Civic engagement and socio-economic proximity in urban areas

ABSTRACT

Diversity boosts innovation and creativity in urban contexts, but it can also undermine civicness by negatively impacting individuals’ trust of other citizens, also hampering economic and institutional performance. By employing a spatial analysis approach using geocoded data from 5776 residents in three major urban Italian areas, we explore whether sharing geographical space in a context of socio-economic difference (in terms of income and education) affects individuals’ choices to be civically engaged. The evidence reveals that the geographical proximity of an individual to others with different socio-economic characteristics decreases their civic engagement. A key challenge when designing urban policies is to reconcile the positive effects of diversity as it influences different economic–societal aspects, with the associated social tensions.

KEYWORDS:

• civic engagement
• urban diversity
• social capital

JEL:

• D7
• D71
• D72

ACKNOWLEDGEMENTS

We thank Giovanni Cerulli for technical suggestions and Emiliano Mandrone for assistance with the data. This paper benefitted from comments made at the VI International Workshop on Computational Economics and Econometrics, CNR, Rome, 26–28 June 2018; the 58th ERSA Congress, Cork, 28–31 August 2018; the XXXIX AISRE Conference, Bolzano, 17–19 September 2018; and the Annual WINIR Conference, Lund, 19–22 September 2019. We are grateful to the Study of Regionalism, Federalism and Self-Government (Issirfa-CNR) for input and support during Dr Andriani’s period as a visiting researcher, 21 May–4 June 2018. Finally, we are indebted with the Editors for their support and suggestions. Usual caveats apply.

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the author(s).

Notes

1 Industrial districts can be defined as local systems characterized by an active co-presence of a human community and a dominant industry comprising a set of small independent firms specializing in different phases of the same production process (Becattini et al., 2014).

2 The Istituto per lo Sviluppo della Formazione Professionale dei Lavoratori (ISFOL – Institute for the Development of Workers’ Professional Training) is a national research institution controlled by the Ministry of Labor and Social Policies. On 1 December 2016 it was restructured and renamed as theIstituto Nazionale per l’Analisi delle Politiche Pubbliche (INAPP – National Institute for the Analysis of Public Policies).

3 We employed the software Google Maps to project the points onto the maps.

4 The component belonging to the largest eigenvalue was extracted, and 76.48% of variance explained. An unrotated factor solution was employed.

5 The analysis was carried out using the software GeoDA (Anselin et al., 2006).

6 Different techniques can be adopted to calculate the weights matrix. In the analysis that follows, the weights matrix was created using a great circle distance (arc distance in GeoDA software) between units with a bandwidth corresponding to the maximum distance observed (27.48 miles).

7 We used the software ArcGIS to partition individuals basing on geographical distance. The HDBSCAN method was applied (Campello et al., 2013).

8 ‘High’ and ‘low’ classes are defined according to the mean value of the distribution. Values are dichotomized in order to enhance the visualization in the maps.

9 Estimations were carried out using STATA 15 software (StataCorp. 2017. Stata Statistical Software: Release 15. College Station, TX: StataCorp LLC).

10 In these estimates we measure proximity by employing the square root of the inverse of the geographical distance. Replicating the same model using a different method to transform geographical distance in geographical proximity, namely a linear method, an inverted sigmoid and convex, do not affect the results; hence, we do not report the table here.

11 The coefficient of proximity heterogeneity can only be interpreted in terms of its correlation with the dependent variable CE. It cannot be interpreted in terms of proportional odds because of the way both the dependent and the explanatory variables are derived. Our measure of CE is obtained by aggregating (by addition) the frequency of participation to associational activities; the more frequent individuals participate in such activities, the more they are ‘civically engaged’. Yet, we cannot claim that a person who scores 6 in terms of CE is three times more engaged than someone who scores 2. Regarding our measure of proximity heterogeneity, this is a composite measure of geographical proximity and socio-economic heterogeneity, which in turn is obtained by aggregating education and income levels. Following these considerations, we can limit our interpretation as follows: any increases/decreases in geographical socio-economic heterogeneity can be associated with higher/lower attitude to CE.

Additional information

Funding

This work was supported by the BEI Small Research Grant, Birkbeck University of London.