The Evolution of Segregation in Large Italian Metropolises, since 1991

This comparative research aims at exploring the dynamic of residential segregation in large Italian metropolises (Milan, Rome and Naples) since 1991. At the core of our framework is the idea that social relationships, attitudes and political behaviour are influenced by the social environment, which comprises both a population structure (a geographical space containing a specific set of individuals/families) and an opportunity structure (opportunities for work, education, health, social capital accumulation, sport and leisure). From this perspective, we look at the most relevant characteristics of metropolitan populations exploring differences and relationships between individuals in a relational manner. Indeed, the key characteristics of the opportunity structure of a metropolis relate to the spatial relations that exist between existing opportunities and specific social groups. The analysis of these spatial relations can help us identify relatively enduring “structural effects”, which are the object of the research.

The Italian case is largely under-represented in the international literature about urban segregation and the social division of metropolitan spaces. The few available research spans across the last four decades and has focused mainly on the residential patterns of ethnic minorities (Petsimeris, 1998; Scarpa, 2016; Mingione et al., 2008), and sometimes on social segregation (Cousin & Préteceille, 2008; Barbagli & Pisati, 2012). These studies, however, often focus on central municipalities rather than on whole metropolises, and rarely take into account comparative perspectives at national or international level.

• We do not really know the dynamics of social change in Italian cities. The only case studies currently available are mainly on ethnic segregation (Petsimeris, 1998; Mingione et al., 2008; Scarpa, 2016);
• For instance, we want to understand where the upper and upper-middle classes have gone in the last 30 years: have they moved to the suburbs, in small municipalities highly segregated? Or have they remained in the main city centers? Does it depend of which fraction of the upper and upper-middle classes we are looking at?
• This type of wide-range quantitative research, based on descriptive statistics, also aims to produce hypotheses for subsequent qualitative research.

Much of the studies on Italian metropolises analyze the phenomenon of segregation only at municipal level. The only exception is Cousin & Préteceille (2008). This happened because the data is generally presented for administrative entities like municipalities, and therefore not foremost linked to a theoretical reflection. The Italian institute of statistics, ISTAT, does provide a functional definition of territorial systems with statistical and geographical updates every ten years. This definition draws on the concept of Labour Market Area, and is being shared at European level with other national institutes of statistics, including the UK’s Office for National Statistics. However, the concept of Labour Market Area is quite different from the main definitions of metropolitan areas used across Europe, and the methodology to identify such LMAs is acknowledged to be hardly applicable to the largest European agglomerations, like Paris and London. We therefore define the perimeters of the three metropolitan areas we study according to the OECD definition (OECD, 2013). Our choice of the geographical definition of metropolitan areas provided by OECD is motivated by an overview of existing methods and approaches to define metropolitan areas at national and international level. The OECD definition is indeed homogeneous at international level. The OECD metropolitan dataset was thus chosen because it offers both high scientific accuracy and easy cross-national comparability.

The Italian census of population relies on a single set of elementary units for both collection and output, the sezioni di censimento, and the office of national statistics (known as ISTAT) has not developed an output geography.

• The sezioni are rarely used for statistical analysis and mapping, due to their very small populations and because ISTAT does not release a full set of data at this fine level of spatial resolution. Until recently, ISTAT did not release any data to researchers or the public at the level of the sezioni.
• In order to make full use of small area data in Italy, there is therefore a need for a new output geography situated between the sezioni and the comuni (municipalities).
• Indeed, sociological analysis based on large spatial units such as the latter is useful, but reveals only macroscopic differences in the distribution of opportunities and social groups (e.g. between the richest areas of the city and the poorest, or between particularly affluent and disadvantaged rural areas). They are therefore not suitable for applications based on the use of small areas as a proxy for individual characteristics, or for exploring the micro-spatial structuring of residential patterns, which is of far-reaching policy relevance.
• Moreover, the geography of the sezioni di censimento in Italy is quite unlike that of the spatial units found in other countries. No lower population threshold is applied, with the result that a very large proportion (in the region of one third) of sezioni have a population below 10.
• Because they are used primarily for enumeration purposes, the boundaries of the sezioni are also variable over time; all three censuses carried out since 1991 have been preceded by extensive redefinition of the boundaries of the sezioni even of some that were not split or merged).

We had therefore to develop ourselves a stable output geography for the 1991, 2001 & 2011 Italian censuses.

• The far-reaching revision of boundaries does not imply that the località and sezioni can never be compared across census waves.
• In the best case, boundary revision merely involves technical considerations regarding how to represent the distribution of settlements and nuclei, and the units involved are comparable.
• This comparability cannot be established by inspecting the boundary data… and requires either: (a) use of auxiliary information or (b) inspection of aerial photos.

We work on census microdata at the individual level to create a new variable of socio-professional positions.

• The new classification has been created combining individual occupation, type of contract and sector of activity.
• The combination of the three produces a more informative socio-professional classification, well adapted to compare the social segregation dynamics of Italian metropolises with those of other European metropolises (particularly London and Paris).

To analyze residential segregation, we then intend to build a typology of spaces based on the weight of the various socio-professional categories within the active resident population of each space (cf. Préteceille). This method of neighborhoods classification is based on a combination of ascending hierarchical classification techniques and correspondence factor analysis.

the team

• Bruno Cousin, Sciences Po, CEE, Paris, France 
• Matteo Del Fabbro, France and Politecnico di Milano, Department of Architecture and Urban Studies, Milan, Italy and Associate Researcher at Sciences Po, CEE, Paris
• Niccolò Morelli, Department of Sociology and Business Law, University of Genova, Italy

• Matteo Piolatto, Department of Economics and Management, University of Brescia, Brescia, Italy
• Jonathan Pratschke, Department of Economics and Statistics, University of Naples “Federico II”, Napoli, Italy.
• Tommaso Vitale, Sciences Po, CEE, Paris, France