Moving away from opportunity? Social networks and access to social services

Abstract

Chinese

Strong social connections often deter residential mobility beyond reach of the social network. A missing link in the body of research on this subject is the significance of the role of social networks in pooling resources for costly services and neighbourhood-level access to social services. Few have explored whether assistance from local social service agencies may substitute for practical help from social networks, thereby enabling low-income assisted renters to locate housing in more desirable neighbourhoods. Relying on data from the Moving to Opportunity experiment, this article examines the impact of social networks and social services on the dynamics of residential mobility. I find that the existence of social networks in the place movers left behind tends to increase the likelihood of moving back, but this likelihood varies with current access to social service providers and distance moved. These findings suggest that policy efforts in spatial dispersion of poverty should pay close attention to the geography of social services.

Keywords

housing migration planning policy poverty/exclusion

Introduction

Housing assistance programmes often address more than the provision of affordable housing. Deconcentrating assisted renters may play a substantial role in achieving greater racial and economic diversity in neighbourhoods. While retaining approaches to the creation and preservation of affordable housing, the US Department of Housing and Urban Development (HUD) has expanded the scope of housing policy through the Housing Choice Voucher programme to ‘expand housing choice outside areas of poverty or minority concentration’ (US Department of Housing and Urban Development, 2001). The Moving to Opportunity (MTO) experiment, a 15-year housing mobility demonstration conducted by HUD, was designed in the expectation that housing dispersal would alleviate individual poverty through enhanced employment and earnings (Goetz and Chapple, 2010). However, the MTO literature found that even when participants are initially required to use their voucher in low-poverty neighbourhoods, many eventually returned to high-poverty neighbourhoods once the requirement had been lifted (Sanbonmatsu et al., 2011). Explanations for the inability to maintain residence in low-poverty neighbourhoods include rising housing costs (Rosenblatt and DeLuca, 2012), a lack of access to vehicle and public transportation (Dawkins et al., 2015; Shroder, 2002) and pre-existing social ties to neighbours and families (De Souza Briggs et al., 2010).

Social networks play a key role in the relocation of housing voucher recipients. The vast body of literature on the link between social networks and residential mobility of low-income households finds that social relations are the driving factor in residential decisions (DaVanzo, 1981; Fischer, 1982; Kadushin and Jones, 1992; Stack, 1974). Nevertheless, little attention has been paid to the links between social networks, social service accessibility and residential mobility. Without adequate access to social services, living in an unfamiliar neighbourhood may be particularly difficult if households depended on their social networks in a prior neighbourhood for costly services such as childcare. Despite the growing evidence of the spatial mismatch between urban poverty and social service activities, access to social services has been overlooked in policy discussions, and no studies to date have explored the impacts of households’ social networks and neighbourhood-level social services on the residential mobility patterns of low-income households.

This article examines the links between social networks, access to social services and residential mobility for participants of the MTO programme. Focusing on the probability of moving back to the original neighbourhood, I place emphasis on the role of social networks and social services as they shape these residential trajectories. I estimate logit models to examine how both social networks in the place that households left behind and access to social services in the newly selected neighbourhood influence subsequent mobility decisions. I find that social networks, access to social services and distance moved are important factors that shape residential spells and transitions to new neighbourhoods. Holding other variables constant, the likelihood of moving back to the baseline neighbourhood from their second neighbourhood is 167 per cent higher when there is kinship or friendship in the baseline neighbourhood, 243 per cent higher if they moved from a neighbourhood adjacent to the baseline neighbourhood and 23 per cent lower with one additional social service provider. In the conclusion section, I discuss policy implications to promote the expansion of housing options in more desirable locations.

Literature review

Social networks and residential mobility

Residential mobility is generally viewed as a way to bridge the gap between one’s desired housing bundle and the actual housing bundle one consumes. The decision of whether to move can be seen as weighing satisfaction with current housing relative to the anticipated satisfaction with alternatives (Speare, 1970, 1974). In many cases, staying in the existing neighbourhood may reflect that a family lacks the resources to move to better housing or to a desirable neighbourhood. Previous studies regarding residential mobility in urban poverty areas find that minority households with a low income tend to remain in high-poverty neighbourhoods despite mobility intentions (Gramlich et al., 1992; South et al., 2005). Residential mobility is also viewed from a life-course perspective, in that life events are potential causes of mobility decisions (Clark, 2005; Clark and Withers, 1999; Coulton et al., 2012).

People often make mobility decisions within the context of social relations. As the strength of weak ties¹ connects different social groups with each other (Granovetter 1973), weak ties among neighbours centred on the residential neighbourhood can be effective and useful in accomplishing shared goals, such as keeping the streets safe or keeping an eye on local children (Sampson, 1999; Sampson et al., 1997). Such intra-neighbourhood social networks are based on informal channels of trust and repeated local social interaction, and therefore are difficult to replicate in new surroundings (DaVanzo, 1981). Social networks also uniquely shape the informational context of decision making that affects the perception of neighbourhoods. Galster and Killen (1995) conceptualised the role of local social networks in the geography of opportunity that filters information and guides individuals to form subjective perceptions of the opportunity structure. The resources through social networks can be described as a form of ‘location-specific social capital’, which enables people to depend on their networks regarding in-kind services or job searches (De Souza Briggs, 1998; Granovetter, 1995).

Social networks for low-income households are particularly distinctive. Spring et al. (2017) shows that kin proximity is an important driver of residential mobility and neighbourhood choice for those from low-income households. Low-income households tend to live closer to relatives and friends for companionship or practical help. On average, social networks are not only more localised, but also more exhausted than those of higher-income people (Fischer, 1982; Kadushin and Jones, 1992). De Souza Briggs et al. (2010) referred to social networks of low-income households as the ‘weakness of strong ties’, based on Stack’s (1974) findings of a dysfunctional ‘culture of poverty’ among the persistently poor. Staying close to kinship or friendship is central to the lives of many poor families for mixed reasons. Strong social networks may provide emotional and practical support, but they typically come with enormous obligations as well. The survival of poor families demands the sacrifice of economic mobility.

Several studies regarding low-income assisted renters look into mobility patterns focusing on social networks. Many housing voucher holders had weak attachments to their new neighbourhoods, and moved in close proximity to their old neighbourhoods (Varady and Walker, 2000; Varady et al., 2001). Findings that come from the Moving to Opportunity (MTO) experiment show that social networks persisted at the core of most participants’ lives regardless of relocation, and influenced how they structured their daily lives. The MTO demonstration expected participants to relocate to quality neighbourhoods and form relationships with their higher-income neighbours, thus building the bridging social networks that leverage information used to access employment and other resources (Curley, 2009). However, most MTO participants who initially gained access to vouchers gave up the opportunity to use the voucher, thereafter moving back to the original neighbourhoods or to similarly distressed neighbourhoods (De Souza Briggs et al., 2010). The MTO Interim Evaluation (Orr et al., 2003) and Cove et al. (2008) find that a few MTO movers converted their new housing locations into valuable new social capital, while many reported regular contact with the previous neighbourhood and formed limited relationships with their new neighbours.

De Souza Briggs et al. (2010) identified major types of neighbourhood mobility and social relations. The most common pattern was households moving to neighbourhoods where relatives lived. These relatives played a central role in daily life, and households received companionship and vital practical support. Some households were transplanted into a new neighbourhood while weakening social ties to the place they left behind. Some households moved to a new neighbourhood by putting distance between themselves and their constantly needy relatives or friends. Although these patterns well explain the relationship between social networks and residential mobility, it is yet unexplored whether seeking assistance for costly services is a primary reason for low-income households to live in close proximity to their social networks, and whether living in areas with accessible social services prevents low-income households from relocating to their initial neighbourhoods.

Access to social services

Access to social services may play a critical role in the residential mobility of low-income assisted renters, especially as they move away from their social networks. A deeper understanding of the importance of social networks calls into question whether voucher households would have stayed in a new neighbourhood if they had had access to social services. If they mainly seek social networks for the purpose of pooling resources for childcare, transportation assistance and other human services, assistance from a local social service agency might be a substitute for practical help from social networks. Edin and Lein (1997) found that 77 per cent of welfare-reliant single mothers depended on family members or friends to reduce a substantial gap between their welfare income and living expenses, and 31 per cent of the sample also received additional cash and in-kind assistance from public or private agencies outside the welfare system.

Two works by Scott Allard (2009, 2017) place emphasis on the geography of social services. Since the provision of social services most often occurs through a local government or non-profit organisation, the location of social service providers varies by the government’s discretion over the administration, capacity and interest of the non-profit sector (Allard, 2009). It is critical for low-income households to have adequate access to social service providers because human service programmes often require a person’s visit to their office or attendance at sessions. Having no agencies nearby is likely to lead to low-income households being disconnected from a service-based system of assistance. Despite the importance of the availability and accessibility of social services, their spatial distribution has mismatched the geography of poverty and need. When the number of clients served by agencies and the number of poor persons were taken into account together, social services were inaccessible in high-poverty neighbourhoods and in predominantly minority neighbourhoods (Allard, 2009).

A more serious problem lies in that recent growing suburban poverty may result in the exacerbation of mismatches between social service providers and demand in the future. Emerging slowly from the inner-ring suburbs, suburban poverty has shifted spatial distribution of demographic characteristics across large metropolitan areas since the 1970s (Cooke and Marchant, 2006; Madden, 2003). While the number of poor suburban residents doubled between 2000 and 2012, over two times faster than in cities (Kneebone and Berube, 2013), Allard (2017) shows that the responsiveness of safety net programmes has not kept pace with the significant rise of poverty in suburbs, and that the majority of non-profit social service activities take place in the urban centre. Median non-profit human service expenditures per low-income person remained much higher in central-city areas compared with suburban counties between 2000 and 2010. Low levels of per-capita funding in suburbs are likely to result from limitations on the availability of adequate facilities for social service programmes and the capability to which agencies achieve economies of scale for service delivery and fee revenue.

In sum, low-income households are likely to live closer to their social networks not only willingly for emotional or practical help, but also unwillingly for family obligations. Strong social connections often deter residential mobility beyond reach of the network, even if housing vouchers are provided. In the MTO experiment, using a voucher with geographical restriction was ineffective in the formation of new social networks in less advantaged neighbourhoods. Many participants maintained existing social networks, and after a short-lived lease-up, moved back to their original neighbourhoods or to neighbourhoods where family members or friends lived. However, we know little about the effects of social resources and access to social services on residential mobility. This article contributes to this strand of literature by emphasising the relationship between social networks, access to social services and the dynamics of mobility, accounting for differences between moves to the original neighbourhoods and elsewhere.

Data and methods

My analyses address two research questions related to the links between social networks, access to social services and residential mobility: What aspects of social networks bind low-income households to their original neighbourhoods? And how do both social networks in an original neighbourhood and access to social services in a new neighbourhood influence the subsequent residential choice? I rely on data from the Moving to Opportunity (MTO) experiment. Across five US metropolitan areas – Baltimore, Boston, Chicago, Los Angeles and New York – over 4600 public housing residents in high-poverty neighbourhoods were randomly assigned to one of three groups. The Experimental group received a geographically-restricted voucher that could be used in neighbourhoods with poverty rates less than 10 per cent. The Section 8 group received a voucher without any restrictions. The Control group received nothing. This randomised experimental research design eliminates potential selection bias associated with residential location choices.

To address the propensity for moving back to a baseline neighbourhood, I explore various descriptive metrics of neighbourhood mobility, focusing on the first two residential moves of MTO participants. Drawing upon data from the MTO residential spell database and Census TIGER shapefiles, I track all MTO participants as to whether they leave their baseline neighbourhood and whether they move back to their baseline neighbourhood. I categorise their residential mobility into three patterns: (1) moved elsewhere and re-entered the baseline neighbourhood, (2) moved to one of the adjacent neighbourhoods that surround the baseline neighbourhood, and (3) moved to a neighbourhood other than the baseline neighbourhood or its nearby neighbourhoods.

To differentiate various aspects of social networks with respect to residential mobility, I consider two dimensions of social networks: kinship/friendship, and social resources. In empirical studies, various measures of social networks have been employed. Duration of residency is used in early studies as an indirect proxy of social networks (Lansing and Mueller, 1967; Speare, 1970). Several studies employ direct measures of social networks, such as the number of relatives and friends (Landale and Guest, 1985; Myers, 2000) and the frequency of neighbours’ visits (Connerly, 1986). The presence of family members and friends shows that strong social networks exist when living in the same neighbourhood with friends and family members. On the other hand, Dawkins (2006) employs measures of social resources. Social resources capture a different aspect of social networks. Provided that kinship and friendship represent the density of social networks, social resources measure how such networks would practically benefit the household. Having a separate measure of social resources isolates the marginal effect of the dependence on social networks for costly services, distinct from the effect of social networks shaped by emotional support or family obligations.

To examine the impact of social networks and access to social services, I employ a logistic regression model and a multinomial logistic regression model. While the logit model characterises residential mobility behaviour, the multinomial logit model examines the propensity for mobility in moving back to the baseline neighbourhood over the propensity for mobility in staying away for the baseline neighbourhood. Alternatives in these models do not depend on each other, and there is no possible additional alternative which maintains the Independence of Irrelevant Alternatives (IIA) and enables regressions to produce valid estimators. In the first move model, the dependent variable of the logit model is equal to one if the household moved from the first residence. The dependent variable of the multinomial logit model is defined by the distance to the initial neighbourhood: those who (1) moved within the baseline neighbourhood, (2) moved to one of the adjacent neighbourhoods that surround the baseline neighbourhood, and (3) moved to another neighbourhood.

In the second move model, the dependent variable of the logit model is equal to one if the household moved from the second residence, while the following MTO participants are excluded: (1) those who never moved from their first residence, (2) those who relocated within the baseline neighbourhood, and (3) those who made a second move after spending more than two years in the second neighbourhood. The cost of moving is a significant factor in the residential mobility decision (Weinberg et al., 1981), and the frequency of mobility declines with the length of residence particularly for renters (Morrison, 1971; Quigley and Weinberg, 1977; Speare, 1970). Movers who stay longer in a new neighbourhood may not be impacted by their previous social networks when they decide to move. Since most households who made a subsequent move after the first move spent around two years in their second neighbourhoods, I focus on the examination of factors that influence the propensity of frequent mobility. In the multinomial logit model, the dependent variable is determined similarly: those who (1) moved back to the baseline neighbourhood, (2) moved to one of the adjacent neighbourhoods that surround the baseline neighbourhood, and (3) moved to another neighbourhood.

All variables employed in the analysis are displayed in Table 1.

Table 1.

Variable descriptions for the logit and multinomial logit regression analyses.

Variable	Definition
Dependent variables
First move	1 = moved from the first residence
Within baseline neighbourhood	1 = moved within the baseline neighbourhood
Adjacent neighbourhood	1 = moved to an adjacent neighbourhood to the baseline neighbourhood
Another neighbourhood	1 = moved to another neighbourhood
Second move	1 = moved from the second residence
Baseline neighbourhood	1 = moved back to the baseline neighbourhood
Adjacent neighbourhood	1 = moved to an adjacent neighbourhood to the baseline neighbourhood
Another neighbourhood	1 = moved to another neighbourhood
Independent variables
Treatment effects
Experimental group	1 = Experiment group
Section 8 group	1 = Section 8 group
Metro controls
Baltimore	1 = lived in Baltimore
Boston	1 = lived in Boston
Los Angeles	1 = lived in Los Angeles
New York	1 = lived in New York
Household characteristics
Vehicle access	1 = the household owned a usable car, van or truck
Income	Household income (US$1,000)
No. children in HH	No. children in the household
Age of HH head	Age of the household head
HH head Black	1 = household head is Black
HH head Hispanic	1 = household head is Hispanic
HH head female	1 = household head is female
HH head married	1 = household head is married
HH head has high school degree	1 = household head has high school degree or GED
HH head employed	1 = household head is employed
Social networks
Kinship/friendship in the baseline neighbourhood	1 = HH has family member(s) or friend(s) living in a baseline neighbourhood
Social resources score in the baseline neighbourhood	Standardised score based on item-response model estimation of social resources
Neighbourhood characteristics
No. children and youth services providers	No. children and youth services providers
Average children and youth services expenditure	Average children and youth services expenditure (US$1,000)
Poverty rate	Poverty rate
Distance to central city	Distance to central city (miles)
Transit access index	HUD FHEA transit access index
Nearby the baseline neighbourhood	1 = the second neighbourhood is adjacent to the baseline neighbourhood

All household-level data come from the MTO baseline survey conducted between 1994 and 1998. Experimental and Section 8 groups are included to examine the treatment effect. An indicator of the household’s metropolitan location is included, with Chicago omitted as the reference category. Household characteristics include vehicle access, income and the number of children in the household. The measure of vehicle access is equal to one if anyone in the household owned a car that runs. Since 99 per cent of the total MTO households had one or more children, those without children are excluded from the analyses. Characteristics of the household head include age, race and ethnicity, gender, marital status, education, and employment status.

With regard to social networks in the original neighbourhood, kinship and friendship are defined as the presence of family members and friends living in the baseline neighbourhood identified by census tract code. To measure social resources, I compute a standardised score of social resources using the Item Response Theory estimation. Sampson et al. (1997) measured social cohesion and informal social control based on a range of conceptually related questions. Similarly, social resources can be quantified by a collection of responses in the MTO survey. I consider three questions that are converted into binary responses: (1) a household borrows things from a neighbour, (2) a household rides with neighbours or carpools, and (3) a household receives childcare from a relative or neighbour. The higher the score, the greater the dependence on social networks for costly services.

Two variables measure access to social services. The number of social service providers captures whether a person located in a given neighbourhood is physically nearby an agency that offers relevant social services. Average social service expenditure takes the size of social services providers into account, assuming that spending more on social services would lead to greater accessibility for those in need. Relying on data from the National Center for Charitable Statistics regarding the 2001 Internal Revenue Service (IRS) filings of non-profit social service organisations, at the census tract level I compute a weighted average of the surrounding zip code’s number and average expenditure of social service providers. HUD zip code crosswalk files allocate census tracts to zip codes, enabling researchers to consider census tract-level data using data available only at the zip code level, with the use of the weighted average method based on the ratio of addresses in the zip to the total number of addresses in the entire zip. Referring to Allard (2017), I include non-profit organisations with annual revenues under US$10 million only in the calculation, assuming that they are more likely to provide services in the neighbourhoods where they file their IRS forms, whereas large organisations tend to be national headquarters or advocacy groups.²

To be comparable with social services that often provided through social networks, I first considered selected non-profits operating in one of the following fields defined by the National Taxonomy of Exempt Entities (NTEE): children and youth services; family services; personal social services; emergency assistance; and residential care and adult day programmes. Table 2 presents the average number and expenditure of social services providers by classification. In all census tracts where MTO households lived during their first two residencies, children and youth services were the most available and largest services. On average, more than 1.5 children and youth services providers were available in the census tract, in contrast to fewer than 0.4 agencies for other services. Despite a deep decline from the first neighbourhood to the second neighbourhood, average children and youth services expenditure showed the highest among all of the selected social services. Because households without children have been excluded from the analyses, I include the number and average expenditure of children and youth services providers as salient measurements of access to social services in order to examine whether social service provision would be an appropriate substitute for social networks with regard to residential mobility decisions.

Table 2.

Number and average expenditure of social service providers by classification.

Social services	No. social services providers		Average social services expenditure
	First move	Second move	First move	Second move
Children & youth services	1.65	1.56	US$1,374,623	US$626,930
Family services	0.39	0.26	US$308,889	US$400,518
Personal social services	0.03	0.04	US$12,013	US$5,551
Emergency assistance	0.09	0.09	US$11,977	US$11,729
Residential care & adult dayprogrammes	0.31	0.28	US$159,470	US$201,786
Total/average	2.47	2.23	US$1,011,536	US$732,203

For additional neighbourhood characteristics computed at the census tract level, I include measures of poverty rate and distance to the central city from the 2000 Census. The transit access index provided by the HUD Fair Housing Equity Assessment database captures the availability of public transportation.³ To explore the effect of distance between the baseline neighbourhood and the second neighbourhood on residential mobility decisions, a dummy variable indicating that the second neighbourhood is adjacent to the baseline neighbourhood is included in the second move model.

Social networks, access to social services and the dynamics of residential mobility

Figure 1 describes neighbourhood mobility patterns shaped as dependent variables in logit models. About 88 per cent of the sample moved from the baseline neighbourhood. Among these movers, about 9 per cent moved within the baseline neighbourhood, about 6 per cent moved to one of neighbourhoods adjacent to the baseline neighbourhood and about 85 per cent moved to another neighbourhood. In the second residence, households who previously moved closer to the baseline neighbourhood are more likely to stay (37 per cent) than households who previously moved to neighbourhoods that were neither the baseline neighbourhood nor nearby the baseline neighbourhood (23 per cent). Movers who lived in a neighbourhood adjacent to the baseline neighbourhood tended more than those who lived further from the baseline neighbourhood to move back to the baseline neighbourhood (22 vs 7 per cent) or to a nearby neighbourhood (19 vs 4 per cent).

Figure 1.

Descriptive statistics for dependent variables used in the logit and multinomial logit regression analyses, MTO participants.

Table 3 presents descriptive statistics for variables used in the logit models. On average, MTO participants were predominantly unmarried, unemployed minority women with three children, with an income of less than US$10,000. About 68 per cent of the households had family members or friends living in the baseline neighbourhood. The average social resources score was 0.81, indicating that most MTO participants tend to use their social networks for costly services. Comparing the census tract characteristics between the first and second neighbourhoods of MTO participants, I find that after the first move, households are likely to live in neighbourhoods with a smaller number and size of children and youth services providers, lower poverty rates, longer distance to the central city and less accessibility to public transportation.

Table 3.

Descriptive statistics for variables used in the logit and multinomial logit regression analyses, MTO baseline survey.

Variable description	Mean	Std dev	Min	Max
Treatment effects
Experimental group	0.39	0.49	0	1
Section 8 group	0.29	0.46	0	1
Metro controls
Baltimore	0.14	0.35	0	1
Boston	0.21	0.41	0	1
Los Angeles	0.22	0.42	0	1
New York	0.23	0.42	0	1
Household characteristics
Vehicle access	0.18	0.39	0	1
Income	9.30	5.50	0	133
Age of HH head	33.67	9.35	17	87
HH head Black	0.63	0.48	0	1
HH head Hispanic	0.31	0.46	0	1
HH head female	0.93	0.25	0	1
HH head married	0.11	0.32	0	1
No. children in HH	2.87	1.64	1	15
HH head has high school degree	0.59	0.49	0	1
HH head employed	0.27	0.45	0	1
Social networks
Kinship/friendship in the baseline neighbourhood	0.68	0.47	0	1
Social resources score in the baseline neighbourhood	0.81	0.18	0	1
Neighbourhood characteristics
First neighbourhood
No. children and youth services providers	1.65	1.51	0	7
Average children and youth services expenditure	1376.25	1467.78	0	4974
Poverty rate	49.19	14.75	0	93
Distance to central city	4.10	2.82	0	19
Transit access index	74.92	35.11	0	100
Second neighbourhood
No. children and youth services providers	1.56	1.68	0	7
Average children and youth services expenditure	627.03	933.84	0	4974
Poverty rate	26.57	14.66	0	91
Distance to central city	6.95	5.75	0	55
Transit access index	69.49	31.40	0	100
Nearby the baseline neighbourhood	0.08	0.28	0	1

Table 4 reports odds ratios for the logit coefficients from the first move and second move models. The goodness of fit is somewhat low, but is comparable with other studies of residential mobility (Dawkins, 2006; Oh, 2003). Results show that those in the Experimental and Section 8 groups show higher probabilities of moving, compared with the control group. For metropolitan controls, those in Boston, Los Angeles and New York are less likely to move than those in Chicago. Controlling for other determinants of residential mobility, these effects likely capture metropolitan differences in housing and labour markets, transportation systems and other amenities. A few household characteristics are associated with residential mobility. In the second move model, younger households and households with more children are more likely to move.

Table 4.

Logit results of the first move and second move models, odds ratio.

Variable	First move	Second move
Treatment effects
Experimental group	1.992***	1.809***
Section 8 group	2.446***	2.193***
Metro controls
Baltimore	0.813	0.771
Boston	0.277***	0.792
LA	0.205***	0.521***
NYC	0.077***	0.407***
Household characteristics
Vehicle access	1.143	1.118
Income	0.989	0.982
Age of HH head	0.952***	0.954***
HH head Black	1.280	1.511*
HH head Hispanic	1.134	1.129
HH head female	1.150	1.240
HH head married	1.302	0.942
No. children in HH	1.026	1.126***
HH head has high school degree	1.007	1.027
HH head employed	0.950	1.043
Social networks
Kinship/friendship in the baseline neighbourhood	1.047	0.823
Social resources score in the baseline neighbourhood	1.061	0.717
Neighbourhood characteristics
No. children and youth services providers	0.935	0.858***
Average children and youth services expenditure	1.000**	1.000*
Poverty rate	0.985***	0.986***
Distance to central city	0.989	1.013
Transit access index	1.003	1.004*
Nearby the baseline neighbourhood		0.590**
Constant	156.692***	3.799**
Number of observations	4083	1432
Wald chi-square	386.87***	159.63***
Pseudo R²	0.169	0.100

Notes: *** p < 0.01, ** p < 0.05, * p < 0.1.

While none of the social networks variables are statistically associated with residential mobility, the effect of distance between the first and second neighbourhoods is significant at 5 per cent. The odds of moving are 41 per cent lower for those who lived in neighbourhoods adjacent to the baseline neighbourhood, compared with those who lived further from the baseline neighbourhood. Despite their insignificance, the effects of social networks in the first move model are the opposite of the effects in the second move model. While holding all other factors constant, MTO participants who have stronger social networks in their first neighbourhood are more likely to move from the first residence, but less likely to move from their second residence.

Access to social services is significant for second movers. With one additional children and youth services provider, the odds of moving are 14 per cent lower for those who moved once from their baseline neighbourhood. This finding might be affected by the geography of social services. Because most social services are clustered in the urban centres (Allard, 2009, 2017), the effect of access to social services suggests that high accessibility of social services near the baseline neighbourhood decreases the odds of moving. Simultaneously, controlling for distance between the first and second neighbourhoods implies that households tend to stay in the second neighbourhood even further from the first neighbourhood if there is an adequate number of agencies that provide social services.

Reported in Table 5, multinomial logit results show relative risk ratios of factors that affect the likelihood of moving back to the baseline neighbourhood or moving to a neighbourhood adjacent to the baseline neighbourhood. In the first move model, those in the Experimental and Section 8 groups are less likely to move within a baseline neighbourhood or to enter one of the adjacent neighbourhoods that surround the baseline neighbourhood, compared with moving to another neighbourhood. The second move model presents the influence on several metropolitan controls, which indicates that those living in Boston, Los Angeles or New York are more likely to move in closer proximity to a baseline neighbourhood. Among all household characteristics, households whose household head is older and has a high school degree contribute the most to mobility towards the baseline neighbourhood in the second move model.

Table 5.

Multinomial logit results of the first move and second move models, relative risk ratio.

	First move		Second move
Variable	Within baselineneighbourhood	Nearby baselineneighbourhood	Baseline neighbourhood	Nearby baselineneighbourhood
Treatment effects
Experimental group	0.424***	0.435***	0.911	0.857
Section 8 group	0.391***	0.508***	1.786	0.507
Metro controls
Baltimore	1.093	1.108	1.557	3.387*
Boston	4.561***	1.543	8.116***	3.592
LA	1.682*	2.120**	4.346***	4.547**
NYC	1.522	0.843	4.165**	1.823
Household characteristics
Vehicle access	0.467***	0.894	1.084	0.974
Income	1.035***	1.013	0.969	0.999
Age of HH head	1.011	1.036***	1.039**	1.025
HH head Black	0.991	1.492	1.797	0.497
HH head Hispanic	1.317	1.549*	0.630	0.348
HH head female	0.888	0.972	0.644	0.570
HH head married	1.133	0.746	0.534	1.724
No. children in HH	0.997	1.016	1.078	0.952
HH head has high school degree	0.910	0.938	2.156**	1.614
HH head employed	0.843	1.022	0.722	1.467
Social networks
Kinship/friendship in the baseline neighbourhood	1.192	1.083	2.671***	2.443**
Social resources score in the baseline neighbourhood	0.914	1.830	3.956*	1.238
Neighbourhood characteristics
No. children and youth services providers	1.065	1.214***	0.766**	1.025
Average children and youth services expenditure	1.000	1.000	1.000	1.000
Poverty rate	1.028***	1.002	0.986	1.002
Distance to central city	1.048	0.908	0.949	1.098***
Transit access index	1.002	1.001	1.010	1.029**
Nearby the baseline neighbourhood			3.430**	2.911*
Constant	0.011***	0.011***	0.001***	0.001***
Number of observations	3603		696
Wald chi-square	264.80***		120.96***
Pseudo R2	0.070		0.151

Notes: *** p < 0.01, ** p < 0.05, * p < 0.1.

The presence of family members or friends is highly associated with the likelihood of moving in the second move, compared with the likelihood of moving to another neighbourhood. For those with a strong kinship or friendship in the baseline neighbourhood, the relative risk of moving back to the baseline neighbourhood is 167 per cent higher, whereas the risk of moving to a neighbourhood adjacent to the baseline neighbourhood is 144 per cent higher. Holding other variables constant, the social resources score in the original neighbourhood has a significant effect at a significance level of 0.1 in the second move model. While the effect appears weak, past dependence on social networks for costly services tends to influence movers’ decision to move back. This finding may extend the literature by isolating the impact of social resources from the impacts of kinship and friendship.

Access to social services has a significant, negative effect on moving back to the first neighbourhood in the second move model. With one additional children and youth services provider, the relative risk of moving is 23 per cent lower for those who have moved back to the baseline neighbourhood. Linking with the finding from the logit model suggests that if households have better access to social services, they tend to either stay in the current neighbourhood or, when they move, choose a distant neighbourhood over the baseline neighbourhood. Thus, exposure to better access to social services may prevent movers from relocating to the original neighbourhood. The results of distance moved are also consistent with the results of access to social services. The relative risk of moving back to the baseline neighbourhood is more than two times higher, compared with moving to distant neighbourhoods.

To further examine whether assistance from social service agencies may substitute for practical help from one’s social networks, post simulations of the multinomial logit models are conducted. I include the interaction terms of social resources score, the number of social service providers and distance moved in the second move model (not reported here). Figure 2 displays the predicted probabilities of moving back to the first neighbourhood for different levels of social networks and access to social services, based on distance between the first and second neighbourhoods. The results suggest that MTO movers with higher social resources scores tend to move back to the baseline neighbourhood. For those with a zero social resources score, if their second neighbourhood is adjacent to their first neighbourhood, 15 per cent of movers move back to the first neighbourhood when there is one children and youth services provider available, whereas only one per cent from neighbourhoods with five children and youth services providers move back. Although the difference declines as the social resources score inclines, households with greater access to social services are more likely to move elsewhere other than the baseline neighbourhood. This result implies that, to a certain extent, access to social services might prevent short distance movers from moving back to the original neighbourhood. However, local social services might not effectively replace practical help from social networks for short distance movers who had largely depended on their social networks. If households had completely depended on their social networks, 19 per cent of households from neighbourhoods with one children and youth services provider move back to the baseline neighbourhood, while 51 per cent from neighbourhoods with five children and youth services providers move back.

Figure 2.

Predicted probabilities of moving back to a baseline neighbourhood by social resources score and access to children and youth services.

For those whose second neighbourhood is far from the baseline neighbourhood, households are more likely to move back to the first neighbourhood if the number of social services providers in the second neighbourhood is lower, regardless of the extent of social network dependency. The disparity in the likelihood of moving back inclines with social resources score between those living in neighbourhoods with higher access to social services and those living in neighbourhoods with lower access to social services. For movers with the highest levels of social resources score, living in neighbourhoods with one children and youth services provider results in 13 per cent moving back, and living in neighbourhoods with five children and youth services providers results in 7 per cent moving back. On the other hand, for movers with the lowest levels of social resources score, a modest percentage (3%) or zero per cent of households move back to the baseline neighbourhoods, which possibly reflects the fact that access to social services could be a substitute for social networks.

Conclusion

This article examines the links between social networks, access to social services and residential mobility for MTO participants. Emphasis is placed on the likelihood of moving back to the original neighbourhood after random assignment, and its influential factors, focusing on social networks and access to social services. Estimates from logit models suggest that social networks influence subsequent mobility decisions. When households make a move, movers tend to choose the baseline neighbourhood over other location options if they have strong kinship or friendship in the baseline neighbourhood. I also find evidence of access to social services as one of the most essential neighbourhood characteristics that affect the dynamics of residential mobility. If MTO households moved to areas in closer proximity to local social service agencies, they might stay in these neighbourhoods, depending on the extent of dependence on social networks. Accounting for social networks, access to social services and distance moved together, it is likely that access to social services is a substitute for social networks for its role in pooling resources for costly services.

My findings point to the importance of access to social services in helping low-income households determine a residential location with voucher use. In consideration of policies designed to enhance mobility towards high opportunity, it is critical to connect households who rely on their social networks as social resources with local social services that would play a substitute role. If low-income assisted renters are able to move to high-opportunity neighbourhoods with adequate access to social services, in particular children and youth services, they are more likely to settle in the new neighbourhood. In this effort, the geography of social services should be of more concern and discussed in the realm of housing policy. Expanding my focus on access to social services, more empirical work is necessary to determine what types of assistance from local social services providers help low-income households move towards high-opportunity neighbourhoods and how these shape residential trajectories with respect to voucher use.

My findings also suggest that housing search assistance may be more effective in leading low-income households to their desirable neighbourhoods with the provision of information and available resources. Moving back to the original neighbourhood from the neighbourhood with a lack of access to social services implies that the return move might be a response to a failure to find a substitute for support from social networks in a new neighbourhood. Expansion of housing search assistance with the provision of information about social services options across the metropolitan area may help to realise greater improvements in neighbourhood outcomes, thereby preventing voucher households from relocation to the original neighbourhood. This policy implication extends the findings by Shroder (2002) that the intensity of housing counselling services increased the likelihood of lease-up in the MTO programme. While providing housing search counselling may be expensive, expanding such assistance to account for the geography of social service accessibility may have more significant impacts on poverty de-concentration than residential mobility programmes with various types of constraints on the use of housing subsidy. Future research should test whether spatial dispersion of social services would have beneficial impacts on the spatial dispersion of voucher households and the changing geography of metropolitan poverty.

Footnotes

Acknowledgements

The author would like to thank Drs Casey Dawkins, Gerrit Knaap, Willow Lung-Amam, Ariel Bierbaum, Peter Reuter (University of Maryland, USA) and Rolf Pendall (University of Illinois at Urbana-Champaign, USA) for their valuable feedback on earlier versions of this article. The author would also like to thank the editors and anonymous reviewers who provided several suggestions which greatly improved the article.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

This work was supported by the US Department of Housing and Urban Development (grant number FR-5415-N-24A).

Notes

ORCID iD

Jae Sik Jeon

References

Allard

(2009) Out of Reach: Place, Poverty, and the New American Welfare State. New Haven, CT: Yale University Press.

Allard

(2017) Places in Need: The Changing Geography of Poverty. New York, NY: Russel Sage Foundation.

Clark

WAV

(2005) Intervening in the residential mobility process: Neighborhood outcomes for low-income populations. Proceedings of the National Academy of Sciences 102(43): 15,307–15,312.

Clark

WAV

Withers

(1999) Changing jobs and changing houses: Mobility outcomes of employment transitions. Journal of Regional Science 39(4): 653–673.

Connerly

(1986) The impact of neighborhood social relations on prospective mobility. Social Science Quarterly 67(1): 187–194.

Cooke

Marchant

(2006) The changing intrametropolitan location of high-poverty neighborhoods in the U.S, 1990–2000. Urban Studies 43(11): 1971–1989.

Coulton

Theodos

Turner

(2012) Residential mobility and neighborhood change: Real neighborhoods under the microscope. Cityscape 14(3): 55–89.

Cove

Turner

De Souza Briggs

, et al. (2008) Can Escaping from Poor Neighborhoods Increase Employment and Earnings?Washington, DC: The Urban Institute.

Curley

(2009) Draining or gaining? The social networks of public housing movers in Boston. Journal of Social and Personal Relationships 26(2–3): 227–247.

10.

DaVanzo

(1981) Repeat migration, information costs, and location-specific capital. Population and Environment 4(1): 45–73.

11.

Dawkins

(2006) Are social networks the ties that bind families to neighborhoods? Housing Studies 21(6): 867–881.

12.

Dawkins

Jeon

Pendall

(2015) Vehicle access and exposure to neighborhood poverty: Evidence from the Moving to Opportunity program. Journal of Regional Science 55(5): 687–707.

13.

De Souza Briggs

(1998) Brown kids in white suburbs: Housing mobility and the many faces of social capital. Housing Policy Debate 9(1): 177–221.

14.

De Souza Briggs

Popkin

Goering

(2010) Moving to Opportunity: The Story of an American Experiment to Fight Ghetto Poverty. Oxford: Oxford University Press.

15.

Edin

Lein

(1997) Making Ends Meet: How Single Mothers Survive Welfare and Low-Wage Work. New York, NY: Russell Sage Foundation.

16.

Fischer

(1982) To Dwell Among Friends: Personal Networks in Town and City. Chicago, IL: University of Chicago Press.

17.

Galster

Killen

(1995) The geography of metropolitan opportunity: A reconnaissance and conceptual framework. Housing Policy Debate 6: 7–43.

18.

Goetz

Chapple

(2010) You gotta move: Advancing the debate on the record of dispersal. Housing Policy Debate 20(2): 209–236.

19.

Gramlich

Laren

Sealand

(1992) Moving into and out of poor urban areas. Journal of Policy Analysis and Management 11(2): 273–287.

20.

Granovetter

(1973) The strength of weak ties. American Journal of Sociology 78(6): 1360–1380.

21.

Granovetter

(1995) Getting a Job: A Study of Contacts and Careers. Chicago, IL: University of Chicago Press.

22.

Kadushin

Jones

(1992) Social networks and urban neighborhoods in New York City. City and Society 6(1): 58–75.

23.

Kneebone

Berube

(2013) Confronting Suburban Poverty in America. Washington, DC: Brookings Institution Press.

24.

Landale

Guest

(1985) Constraints, satisfaction and residential mobility: Speare’s model reconsidered. Demography 22(2): 199–222.

25.

Lansing

Mueller

(1967) The Geographic Mobility of Labor. Ann Arbor, MI: Institute for Social Research.

26.

Madden

(2003) The changing spatial concentration of income and poverty among suburbs of large U.S. metropolitan areas. Urban Studies 40(3): 481–503.

27.

Morrison

(1971) Chronic movers and the future redistribution of population: A longitudinal analysis. Demography 8(2): 171–184.

28.

Myers

(2000) The impact of religious involvement on migration. Social Forces 79(2): 755–783.

29.

(2003) Social bonds and the migration intentions of elderly urban residents: The mediating effect of residential satisfaction. Population Research and Policy Review 22(2): 127–146.

30.

Orr

Feins

Jacob

, et al. (2003) Moving to Opportunity: Interim Impacts Evaluation. Washington, DC: US Department of Housing and Urban Development.

31.

Quigley

Weinberg

(1977) Intra-urban residential mobility: A review and synthesis. International Regional Scence Review 2: 41–66.

32.

Rosenblatt

DeLuca

(2012) We don’t live outside, we live in here: Neighborhood and residential mobility decisions among low-income families. City and Community 11(3): 254–284.

33.

Sampson

(1999) What “community” supplies. In: Ferguson

Dickens

(eds) Community Development. Washington, DC: Brookings Institution Press.

34.

Sampson

Raudenbush

Felton

(1997) Neighborhoods and violent crime: A multilevel study of collective efficacy. Science 277(15): 918–924.

35.

Sanbonmatsu

Ludwig

Katz

, et al. (2011) Moving to Opportunity for Fair Housing Demonstration Program: Final Impacts Evaluation. Washington, DC: Office of Policy Development and Research, U.S. Department of Housing and Urban Development.

36.

Shroder

(2002) Locational constraint, housing counseling, and successful lease-up in a randomized housing voucher experiment. Journal of Urban Economics 51: 315–338.

37.

South

Crowder

Chavez

(2005) Exiting and entering high-poverty neighborhoods: Latinos, Blacks, and Anglos compared. Social Forces 84(2): 873–900.

38.

Speare

(1970) Home ownership, life cycle stage, and residential mobility. Demography 7(4): 449–458.

39.

Speare

(1974) Residential satisfaction as an intervening variable in residential mobility. Demography 11(2): 173–188.

40.

Spring

Ackert

Crowder

, et al. (2017) Influence of proximity to kin on residential mobility and destination choice: Examining local movers in metropolitan areas. Demography 54(4): 1277–1304.

41.

Stack

(1974) All Our Kin: Strategies for Survival in a Black Community. New York, NY: Harper and Row.

42.

US Department of Housing and Urban Development (HUD) (2001) Guidance for HUD Field Offices to Conduct Annual Section 8 Management Assessment Program (SEMAP) Assessments. Washington, DC: HUD. Available at: https://www.hud.gov/sites/documents/PIH2001-6.PDF (accessed 13 May 2019).

43.

US Department of Housing and Urban Development (HUD) (2012) FHEA Data Documentation (Draft). Washington, DC: HUD.

44.

Varady

Walker

(2000) Vouchering out distressed subsidized developments: Does moving lead to improvements in housing and neighborhood conditions? Housing Policy Debate 11(1): 115–162.

45.

Varady

Walker

Wang

(2001) Voucher recipient achievement of improved housing conditions in the US: Do moving distance and relocation services matter? Urban Studies 38(8): 1273–1304.

46.

Weinberg

Friedman

Mayo

(1981) Intraurban residential mobility: The role of transactions costs, market imperfections, and household disequilibrium. Journal of Urban Economics 9: 332–348.