On the Individual and Social Determinants of Neighbourhood Satisfaction and Attachment

Abstract

Corrado G., Corrado L. and Santoro E. On the individual and social determinants of neighbourhood satisfaction and attachment, Regional Studies. This paper explores the determinants of perceived neighbourhood quality, emphasizing the nexus between two distinct but correlated dimensions along which individuals evaluate their neighbourhoods: satisfaction, traditionally viewed as a cognitive judgement based on the attainment of some standard or aspiration; and attachment, which relates to an affective evaluation. A bivariate probit model is estimated using data from the British Household Panel Survey (BHPS). Significant residual correlation is estimated, which suggests that satisfaction and attachment tap both cognitive and affective aspects of individual attitudes. In addition, average satisfaction and attachment at the neighbourhood level emerge as crucial determinants of individual responses.

Corrado G., Corrado L. and Santoro E. 社区满意度与归属感的个体以及社会决定因素，区域研究。本文探讨了影响对社区品质感受的决定要素，强调了个体如何在相异却相互关联的两个不同层面上对社区进行评估：满意度，传统上被认为是一种基于获得某些标准或者是需求的认知判断；而归属感，是与情感评价相关。在英国住房面板数据基础上我们对 二元常态 probit 模型（bivariate probit）进行了预测。显著的冗余相关性表明满意度以及归属感与个体态度中的认知与情感因素同时相关。同时，社区层面的平均满意度以及归属感是影响个体反映的决定性要素。

社区 满意度以及所属感 社会互动 二元常态 probit

Corrado G., Corrado L. et Santoro E. Sur les déterminants individuels et sociaux de la satisfaction et de l'attachement locaux, Regional Studies. Cet article cherche à examiner les idées que l'on se fait de la qualité du voisinage local, en soulignant la connexion entre deux dimensions, à la fois différentes et en corrélation, tout le long desquels les individus évaluent leur vosinage: la satisfaction, considérée traditonnellement comme un jugement cognitif fondé sur l'acquisition de quelque norme ou aspiration; et l'attachement, qui se rapporte à une évaluation affective. A partir des données provenant de la British Household Panel Survey (une enquête à échantillon constant sur les ménages britanniques), on évalue un modèle probit à deux variables. On évalue une corrélation résiduelle importante, ce qui laisse supposer que la satisfaction et l'attachement captent à la fois les aspects cognitifs et affectifs des opinions individuelles. En outre, il s'avère que la satisfaction et l'attachement moyens sont au niveau local des déterminants préalables des réponses individuelles.

Voisinage Satisfaction et Attachement Interactions sociales Modèle probit à deux variables

Corrado G., Corrado L. und Santoro E. Individuelle und soziale Determinanten für die Zufriedenheit und Verbundenheit in Wohnvierteln, Regional Studies. In diesem Beitrag untersuchen wir die Determinanten der subjektiven Qualität von Wohnvierteln unter besonderer Berücksichtigung der Verknüpfungen zwischen zwei unterschiedlichen, aber miteinander zusammenhängenden Dimensionen, anhand derer Privatpersonen ihre Wohnlage bewerten: Zufriedenheit, die traditionell als kognitive Beurteilung aufgrund der Erreichung eines gewissen Niveaus oder einer Aspiration betrachtet wird, sowie Verbundenheit, die sich auf eine affektive Bewertung bezieht. Mit Hilfe von Daten des British Household Panel Survey wird ein bivariates Probit-Modell geschätzt. Die Schätzung ergibt eine signifikante residuelle Korrelation, was darauf schließen lässt, dass sich die Zufriedenheit und Verbundenheit sowohl aus den kognitiven als auch aus den affektiven Aspekten der individuellen Einstellungen ergeben. Darüber hinaus erweisen sich die durchschnittliche Zufriedenheit und Verbundenheit auf der Ebene des Wohnviertels als wesentliche Determinanten für die individuellen Antworten.

Wohnviertel Zufriedenheit und Verbundenheit Soziale Wechselwirkungen Bivariates Probit-Modell

Corrado G., Corrado L. y Santoro E. Determinantes individuales y sociales para la satisfacción y la conexión en los barrios, Regional Studies. En este artículo analizamos los determinantes de la calidad percibida de los barrios, prestando atención al nexo entre dos dimensiones distintas pero relacionadas a través de las cuales las personas valoran sus barrios: la satisfacción, que tradicionalmente se ha considerado un juicio cognitivo basado en el logro de algún nivel o alguna aspiración; y la conexión, que está relacionada con una valoración afectiva. Calculamos un modelo probit bivariado usando datos de la encuesta de panel británica BHPS (British Household Panel Survey). El resultado del cálculo es una correlación residual significativa, lo que indica que la satisfacción y la conexión están relacionadas tanto con aspectos cognitivos como con aspectos afectivos de las actitudes de las personas. Además, la satisfacción y la conexión medias con respecto al barrio parecen ser determinantes fundamentales en las respuestas de las personas.

Barrio Satisfacción y conexión Interacciones sociales Probit bivariado

Keywords:

• Neighbourhood
• Satisfaction and attachment
• Social interactions
• Bivariate probit

JEL classifications::

• C35
• D10
• R23

Acknowledgements

The authors wish to thank Roberta Distante, the Editors, and two anonymous reviewers for their helpful comments and suggestions. Any remaining errors are the authors’ own. Luisa Corrado acknowledges the support of the Marie Curie IE Fellowship 039326 and the Marie Curie Excellence Award 2007. Emiliano Santoro gratefully acknowledges financial support from the UniCredit & Universities Foscolo Foundation.

Notes

On these aspects, see Connerly and Marans (1985), Ringel and Finkelstein (1991), and Bell et al. (2005).

In addressing these questions, it is important to acknowledge that the distinction between satisfaction and attachment reflects the standpoint of a vast literature exploring various aspects of subjective well-being (Kahneman and Deaton, 2010).

Individuals can derive financial and/or emotional support from social networks. Under this perspective, changing neighbourhood is likely to deplete the stock of social capital.

In this respect, Iyer et al. (2005) recognize the importance of pursuing a more region-specific approach towards examining social capital development.

For a comprehensive list and classification of the covariates, see the third section.

Manski (1993) considers three factors affecting individual choices (1) endogenous effects, intended as the propensity of an individual to conform to the average behaviour in that individual's peer group; (2) exogenous effects, that is, those effects exerted by the characteristics of the reference group; and (3) correlated effects, or the propensity of the individuals who belong to the same group to behave similarly because they have similar characteristics or just face and live in the same institutional and social environment.

As in Blume and Durlauf (2005), it is assumed that the agents’ information set includes values of other determinants within their group, as well as the equilibrium expected choice in their group. Agents are assumed to be unable to observe the choice of others or their random payoff terms (u _kig , k = 1, 2).

Children in the original household were interviewed when they reached sixteen years of age.

The BHPS aims to provide cross-sectional population estimates for the lifetime of the study in order to facilitate longitudinal research. It provides timely data on the short-term processes of change at the individual and household level for several domains (that is, labour markets, income, savings and wealth, household and family organization, housing, consumption, health, social and political values, and education and training). For further details, See Institute for Social and Economic Research (2006).

In turn, these contain, on average, 2500 delivery points, which are approximately equivalent to addresses.

For the variable proxying individual neighbourhood satisfaction (attachment), the fraction of missing values is around 4.7% (5.2%).

Standard measures of socio-economic and occupational status are not included. This choice is justified by the strong correlation of these variables with household income.

Social capital displays strong correlation with homeownership. Because of high transaction costs in the property market, homeowners tend to be relatively less mobile, and low levels of mobility predict high levels of social capital (Glaeser et al., 2002). Several works have emphasized that homeownership increases with social capital as homeowners usually invest more in long-term relationships with their neighbours (in particular, see Rohe and Stewart, 1996; and DiPasquale and Glaeser, 1999).

All correlation coefficients are between individual-level variables, with the exception of contextual effects such as crime, pollution, light, street and neighbours’ noise, which are measured at the neighbourhood level.

The rest of this paper will use, with no distinction, the expressions ‘attachment’ and ‘preference to stay’, as the latter summarizes the variable used to proxy the former.

The selection of a discrete-choice setting is also driven by a further consideration. As suggested by Blume and Durlauf (2005), whereas the reflection problem highlighted by Manski (1993) arises naturally in linear-in-means frameworks – producing a lack of identification – it does not necessarily generalize to alternative data structures such as binary-choice models. The logic of the reflection problem as it emerges in the linear model indicates why identification does not fail in binary-choice models, as the dependent discrete variable that proxies the latent variables, bounded between 0 and 1, cannot be linearly dependent on contextual and exogenous factors. Non-linearity is crucial in this sense.

For more details on the implementation of bivariate probit techniques, see Wooldridge (2002) and Monfardini and Radice (2008). As to the application of these models to the analysis of social interactions, see Jaenicke (2009). Bi-probit for STATA was used to perform the estimation.

Without loss of generality, the time subscript was reintroduced to account for the difference between individual fixed effects and other variables.

Once quasi-fixed effects are considered, the endogeneity problem associated with this regressor is removed. Alternatively, a ceteris paribus analysis that does not account for quasi-fixed effects may not allow one to determine the causal effect exerted by the potentially endogenous variable (average satisfaction) on the dependent variables (individual satisfaction and attachment), the reason being that variations in the regressor reflect into variations in the error term, thus confounding the effect of the regressor on the dependent variables.

This is also the case in the Netherlands, Denmark and Ireland. Diaz-Serrano (2006) shows that analogous results can alternatively be obtained in two ways: firstly, by giving an incentive to private homeowners to offer their flats in the rental market, which in turn implies a larger supply of dwellings for rent, and hence lower rents (this is the case of Germany); and secondly, by providing housing allowances to low-income renters (this is the case of Belgium, France and Finland).