Internal migration in Italy in recent years: a top-down approach

Internal migration has played a key role in determining the physiognomy of Italy in the post-war period (Golini 1974; Rees et al. 1998; Golini 1999). The most intense emigration from the South (traditionally the poorest area of the peninsula) to the richest regions of the Centre-north (Baldi and Cagiano de Azevedo 1999; Pugliese 2002) took place in the twenty years between 1955 and 1975 (Bonaguidi 1987). The internal migration flows are important and necessary to properly describe the economic and political phenomena during those years (Bonifazi 2013a). In the 1970s and 1980s, there was a reduction in interregional migration flows (Di Comite 1992; Piras and Melis 2007). As a result, scholars began to focus more on international migration flows and neglected the internal ones (Pugliese 2011). However, since the nineties internal migration has regained importance (Matinotti 1993; Bonifazi 1999). Internal migration has radically changed over the years; migratory behaviour has become much more complex and less readable with the old approaches (Bubbico, Morlicchio and Rebeggiani 2011) to the point that many scholars have wondered if the traditional ones are still adequate, or have become insufficient (Raymer, Bonaguidi and Valentini 2009). Abroad, there is also a perceived need to identify new data, approaches, and analysis methods that help to grasp the ongoing changes (Caselli Vallin and Wunsch 2001; Rogers, Raymer and Little 2010). This thesis is based on these considerations and attempts to contribute to the current reflection on internal migration in Italy in recent years. The analysis will focus mainly on: the evolution over time of internal migration, age profiles, gender differences and differences between natives and immigrants. In the first three chapters, a central role will be dedicated to internal migration focusing on the place of birth. Specifically, the place of birth will not only be used to distinguish natives from immigrants, but also to identify the migratory patterns of each Italian region of birth (or Macroregions: Northwest, Northeast, Centre and South). Therefore, “place of birth” is a locution that will indicate individuals born in the 20 Italian regions (born in Piemonte, born in Valle d’Aosta, born in Lombardia, etc.) and individuals born abroad. We do not talk about “region of birth” because we include in our definition also people born abroad. The first chapter serves as an introduction to chapters two and three. It is devoted to describing the procedure utilised to allocate the region of birth to the resident population in Italy. Indeed, the resident population in Italy issued by the National Institute of Statistics (hereafter ISTAT) is also distinguished by birth region only in the census years (as well as by sex, age and region of residence). It was therefore necessary to allocate the place of birth (individuals born in the 20 Italian regions and individuals born abroad) to the resident population in Italy from 2002 to 2015 using the 2001 and 2011 censuses as starting points. The population obtained, therefore, is no longer distinguished only by sex, age and region of residence, but also by place of birth in each of the 14 years in the selected time interval (2002-2015). These estimates, on the one hand, made it possible to calculate the denominator necessary for the construction of the different migration rates by ages (necessary for the construction of the multiregional life table used in chapter 2). On the other hand, we were able to implement an application of the gravity model using as explanatory variables also the distinct population by place of birth (for a more detailed description see chapter 3). After describing the calculation procedures adopted for the allocation of the birth region to the resident population in Italy from 2002 to 2015, a description of the resulting population was made. In particular, we chose to use a lifetime migration approach (Livi Bacci 1999) based on the comparison between the place of birth and the place of residence of the populations obtained. We obtained a structural framework of migration flows and considered those who reside in a different region than the birthplace as a proxy of the migrant population. With this logic, in the first chapter, internal migration in Italy is described, starting from the national level, then the Macroregions (Northwest, Northeast, Centre and South) and finally the regional level. Before outlining each of the chapters, some clarification is needed. As well-known, there is no single definition of what internal migrations are (Livi Bacci 1999; Preston, Heuveline and Guillot 2001). Depending on the target, it may change the type of migration studied. For example, by limiting ourselves to internal migration that involves a change of residence, the definition of internal migration can be based on distance covered by a person moving. In this case, internal migrations can be defined as all changes in residence regardless of the geographical distance covered, or a minimum distance that can be established. Movements between different administrative units in the territory (Macroregions, regions and municipalities) can also identify internal migration. Sometimes, the geographic area is divided into areas that do not correspond to traditional administrative divisions (e.g., moving between urban areas and rural). In this thesis, we will always refer to the shifts between traditional administrative units (Macroregions, regions and municipalities). In each chapter, there is an analysis devoted to a different administrative level and the overall idea is, therefore, to move from a more general level (firstly, a national level and secondly, to the Macroregions), and arrive at an analysis that moves towards a more specific level (migration between municipalities). Except for chapter 1, all chapters are organised in the form of journal articles. Therefore, each of them has a very similar structure: an introduction, a literature review, a commentary on results (sometimes preceded by a section on a descriptive data analysis) and finally conclusions and literature references. For this reason, inevitably, some concepts in a chapter can be repeated in another. A second important clarification concerns the content of chapters. Chapter 1, as explained above, introduces chapters 2 and 3. Therefore, it could not be structured in the form of a journal article. In this chapter the procedure for estimating the place of birth (individuals born in the 20 Italian regions and individuals born abroad) of the resident population in Italy is explained. The models proposed in chapters 2 and 3 will be built using the results obtained in the first chapter. Given that it was not possible to also allocate the municipality of birth to the resident population, in chapter 4 (which deals in particular with migration between municipalities), the variable place of birth will only be used to distinguish the native (born in Italy) from the immigrant (born abroad). While chapter 1 is a necessary premise for subsequent chapters up to the third, chapter 4, which addresses the topic of migration between municipalities, is independent from the other chapters and uses a different source of data (European Labour Force Survey). Specifically, chapter 2 is devoted to the study of migration between Macroregions in Italy from 2002 to 2013. The time span was divided into four periods of 3 years. This choice was made to have equal ample intervals of 3 years and make sure that while distinguishing migrations, apart from gender, age, distribution of origin and destination, distinguishing them also by birthplace, where the number of migration flows was always greater than zero. The approach chosen to study such migrations was that of Rogers’ multiregional life tables (1973). However, having distinguished the resident population by birth region (through the procedure outlined in chapter 1), we were able to build tables that took into account the Macroregion of birth in our collective analysis. Therefore, we used the multiregional life table built using the place of birth dependent approaches (Ledent 1980). Using the place of birth makes the results more accurate because they can take into account the fact that a birthplace is a very important determinant of both international and internal migration (Long and Hansen 1975; Ledent 1980). Through the construction of multiregional life tables, one achieves an accurate measure of internal migration: the years of life expectancy at birth lived in the four Italian Macroregions for each birth cohort. In other words, unlike uniregional life tables, it is possible to follow the story of a cohort of 100,000 people, with regards to not only their mortality but also their internal migration, by observing the years of life expectancy at birth for each birth cohort that lives in each Macroregion. International literature has shown that this indicator is more accurate than traditional migration rates and, in our case, provides more accurate results than multiregional life tables built without taking into account the place of birth (Philipov and Rogers 1981; Halli and Rao 1992; Jozwiak 1992). An analysis of the evolution of internal migration over a period of 13 years through multiregional life tables built using the place of birth dependent approach has no precedent in Italy. What kind of results are achieved by using this model? What are the ages and sex distribution of the birth cohorts in each Macroregion? What is the time evolution that emerges from these analyses? How have migratory models changed over time for each birth cohort distinctly by sex and age? How have the migration flow destinations changed migrations between Macroregions? The use of the multiregional life table will enable us to answer these questions. Additionally in chapter 3, we deal with interregional migrations, but in this case only refer to 2014. However, the approach used tries to go deeper than the analysis in the second chapter. We used the gravity model. As well-known, the Newtonian model applied to migrations is based on direct proportionality to the masses (in our case represented by the populations) and indirect proportionality in relation to the distance between them. As in the previous two chapters, in this case, a central role will be afforded to place of birth (migrants born in the 20 Italian regions and migrants born abroad). In fact, both migration flows and populations will be distinguished by place of birth. More specifically, with reference to migration flows, a matrix will be considered in which, in addition to considering the 20 regions of origin and the 19 destination regions, the matrix will be subdivided into the 21 places of birth. The final vector will thus be 20 * 19 * 21 = 7,980 rows. In addition, in the final model proposed in the chapter, two types of populations will be used in relation to each origin and destination region: the population residents and the distinct population residents according to the place of birth (residents in the region who are born in the same region as those who migrate). A further explanatory variable is information on the gross domestic product (GDP), referring both to the region of origin and destination. The introduction of GDP will allow us to control for the role played by the richness of each Italian region. Finally, the distances will be calculated with regard to the centroids with the Vincenty method (1976). This kind of approach will allow us to ascertain how much and how the internal migration flows are determined by the size of the population residents (also distinguished by place of birth). In addition, we can investigate the role of distance in interregional migration and whether it is different depending on the place of birth. We can also study the role of the geographical location for each region. Finally, the gravity model will allow us to verify, for each region of birth, what role GPD plays in relation to the region of origin and destination. Chapter 4 is the last of the thesis. This section focuses on the analysis of migration between municipalities. The chapter represents the natural end of the path we have set ourselves, namely to proceed from the more general administrative level (migration between Macroregions) to the more particular one (migration between municipalities). Using the data for Italy provided through the European labour force survey (ILFS), the demographic characteristics (sex, age and place of birth), socio-economic characteristics (employment, income, education, area of residence) and household characteristics (type of personal relationship and parenthood) of internal migrants in Italy are analysed using multivariate analysis. In particular, we will study how these characteristics modify the propensity to migrate for the population residents in Italy (Basile and Causi 2005). We included a set of logistic models in order to control for compositional effects and to analyse the main determinants of migration. The same model has been estimated for some subgroups of women, employed, interviewees aged 25-34, born abroad and migration from the South to the Centre-north. The results obtained will enable us to verify which variables are most affected by the migration propensity and whether specific migratory patterns can be identified by demographic, socio-economic and household characteristics. Chapter 4 has already been published. The presentation will be as faithful as possible to the original publication. The only variations are: inserting colour figures (instead of black and white) and a modification of the layout to make it homogeneous with regards to the overall thesis. As already pointed out earlier, chapters 2 and 3 are in the form of journal articles. However, in this case, these are not published articles. For this reason, these sections are characterised by a greater effort to adapt the articles to the explanatory needs of the thesis. There will be references to other chapters of the thesis and in some cases the descriptions will be more detailed than required by an international journal article.