99
Citizens’ advice	1.00	0.99^*
Chemist	1.00	0.98
Corner shop	0.99^*^*^*	0.98^*
Supermarket	0.99	0.98
Bank, building society	1.00	0.99
Pub	1.01^*^*^*	1.02^*
Bus services	0.98^*^*^*	0.98^*^*
Train/tube service	0.99^*	0.99
Children's services
Children's play	1.03^*^*^*	1.04^*^*^*
School meals	0.99	1.01
Youth clubs	1.02^*	1.01
After school club	0.98^*	0.97^*
School transport	0.99	1.00
Nursery	0.98	0.96^*
Older or disabled people's services
Home care	1.05^*^*	1.04
Meals on Wheels	1.01	0.96
Special transport	1.02	0.97
Day centres	1.02	1.02
Chiropodist	1.02	0.99

Notes: Table shows the impact of the continuous neighbourhood deprivation measure on the odds of facing service constraints (exponent of the regression coefficient, B). These are taken from 56 separate logistic regression models with the same controls as previous models including age, gender, ethnicity, health/disability constraints, household composition, and individual income and deprivation. Significance: ^* 5 per cent; ^** 1 per cent; ^*** 0.1 per cent.

The results suggest that there is an association between service constraints and neighbourhood deprivation in half of the general services (using the 1 per cent significance threshold as the cut-off), but only in three cases (libraries, opticians and pubs) are constraints greater in more deprived neighbourhoods, and for these the difference is not large. In five of the general services, problems are slightly lower in more deprived places, including important basic services such as dentists, corner shops and buses. There is slightly more evidence of problems for deprived areas in relation to children's services (one of six having greater constraint but none showing less) and those for older people (one out of five having greater constraint—home care—but all in the same direction). The results for poor adults are essentially the same; the direction of the relationship is the same in almost every case, although the size and significance of the odds ratios vary a little.

Discussion and Conclusions

There is a vigorous on-going debate over the rights and wrongs of ‘social mix’ policies which hinges, on the one hand, on the strength of evidence about damaging ‘neighbourhood effects’ and, on the other, on views about the strengths or value of existing low-income communities—of social ties, support and reciprocity within them. This paper uses the UK evidence to try to shine some light on this by asking whether there is a positive or negative value of living in more deprived communities in terms of access to some kinds of resources. The results do not provide support for either side in the debate.

On social networks and resources, there is no evidence of greater levels of reciprocity in deprived neighbourhoods. Living in a poorer neighbourhood does not provide any greater access to resources through gifts of financial or practical assistance. More deprived individuals tend to receive more gifts but, once we control for this, people in deprived places are no different; the picture is the same if we look only at people who are poor. People who live in more deprived places do report a slightly greater sense of support and some of this comes from having marginally greater contact with family. These networks and support may be an important source of reassurance or emotional assistance. They do not, however, lead to an enhanced ability to access economic or material resources through gifts. On the other hand, we also find no evidence that living in a deprived neighbourhood leads to an additional drain on resources through the giving of gifts. On the contrary, the giving of gifts is somewhat lower in these places. On this measure, people in more deprived neighbourhoods are not generally involved in any greater levels of reciprocity than anyone else.

With services, the hypothesis that usage and access are worse in poor areas is also not borne out for the majority of general services examined here, even those which are market-driven. There are some services where access is worse, but there are more where people in deprived neighbourhoods report fewer constraints. This includes important health-related services such as dentists, as well as transport. There is slightly more support for the constraint hypothesis in the services targeted at children or at older or disabled people (children's play or older people's home care). These differences continue to give cause for concern, since these services are supposed to be more redistributive and needs-based.

Overall, the services findings suggest that the crude ‘de-institutionalization’ thesis does not hold up under UK conditions. We would suggest that such tendencies are countered by (a) the geographical or sparsity factor in service access combined with the greater concentration of the poor in denser urban locations; and (b) the conscious efforts of national and local governments to improve service provision for the most deprived areas, particularly in the period 1998–2010 under the National Strategy for Neighbourhood Renewal. On this evidence, it could even be suggested that low-income households had a modest incentive to remain in more deprived neighbourhoods in order to benefit from better service access although this is likely to be one factor among many affecting mobility decisions. It should also be acknowledged that this part of the analysis does not look at possibly the most important and controversial local service, education. Indeed, the analysis does not look at the quality of services in general beyond judgements about basic access and adequacy.

On balance, this evidence does not provide a compelling argument to pursue social mix policies nor to halt them. For the services we examined, any damaging effects of service constraint are quite selective and might be addressed more effectively through service-specific reforms than through the cost and upheaval of mixed communities’ strategies. On the other hand, if evidence from other research suggested that greater social mix might improve a range of welfare outcomes, our evidence suggests that any resulting social disruptions may be less damaging than some have argued. People in more deprived neighbourhoods do tend to report a slightly higher sense of support, but they do not appear to be involved in networks of reciprocal material aid to any greater extent than anyone else.

The PSE-UK survey makes it possible to study the experience of poverty in different neighbourhood contexts due to its large sample, the significant boost for low-income households and the breadth of coverage of diverse aspects of poverty and social exclusion. The limitations of such a wide-ranging survey are also apparent, notably in the slightly crude measures for some important outcomes: the data do not attempt to assess service quality which is potentially a source of major differences between neighbourhoods; the number of different kinds of gifts given or received is possibly only a weak guide to the economic value of these. There may also be concerns about selection and inferring causality from this cross-sectional survey. Future research could usefully develop stronger instruments for these and employ them in longitudinal designs.

Acknowledgements

The PSE-UK survey involved a large team of researchers led by Professor Dave Gordon at the University of Bristol, including academics from the Universities of Glasgow, Heriot-Watt, Queens Belfast and York, and the Open University. We would like to acknowledge the great collective efforts by this team on the design and delivery of the surveys which underpin this analysis. None of the authors has any conflict of interest in relation to this work.

Additional information

Funding

The research on which this paper is based was funded by an ESRC large grant, the Poverty and Social Exclusion in the UK 2012 Survey (RES-060-25-0052).

Notes

¹ Email: [email protected]

² Email: [email protected]

³ Email: [email protected]