School segregation in 4 Italian metropolitan areas. Rescaling, governance and fragmentation of immigration policy

Post-Fordist migration flows and settlements are characterized by a “superdiversification” process (Vertovec 2007) that has also a territorial dimension, with a number of new destinations mixing openings and closures (Alexander 2007). The analysis of rescaling processes allows to frame the shifting boundaries of socio-economic processes and institutional answers, how this affects the multilevel governance of immigration policy in the interplay between the place-based agency of local actors (immigrants included) and the structural opportunities locally available. In hypothesis, migration mobility, settlement and socio-economic participation are tied with the scalar position of the destination area, that in turn they influence as localized actors. This analytical approach will be used here to analyze school segregation of children of immigration in four Italian metropolitan areas (Milan, Bologna, Rome and Naples) having different socioeconomic structures, migration processes and policies. In particular, we will make two qualitative focus on the cases of Rome and Bologna, providing deeper insights on the relationship between socio-economic participation, settlement trends at neighborhood level, school distribution policies and practices. The research presented will be based on a mixed method approach, with: a quantitative analysis using data on population and pupils (mainly using the decomposed Theil index in order to measure segregation at different scales), to see effects of school offer (structuring segregation among neighborhoods) and individual choices (structuring segregation within neighborhoods) qualitative accounts (in particular vignette interviews), focusing on policies and practices affecting school access of pupils with an immigrant background at local level. Thus, the paper will be structured as follow: after a short theoretical introduction based on the literature on rescaling and on segregation, we will introduce the socio-economic structure and the migration patterns of four Italian metropolitan areas, then focussing on the effects this has on segregation at area, city and neighborhood level. In particular, we will focus on school segregation (in particular for Rome and Bologna), with a deeper analysis on its influencing factors – including the segmented effect of national and territorial policies, that will be exemplified also with some


Introduction
More and more of recent literature is highlighting that immigration policy has a territorial dimension and that localized processes and actors play a role that is not only local, since it affects the boundaries of (national) identities and rights (cfr. Brenner 2010). Furthermore, the study of the multilevel governance of immigration (Zincone e Caponio 2006) showed that the local dimension has a "model-making" role through a more or less creative implementation of national regulations. As a consequence, the devolution of policy tasks to subnational regulation allows more and more a discrimination based on diversity cleavages: in terms of citizenship and naturalization, but also of place of residence and other forms of denizenship (e.g. the type and duration of permits of stay). Thus, local policies play an increasing role in affecting migration and integration processes, interacting with economic processes: outcomes are tied to urban and regional hierarchies and power relationships, to social and demographic profiles, ongoing economic transformations and pathdependencies -but also to the active role as "scale-makers" that migrants have (Glick Schiller e Caglar 2010). To sum up, diverse migration paths, incorporation and participation are tied with the scalar position of the destination areas; on the other hand, migrants do influence themselves that position, being workers, generating wealth, interacting with and making institutions.
Starting from the theoretical background sketched above, in this paper we will analyse school participation and segregation of pupils with immigrant background (PIBs) in Italian schools, focussin on 4 metropolitan areas (Milan, Bologna, Rome, Naples). These cities have a different position in the urban hierarchy, but they are nevertheless attractive for migrant settlements as focal points in wider regional hubs (Pugliese 2010). Such a territorialized model of integration deploys according to the main features of local labour and housing markets, thus producing a peculiar model of micro-ethnicization and micro-segregation -that can be considered part of a national immigration model that didn't develop a nation-wide discourse and practice for migration policy. Hence, in this paper we will analyse the link between cities, immigration, policies and school segregation: we consider the latter as a privileged point of view to study the territorial dimension of social policy, since it is related both to migration settlement processes (shaped by socio-economic factors: housing systems, local spatial configurations and labour markets) and to institutional answers. As a matter of fact, we maintain that policies are an important predictor of selection and integration patterns for both first and second generations of immigration: access and success are not as related to minorities' characteristics as to the adequacy of educational systems in coping with diversity (Dupriez et al. 2008;Gewirtz e Cribb 2008;Oecd 2008). In the end, we will try to explain school segregation in our case cities with the interplay of urban structure, individual choices and institutional models of production of social difference.

Main features
Italy has a comprehensive education system since 1962 (when early tracking was abolished), that was widely confirmed later in many respects, e.g. with the abolition of most special schools in the 1970s. Its system is made up by two basic tiers -a 5-year primary education with a common curriculum and most of teaching hours dispensed by a single teacher; a 3-year lower secondary tier with a common curricula and specialized subject teachers -and a 3-to-5-year upper secondary tier with different paths (see Fig. 1). This system has been traditionally governed in a centralistic, bureaucratic-hierarchical way by the Ministry of Education, notwithstanding a well-know, traditional territorial cleavage (Grimaldi, Serpieri, 2012): a fundamental feature of the country, that doesn't concerns only socio-economic dimensions, but also institutional performance and structures, education and training ones included.
Though, this "central government monopoly" has been eroded by two waves of decentralization: during the 1970s, and later and more substantially during the 1990s-2000s. In the first period, responsibility over vocational training was devolved to the nascent Regions, while grassroots instances were recognized with the creation of representative bodies in schools (1973). In the second period, schools were granted autonomy in management, organization and teaching (1999), with first degrees of flexibilization, that allowed a school-and local-centred share of curricula to be set. 2 Then, decentralization process climaxed with 2001 Constitutional reform (still to be implemented in full), that expanded financial, legislative and management role of Regional governments considerably, also in the field of education and training. From then on, the process slowed down, even with signs of recentralization (Oecd, 2012).

Emerging criticalities
In the intersection of traditional (e.g. class and territorial divides) and new problems (e.g. the destandardization of life courses, the changes in family subsidiarity and labour market structure) and institutional responses (in particular, a decade of reforms in the 2000s 3 ), it is possible to stress some governance weaknesses of the Italian education system that will re-emerge throughout this article: • an inconsistent governance structure (mainly hierarchical and bureaucratic), that results in a blurred school autonomy and in a wavering decentralization, especially as far as the management of resources is concerned; • a decentralization that set up an unclear division of tasks and responsibilities between State and Regions, further institutionalizing a traditionally strong territorial divide; • an unreformed lower secondary tier, a weak point in the whole education path; • a challenge to the traditional integrative and comprehensive school model raised by new profiles of pupils and families.
"Universal" in its welfare aims, the Italian education system proved to be highly selective in practice, with tracking effects and serious problems in granting equality and social mobility, as part of a "blocked society" where social class and divide are strong determinants of success and of reproduction of intergenerational inequalities (Schizzerotto, 2002;Ballarino, Checchi, 2006;Ballarino et al., 2009;Schizzerotto et al., 2011): social mobility is weak, both for traditional categories of disadvantage (e.g. social class) and for more recent ones (e.g. immigrant background).
The participation of a growing number of PIBs is one of the biggest challenges the Italian education system had to face in the last decade: it implied not only a quantitative, but also a qualitative change, due to a large plurality of trajectories -by generations and ages at migration, areas of origin and destinations, schools involved. The number of non-Italian citizens in Italian schools grew from 574,000 in 2007 to 756,000 in 2011, now a share of 8.4% of pupils, whose 44.2% are born in Italy (peaking to 80% in preprimary education). They are mainly from Central and Eastern Europe (Romania, Albania, Moldavia, Ukraine), but also from Northern Africa (Morocco, Tunisia) and Asia(PRC, India, Philippines): though, these nine origins account for less than two third of PIBs, evidence of the above-mentioned plurality.
In the secondary tiers we are interested in, they count for 9.3% in lower secondary and 6.2% in upper secondary (Miur, 2012a) Furthermore, the general problem of school tracking has a strong accent when PIBs are taken into account: vocational education is attended just by 19% of Italian pupils, but by 30% of PIBs born in Italy and 40% of foreign-born PIBs. This means that PIBs are 12,1% of pupils il vocational education, but just 2,7% in general education (ibidem). These weaknesses are also matched with a significantly delayed educational career, especially in vocational schools: more than a PIB out of seven there is 20 or older (Miur, 2012b), as an effect of delays in previous grades, so that the gap in hold back rate is impressive (see Fig. 2).

Fig. 2. Hold-back rate by age and school tier. 4
Source: own calculation on Miur, 2011;Miur, 2012a Nevertheless, tracking, segregation and underachievement of PIBs are not a problem just in Italy (Szalai, 2011), so it is worth noting what is peculiar in this case: first, the gap of performance with the natives is wide in comparative terms, even though mainly due to first generations; second, it is not so related to socio-economic background, but mostly to migrant condition itself, thus possibly to direct discrimination effects (Oecd, 2010). A further analysis of Fig. 2 can help making another point: transitions from a tier to another are tough, but the whole lower secondary education opens the gap for PIBs, thus producing complex transitions in the following steps. So, this general weak point of the Italian educational system has a specific effect on PIBs: being held back increases the chance to drop out, and increases tracking processes toward shorter vocation paths.

School system and locales: educational policies, segregation and minorities in Italy
More and more recent migration flows are shaped like mobility process between locales, with a core relevance of subnational areas, according to their competitiveness and institutional dimensions: this is very clear in the Italian case, where migration is more a territorial issue than an urban issue, due to a very scattered urban and economic fabric (Mingione 2009). This could be also an explanation why school segregation of PIBs hasn't reached the top of the policy agenda: the territorial fragmentation (and scattering of immigrants) might have worked has a shock-absorber of segregation and concentration. Though, also an institutional dimension has to be taken into consideration: since the 1970s -and further from 1999 onwards -the organization of educational institutions, including important responsibilities on access dimensions, supported a transformation toward school autonomy. Also with a problem of scale relations, further challenged by the need to include new -migrant -pupils. Actually, the institutional level that is decided to be the reference point for school access can strongly influence school accessibility and choices: the degree of autonomy left to school can have an effect in selection processes, playing a role in segregation processes. A problem further complicated by the occurrence of a vertical segregation (channelling/tracking, as mentioned above) that meets the horizontal (territorial) one.

Policy priorities and PIBs
Actually, the Ministry of Education has a long record of documents on the integration model to be used with PIBs, that culminated with a paper on "The Italian way to intercultural education and the integration of foreign pupils" (Ministero della Pubblica Istruzione, 2007). Though, for many years, PIBs' tracking, segregation and guidance haven't been an educational policy priority at national level, much more focussed on integration issues in earlier tiers, where numbers have been more relevant. So, the evidence of such a discrimination is a challenge especially for a comprehensive system that refuses special education in its norms and professional cultures. Somehow, we can hypothesize that a comprehensive approach in this field was not resulting from a reflexive thinking on explicit expectations and goals for this target group, but as an isomorphic application of receipts thought for other target groups -a path-dependency more than the outcome of a real and actual debate. This may cause unintended consequences in a context where a comprehensive approach is under constant challenge due to retrenchment policies. As a matter of fact, Ministry guidelines provided principles, but few indications and tools for everyday practices, and they were matched with a very limited re-training of staff and likewise limited action plans and resources. This may be seen as a kind of State withdrawal, in case coupled with a discourse on subsidiarization and the role of local actors in finding proper answers to emerging needs. The consequence of poor institutionalization in a decentralizing comprehensive system is that schools and local authorities built up their know-how incrementally within local policy networks. Moreover, since immigration is a politically sensitive issue, dumping towards local answers seem to work as a "sweep-it-under-the-carpet" strategy: accommodation via microregulation avoids a negative politicization of the issue, heated national debates and hatred. This put local actors under great pressure, without any empowerment and support mechanism: emerging risks, like the ones concerning PIBs, find fragmented, place-specific answers.

Intersections between PIBs coping and the local dimension of educational policy in Italy
As hinted above, segregation hasn't been a policy issue in Italy for a long time. Since no large "ghetto" is present; microsegregation limits negative effects in urban areas; PIBs have also been rivitalizing schools in depopulating, rururban areas; poor media attention was charged on the few existing cases; the problematization of concentration cases was limited. Nevertheless, there are some risks, since the local governance of school provision and access can fail in preventing and coping with segregation cases. Actually, since 1999 school districts have no self-government, and since 2002 districts themselves were abolished. As a consequence, the thenexisting concept of school "catching area" and its regulation was abolished, with no real public debate on that -till school segregation of PIBs entered the policy and public agenda in late 2000s. At first (2009), the Ministry of Education released a circular letter requiring at least school to publish admission criteria; later on (2010) another circular letter definied a maximium threshold of foreign pupils per class (30%). In this respect, the lack of a territorial self-government of schools caused difficulties in planning the educational offer in a way able to meet changing localization patterns of school-age population: for example, the low number of schools in peripheral and deprived areas caused lower-class and disadvantaged pupils to be overrepresented in few periphery schools (Torri and Vitale 2009). Furthermore, today retrenchment policies in the educational system had a role in the cut and merger of schools: it is still an open question if and how much mergers do take into account pupils' distribution.

An example of unsuccessful regulation of concentration: blaming the scapegoat by setting a threshold for PIBs
An example of blame game is tied with the above mentioned Ministry circular letter 2/2010 ("Indications and Recommendations for the integration of pupils with non-Italian citizenship"), that defined a 30% threshold per class for PIBs (especially newcomers with weak language skills). This letter touched a real problem (that we will further hint to in the next paragraphs), i.e. the uneven distribution of disadvantaged pupils, though providing a negative and discriminatory symbolic answer: just a formally tough limit (to show commitment to public opinion), though with unclear implementation rules, no resources and a total devolution of responsibility to local networks. So, it allowed political gains without much effort, as we can see retracing the reason behind this norm. Such a national measure sprang basically just from a single case of a primary school located in a peripheral district in Rome, where PIBs were some 95% of enrolled pupils, as a consequence of local discretionary practices: an activism toward immigration issues in that school, a neglect of it in the nearby ones. What happens in the capital city climbs soon the political agenda, and between 2009 and 2010 some right-winged stakeholders -including the (post-fascist) then-mayor of Rome -labelled this school as a "ghetto to clear up". Notwithstanding divergent discourse on this case -(some) parents and (many) teachers maintained that it was a positive experiment of integration where pupils with different ethnic background (most born in Italy) were taught together -some local media and politicians used this case as a flag to stress negative effects of multiculturalism and immigration policy in Italy. In a conflict over two definitions of "Italianness" -civic or ethnic -the Ministry circular letter ended up supporting the second, even reversing a long-lasting position on intercultural education and migrants' inclusion. Anyway, charged with symbolic consequences, it didn't have great impact and proved to be unable to reach the goal it stated -not even in the school where the issue was raised. Actually, nothing changed there, since parents' association raised a claim in front of the Administrative Court and obtained an exemption, while other claims where raised elsewhere for discrimination.
3. Subnational policies. The regional and municipal level Given the above mentioned national policies, we will frame here first regional policies and then local ones, to disentangle the relationship between scales and practices in affecting settlement and segregation patterns. Thus, we will account for the governance of immigration in the four Regions where our case cities are (Lombardy for Milan; Emilia-Romagna for Bologna; Latium for Rome; Campania for Naples), with a special attention for minors and education. The analysis will focus on social expenditure and policy priorities, while practices will be analysed through a vignette administered in our four case cities and in another town per region (Brescia, Modena, Frosinone, Salerno).
3.1. The regional level 3.1.1. The resources It's not easy to assess social expenditure targeting minors with an immigrant background: they access both dedicated (e.g. mediation) and mainstream policies and services (e.g. free textbooks, scholarships, juvenile social service...), but -especially for the latter -there are rarely data on their share on users. We will use some proxies. The survey made by the national statistical office on municipal social services includes the area "immigration and Roma" (Istat 2012), that makes up a limited share of the total local social expenditure: in 2009, it ranged from 1,1% in Campania to 4,4% in Latium (where anyway it is largely used for the Roma target), passing for 2,1% in Lombardy and 3,3% in Emilia-Romagna. It means a per-capita yearly expenditure of € 25 in Campania and Lombardy, € 60 in Emilia-Romagna and € 70 in Latium. This expenditure have been decreasing in the years -notwithstanding the increase of immigrants and the increase also of general social expenditure -due to the negative politicization of the issue and the end of national earmarked transfers after the implementation of the 2001 Constitutional Reform. Actually, in 2003 the National Fund for Immigration Policy has been merged without earmarks in the National Fund for Social Policy: this caused a disinvestment in favour of more politically rewarding targets (see Einaudi 2007). For example, in those years social expenditure on immigration in Campania and Lombardy dropped dramatically. In some cases, it could be thatgiven the above mentioned politicization of the issue -there has been a shift from dedicated to mainstream policies, where immigrants can access anyway, though with less guarantees. Further information can come from regional deliberations on fund allocation, to understand how much Regions -that according to the reformed Constitution are the main player in the organization of welfare provision -can steer their municipalities. Actually, direct regional resources are quite limited: -In Lombardy, they account for just 20% of total expenditure on integration policy, as a consequence of the drop in mid-2000s (Tosi et al. 2012); -In Emilia-Romagna, they are some 25% (plus and indirect effect, since the Region requires a cofunding from local authorities), a share higher compared to other policy areas (Regione Emilia-Romagna 2006); -In Latium, it is very hard to define a yearly expenditure, due to very variable fundings, difficulties in expenditure and budget allocations not matched with actual transfers; -In Campania, in early 2000s the regional role was utmost, mainly due to a scarce local expenditure, while newest data show that 25% of the (poor) local expenditure on immigrants' integration originates from the Region.

Policy priorities and institutionalization of measures
Integration (especially school integration) is an issue usually included in regional policy priorities, especially in Lombardy and Emilia-Romagna. In the first case, regional guidelines promote educational participation, and language policies (Gambino 2005), being minors and youth the most frequent target (cfr. also Tosi et al. 2012). A dedicated Regional Observatory (Orim) and a big player from the civil society (Ismu) have been also managing a database of projects for intercultural education and made frequent training activities.
In the second case (Emilia-Romagna), in early 2000s teaching Italian to minors, intercultural education and afterschool activites were the third most funded priority (after dedicated counters and intercultural mediation, having anyway consequences also on the educational system). From the mid-2000s onward it became the biggest target, accounting for 1/3 of dedicated social expenditure in 2006. Today, the issue seems less prioritized, but anyway receiving resources with continuity (e.g. with funds from a 2001 Regional Law on the "Right to Education", whose 1/3 of resources targets PIBs, cfr. Regione Emilia-Romagna 2012). Not by chance, some years ago a survey pointed out that education was the area most covered by municipal policies for immigrants: 83% of the municipalities in the Region enacted education measures for PIBs (Pavolini 2006). On the other hand, in Latium a large share of resources is dedicated to Roma, asylum-seekers and refugees, so PIBs are not a priority target. Many local authorities chose to cope with them through the access to mainstream services, though rarely caring accessibility dimensions. In general resources seem to target more emergencies than a steady, continuous action (Marucci and Montedoro 2010). In Campania, the area minors-education-intercultural policy has been the second most endowed regional policy priority, counting for 22% of projects, 14% of resources and 20% of users in the period 2001-2004(Servizio gruppi etnici -Ormel 2004. Though, in general, just the Emilia-Romagna case shows a certain degree of institutionalization in programming and funding regularly this policy area and target (Campomori e Caponio 2009). The policy agenda has been also quite conflicting with the central government, and trying to build up a regional network that -besides a relevant public role -involved also the main players from the civil society.
On the other hand, in Lombardy the regional policy agenda on this issue has been poor, as shown also by the old and inadequate regional law regulating the issue (it dates back to 1988). Programming through regional government's deliberations proved to build up a fragmented and poorly endowed system, where civil society organizations play a pivotal role. In Latium, the role of the Region has been wavering. Low institutionalization in the policy agenda and endowments have been due to weak and changing regional governments. In 2008 a new regional immigration law was passed, but its implementation is yet to come. Finally, in Campania, programming activities followed national priorities, but experienced serious implementation problems: migration is here a lower priority, that has to be framed in a particularly weak local welfare system.

Practices at local level: coping a case
Using the vignette technique, we tested main features of local immigrant policies and how they affect educational access chances. Actually, we asked fourty social workers and civil servants in 8 cities how they would cope the case of J., a 15 years-old newcomer with poor language skills in Italian, whose family requested for school enrolment. This case is challenging since J. is at the limit of the compulsory education age in Italy, and requires a choice between a "social" and a "selective" view of education and schooling. 5 In principle, our interviewees maintain that the right to education is binding (as stated in the norms on this matter), but we can identify some problems due to the age of the pupil, especially if J. is undocumented. In some cases, there's a shrinkage of rights, partly due to the profiling of J. as not particularly "deserving" (adolescent, male, immigrant: an example of social dangerousness!), partly due to a lack of knowledge on relevant norms -both those concerning the educational system (duty and right to education; reform of the school tiers) and the rights of children and adolescents. As a matter of fact, till lower secondary education the right to education is not so problematic (also for undocumented pupils), while upper secondary education and vocational training experience a lack of attention -notwithstanding the need for coping and guiding over-14 PIBs is already stated in the 1998 immigration law. Privileged witnesses we interviewed underline that problems increased after 2009 security laws, compelling immigrants to show a valid ID while dealing with public institutions. Even though this law doesn't apply to "compulsory schooling services", this definition has blurred boundaries, so not rarely educational professionals are scared by possible lawsuits and can deny enrolment of undocumented pupils if they are not able to comply with bureaucratic rules. In other cases, school access is jeopardized by municipal exclusionary policies: e.g. in those municipalities requiring a residence certificate to access services -a certificate that by definition undocumented people cannot have. This doesn't hinder to enter in schools, but to fully access the right to education (e.g. to have free textbooks). These problems add up to governance ones: an integrated support network is often missing. School access is mostly seen as a problem of knowledge to be learned, but -besides specific projectsthere's poor attention on social support.

An overview of practices city by city
The general trend mentioned above can vary according to place-specific characteristics. For example, problems are less tough in areas where: − immigration is not large and not a hot issue in the policy debate; − where politics support pro-immigrant policies; − where there are specific skills spread in all the policy network, among public and private actors.
In Bologna, there's a centre for intercultural education (Cd/Lei) that collects good practices and coordinates intercultural mediation in schools. Besides, many associations and cooperatives can provide complement1ary services. The weak point is the lack of clear and institutionalized coping strategies, due to poor resources, skills and programming. Notwithstanding improvements in recent years, there's a discretion and fragmentation that can be detrimental: e.g., individual plans for school participation not shared with the pupil and his/her family; rejections of enrolment justified by inadequate resources to cope with PIBs' special needs. Modena -the other case in Emilia-Romagna -has a quite similar organization, with a coordination municipal centre, but more difficulties in coping with J. due to his age: some interviewees even suggested to enrol him in adult education. The two cities in Lombardy we studied -Milan and Brescia -show a more hostile relationship between school and non school actors, with a negative politicization of the issue. Some interviewees complain for a residual role of public actors, more focussed on emergencies that on day-by-day criticality. Improvements can be seen also here (e.g. with the definition of StarT centres to coordinate PIBs' inclusion paths in Milan), but in general there's a delegation and a protagonism of civil society organizations that compensates providing standard and good practices -even with a national relevance. In Latium -Rome and Frosinone -the first case is quite structured, though much delegating to Third Sector organizations, sometimes very experienced and effective; measures and services provided by the municipality have been frequent (e.g. the Integra Programme and "Intermundia" centres), though with problems in the continuity of funding and organization. In the latter case, a province with a small share of PIBs, measures are minimal, with some sort of coordination between volunteering schools and a subsidiary, emergency role of municipal welfare offices. In Campania, support chances are more limited: there were no targeted measures, but also difficulties in accessing mainstream policies due to lacking resources and skills. Adult education and civil society organizations partly compensate for these deficiencies.

Coping within a low standardization environment
After this assessment of regional programmes and local practices, we can rank our for cases according to the coping networks they have and the guidance provided by public authorities. The most structured network with a stronger public role is in Emilia-Romagna, while in Lombardy and Latium civil society plays a stronger role -sometimes complementary, sometime substituting absent public authorities. In Campania usually the role of public actors is the weakest, delegating to NGOs. Nevertheless it is worth noting that even when talking about "structured networks" we are analysing cases with a degree of institutionalization surprisingly low: the coping of J. shows relevant drawbacks in responsibilities, tasks, networks and eligibility and accessibility of rights. More than 20 years after the first circular letter of the Ministry of Education (205/1990) on PIBs and intercultural education, coping practices are still too often voluntarist and extemporaneous, variable in time and space. While at first we thought that J. was a problem as an extreme case due to an unusual age for school enrolment, it turned out that even the assumed "usual" wasn't so clearly defined. At micro-level, there are many coping strategies, but strongly fragmented not only among regions, but also among municipalities in the same region.

School segregation
As an example of an educational system that dumps many challenges on the local level in an unregulated way, we present here a quantitative analysis on micro-level school segregation.

Components of segregation
Italian cities we analyse here show an importance of metropolitan central areas for job opportunities and services, but also a relevant role played by surrounding residential belts in accommodating migrants, though without strong ethnic segregation effects. Such a distribution pattern favours a settlement of long-staying immigrant households with children in metropolitan belts, where sometimes the number and type of schools is not consistent with the increase of school population. We will show this point focussing on lower secondary schools, because they were main target of the research projects that fed this paper, but also because it is a frail key point in the Italian educational system: on the one hand, it is still part of comprehensive, general and compulsory education, so localization patterns should be homogeneous enough to cover territorial needs, while holes in the homogeneity could show the intersection between housing and school segregation; on the other hand, they start to be selective, both in its organization by subjects and in its actual effects (it is the level where the gap by class, origin and territory opens dramatically). The analysis will focus on datasets of first and last school years available (2003/04 and 2010/11) at the Ministry of Education's Statistical Office when we submitted our data request: 6 in this time span, foreign pupils more than doubled in the four case Provinces, so that nowadays their share on the school population is meaningful (but in Naples) -even though it is still far from levels reached in European countries with a longer immigration history. The general increase of population hasn't been mirrored in an increase of foreign pupils' segregation level 7 . And H index is not dramatic: even where it is highest (Milan and Naples), it is anyway a fourth of US levels. The level of segregation didn't change so much in a decade, and in Naples even dropped, since it was mainly due to the low number of foreign inhabitants. Even though the level of segregation has been steady, the situation is not static at all. Analysing the territorial components of the H index, we can see a) very different segregation models in the four cities; b) that the type of segregation changed (Fig. 3).  At first, we can see four different concentration models. Bologna has a more "balanced" (and lower) school segregation. Foreign pupils' concentration is due to their settlement in some urban areas (formerly working class ones) and in some municipalities of the Province (the ones mixing more residential and productive functions). All in all, the segregation between and within districts is limited. Segregation between the core municipality and the rest of the Province increases due to the doubling of PIBs and the stability of national pupils there. Mapping the data, we can notice high 8 The H index has been divided into 4 components, measuring the segregation between:

Tab. 1. School population -Italian and foreign citizens and segregation (H index
• Province and core municipality (Hcm); • Different school districts within the core municipality (Hc); • Different school districts within the whole metropolitan area (Hm); • within individual school districts (Hw).
concentration in large schools (squares instead of circles in the map) in the north-western part of the city (Fig. 4). At neighbourhood level, we have an effect related to family reunifications (Comune di Bologna and Info-Bo 2012). This has an effect on concentration in the core municipality, where most of reunifications with teen children takes place. They are placed in few lower secondary schools that have to cope with the highest share of new-comers (e.g. in Navile neighbourhood).

Fig. 4. Share of national and foreign students by school size. School year 2010/11
Source: our elaboration on data from the Ministry of Education Focussing on the wider provincial area, schools (even though not very large ones) with higher shares of foreign pupils are placed in North-Western towns, a large multispecialized areas for agriculture and industry; some schools with high concentration are also located in Imola, the second biggest municipality of the Province. In Milan, the most specific segregation dimension has to be found in the unbalance between the core municipality and the surrounding belt, with effects not so tied with settlement patterns, but mainly to the higher attractivity of schools in the core city: this effect is still important, but less than in 2003 due to the decrease of nationals' living in the province, outside the core city.
In the map referred to the school year 2010/11 (Fig. 4), we can see that most schools with high concentration in Milan are large schools, but we can also notice that schools in the surrounding areas (especially north) are becoming more and more similar to the ones in the core city -thus contributing to a deconcentration effect. On a lower scale, deconcentration took place also in Naples, where anyway foreign pupils grew just in few municipalities. Segregation is mainly due to differences between municipalities in the urban belt: though, no school has more than 30% of foreign pupils and few (small) schools have a relevant share of PIBs.
Last, in Rome pupils living in the urban belt do enrol in the urban belt, and those living in the core city enrol in the core city. The most relevant component is thus the micro one, within every district. Every district in Rome has at least a fairly big school with a high share of PIBs (close to 25%) and small school with a share close to zero. At least six districts have school with a very high share. In particular, the Eastern area of the city (formerly a working class neighbourhood) is more and more inhabitated by foreigners, and then by foreign pupils.
In the urban belt, we can see two types of concentration areas: • municipalities traditionally attracting population coming from Rome in suburbanization processes; • rural depopulating towns, where the high share of PIBs is due to the drop of nationals.

Segragation scales and accessibility in education
In general, the micro-dimension of school segregation seems to be the most important feature in most of our case studies -consistently with the analyses on settlement and residential segregation. This issue requires a clarification on the Italian model of school segregation. Concentration can be due to a variable mix of five factors: • Residential segregation pushing foreigners in some (small urban) segments; • Elusive strategies enacted by national parents, based on ethnic prejudice (white flight); • Sorting mechanisms (Ball et al. 1996), i.e. a quasi-market competition for scarce places in most attractive schools; • Negative side effects of school policies; • Shortcomings in planning and managing complex policies (education, labour market, housing, migration) that also interact each other. This is typically a problem that should be coped at regional scale -but Regional authorities, in our analysis, are those most absent in producing policies and measures for the school integration of PIBs.
To disentangle such a mix of factors, we will show some small scale examples that will epitomize the local outcome of the territorial reorganization of policies at different levels -just some examples that do not describe what is going on in every city as a whole, but show possible, diverse effects. To achieve this goal, from an analytical point of view we have to distinguish between individual enrolment behaviours and the ethnic composition of city districts. Furthermore, these two features should be integrated with an attention on private schools, whose enrolment structure can be quite different (we used here school data from 2003/04 matched with 2001 Census data). In Milan, as in many other Italian cities, enrolments in private schools are quite limited. Though, private schools there are mostly excluding foreign pupils. This implies an overburdening for nearby state schools -as in Via Padova neighbourhood. For the rest, elusive and discriminatory strategies seem limited: just in the city centre the share of PIBs is limited, but it could be based on other accessibility criteria (social, functional-structural and/or due to the low number of available places) 9 . In Bologna, private schools are not accessible to PIBs, and closeness to residence is the main selection criterion and source of concentration. In some neighbourhoods (e.g. Pilastro) this can cause a quite extraordinary level of concentration, that also cumulates with other disadvantaged conditions pertaining the Italian population .
In Rome the situation is somehow different: many small private schools -especially Catholic ones -include foreign pupils in different districts of the city. Though, in Rome it is also common to see unbalances between schools in the same neighbourhood, likely due to parental elusive strategies.
For example, in the middle-class area of Piazza Bologna there are two State schools, but foreign students are enrolled just in one. On the other hand, Esquilino district -a strongly multiethnic area -the concentration in a single school (named "Daniele Manin") cannot be explained without taking into consideration its early attention for integration policies: notwithstanding two other State school in a range of 800 mt., this is the one attracting most PIBs, being also a "Intermundia centre", i.e. one of the few schools that serviced the local education system in the field of intercultural education and PIBs' integration. Last, Naples seems to reproduce same features analysed for the other three cities, though at a lower level, given the limited number of foreign pupils. There's just one case of unbalanced distribution of PIBs in neighbour schools.

Conclusions
The analyses we reported above show the effects of a national context poorly regulated, where the governance of migration -that we analysed from the specific point of view of the school participation of PIBs -is mainly up to the interplay between labour and housing markets and to "molecular" relations taking place at the very local level. We could not find a clear, sole social map of school segregation. The distribution of PIBs follows settlement patterns based on the attractiveness of scattered manufacturing systems or concentrated urban services. The increasing number of pupils enrolled in the metropolitan belt requires a tuning of the educational offer at least at provincial level, to avoid some concentrations due to the scarcity of schools or the poorness of integration policies. Anyway, overall school segregation in lower secondary education is still far from the levels reached in other Western countries. This is still the outcome of a comprehensive educational system and of a limited residential segregation -a positive but not planned effect, notwithstanding the lack of clear integration and housing policies. Nevertheless, there's still much to study about the micro-segregation within neighbourhoods. Sometimes it is due to residential micro-concentrations, in other cases by filtering-out processes excluding PIBs from schools with a better reputation. With such a blurred regulation, diversity management is mostly dumped on single schools, with national goals hard to be implemented without adequate support and shared practices. This is also what can be expected from a comprehensive system at the ropes in a retrenchment era that jeopardize agencies of social cohesion: actually, even the most well-established targets of educational policies in Italy (like the disabled, to be included in mainstream classes with assistant teachers), those that structured the Italian model of integration in the past, are now undergoing tough retrenchment policies. So, in the management of a disadvantaged condition not yet coped enough in social policies -as migration-related issues -this means an even higher pressure on local networks, since schools do not have "in-house" professionals able to face these new challenges.
In governance terms, we can see that these problems are all in all consistent with a welfare model affected by a traditional gap between insiders and outsiders, with an extensive passive form of subsidiarity, where families and local actors (public and private) have a pivotal role in supporting people in need, though without adequate institutional support (Ferrera, 2005;Kazepov, 2008). First, the problem of territorial fragmentation: differences in the provision of services, infrastructures and resources is not just a North/South divide (even though that gap is still a fundamental problem) and has also a micro-dimension. The problem also refers to new unclear and ambiguous relations between State and Regions within the frame of the principle of subsidiarity: more than ten years after the 2001 Constitutional reform, the Regionalist change is dangerously stuck at a crossroad: the central State is not yet a Federal authority with proper coordination tools, Regions are not yet protagonists due to incomplete transfer of powers, especially as far as finances are concerned. The actual output is wavering between neo-centralist attempts that are going to fail considering the new Constitutional framework and dangerous regional flights filling the gap of State measures on their own, with a further fragmentation that challenges equal rights. Second and related, the issue of coordination. School autonomy and the federal constitutional reform redistributed competences, power and responsibility, though with inconsequential resources, paving the way to a "decentralization of penury" and blame-avoiding strategies in the State retrenchment (Kazepov 2010;Mény & Wright 1985): central authorities can devolve responsibilities but not enough resources, keeping budget under control, and can blame local authorities for not being effective; local authorities can blame central ones for not endowing them enough, and they can go on playing hide-and-seek with likely positive effects in electoral terms, but a risk of jeopardizing effectiveness of involved institutions. What emerges is the shifting boarders of citizenship towards increasingly exclusionary policies characterised by ambiguity and cost-cutting tendencies in which the buck is passed to local authorities which are increasingly going to be blamed for whatever does not work. So the unsaid turning point is the real ability of the State to be enabling towards territorial levels and to guarantee a minimum set of nationwide standards. Too often local actors are required to pool resources that are not available, thus leaving weakest locales to their own destiny, with a micro-local fragmentation -a true postcode lottery. In this respect, school and local autonomy can turn into an inadequate tool, if not properly matched with institutional empowerment, capacitation tools, and resources to cope with devolved tasks.