Should Policy Makers Strive for Neighborhood Social Mix? An Analysis of the Western European Evidence Base

Abstract

The paper presents an analytical framework for elucidating the equity and social efficiency criteria that might be used to justify a housing policy aiming for a substantial mix of neighborhood residents by income, ethnicity and/or immigrant status. This framework permits the classification of multivariate statistical studies comprising the Western European evidence base and shows the importance of distinguishing intra- and extra-neighborhood processes when evaluating evidence related to the efficiency criterion. Evaluation of the evidence base in light of this framework reveals that it sufficiently supports a mixing policy aimed at avoiding concentrations of disadvantaged individuals if, and only if, policy makers emphasize equity grounds (i.e. improving the well-being of the disadvantaged absolutely). The evidence base does not support a mixing policy on efficiency grounds, regardless of whether intra-neighborhood social interactions or extra-neighborhood stigmatization/resource restrictions are presumed to be the primary causal mechanism for neighborhood effects.

Keywords:

• Neighborhood effects
• social mix policy

Acknowledgements

In preparing this paper, the author has benefited greatly from the suggestions provided by Roger Andersson, Alex Marsh, Sako Musterd, Wim Ostendorf and three anonymous reviewers. The author also expresses gratitude for the excellent clerical support provided by Richard Ban, Noelia Caraballo and Phyllis Seals. The opinions (and potential errors) contained in this document are his own.

Notes

¹ With only two groups %D = 1 − %A, but it is important to keep these two terms separate because the source of the social externality may only be one group or the other.

² Unusually explicit articulations of such a position can be found in Andersson (2004) and Delorenzi (2006).

³ The statistical studies of neighborhood effects in Europe have employed a variety of datasets: British Household Panel Survey (Bolster et al., 2004; Buck, 2001; Gordon & Monastiriotis, 2006; McCulloch, 2001; Propper et al., 2004), Dutch income tax statistics (Musterd et al., 2003), German Socio-economic Panel (Drever, 2004), Swedish national register (Andersson, 2004; Andersson et al., forthcoming; Brannstrom, 2004; Galster et al., forthcoming; Musterd et al., forthcoming), Finnish national register (Kauppinen, 2004), and tailored resident surveys or administrative databases in particular cities (Farwick, 2004; Friedrichs & Blasius, 2003; Oberwittler, 2004; Van der Klaauw & Van Ours, 2003).

⁴ The three income groups of male neighbors were defined by the 30th and 70th percentiles.

⁵ The Atkinson & Kintrea (2004) qualitative study of key informant opinions in Glasgow is also noted, which found that some espoused the relative deprivation consequence of extreme social mixing within neighborhoods.

⁶ This ‘Townsend index’ consists of a weighted average of tenure, density, unemployment and auto ownership characteristics of the neighborhood.

⁷ For a review and application of these statistical procedures, see Galster et al. (2000).

⁸ Although arguably some non-linearities were evinced at the extreme values of neighborhood conditions, they involved so few neighborhoods that results may not have been robust.

⁹ Low income was defined as the percentage of males in the lowest 30th percentile of the labor income distribution.

¹⁰ This conclusion is consistent with another body of European literature that has found that there are, indeed, minimal social interactions among advantaged and disadvantaged groups who may be sharing the same neighborhood. See especially Blokland-Potters (1998), Kleinhans (2004) and Allen et al. (2005).

¹¹ Starting from a different point and applying a different form of analysis, Berube (2005) reaches a similar conclusion in the British policy context.