Migration propensity of peripheral youth: insights from Italy

ABSTRACT The social and economic consequences of the Covid-19 pandemic are at risk of exacerbating the pre-existing intergenerational and interregional inequalities between and within countries, such as those between core and peripheral areas. In particular, in the latter, especially in Italy, the lack of opportunities and access to essential services may affect the current and new generations to come, while also compromising the development of the country as a whole. Against this backdrop, understanding young people's aspirations and needs is vital to produce evidence-based knowledge to inform policies promoting opportunities to stay in or return to these territories. This paper investigates young people's migration propensity, uncovering the individual characteristics and the factors explaining their propensity to leave their place or to stay and work there. Data come from a nationwide survey carried out on a representative sample of about 950 young people residing in peripheral areas of Italy, via a questionnaire designed and administered as part of the research-action project ‘Giovani Dentro’. The study provides new and timely information about the difficulties, desires and plans of young people who choose to stay in or return to these territories.


INTRODUCTION
Addressing the economic and social challenges of peripheral areas is on top of the policy agenda in many countries (European Commission, 2022). In this scenario, the role of young people in these places, long affected by depopulation and ageing, has taken central stage. The attraction and retention of young people are indeed increasingly recognized as one of the (if not 'the') key ingredients to slow (and maybe in some places revert) the current dynamics and stimulate economic development (Corcoran et al., 2010;Stockdale, 2006). Larger metropolitan areas, in particular, have attracted the youngest and brightest (Artz, 2003;Gibson & McKenzie, 2012), leaving rural, peripheral areas and smaller towns struggling. The European regional divide replicates a geography of diverging employment accessibility and life chances for young people, undermining the convergence of peripheral Southern regions (Cefalo & Scandurra, 2021). In the Italian Southern regions (i.e., the 'Mezzogiorno'), the long-standing socio-economic backwardness and youth out-migration are persistent challenges fuelling the historical North/ South divide (Cannari et al., 2019; Associazione per lo Sviluppo dell'Industria nel Mezzogiorno (SVIMEZ), 2019; Viesti, 2021).
Many now wonder whether Covid-19 is actually opening up new opportunities for non-core places due to the raised concerns about the fragility and vulnerability of our consolidated way of living (Cotella & Vitale-Brovarone, 2020;Gurrutxaga, 2021). For instance, the changes in individual preferences towards less densely populated locations, with a more accessible housing market, better environmental amenities and quality of life, coupled with the increased possibility of working from home, may influence the migration decisions of young and educated people. Therefore, analysing the determinants of young people's migration propensity, especially to and from peripheral areas, via a fresh understanding, which leverages lifestyles and the sense of belonging and attachment (MacKinnon et al., 2022;Sandbu, 2020;Tomaney, 2015), is of paramount importance, also in light of the discontent spreading in many countries (Faggian et al., 2021;McCann, 2020;Rodríguez-Pose, 2018).
Within a framework of place-sensitive strategies for peripheral areas (Barca et al., 2012;Iammarino et al., 2019), we call for a deeper focus on the relationships between people and 'places' to finally understand how to bring opportunities to stay or return.
In the well-established migration literature, plenty of contributions point to the positive selection of migrants as compared with the general population: it is the youngest, most educated, extrovert, entrepreneurial and risk-taking people who are more likely to leave (Crown et al., 2020;Gabriel & Schmitz, 1995). With reference to young people, most studies focus on the reasons behind the decision to movethat is, life-course events or entry in the labour force coupled with labour market vitality (Faggian & McCann, 2009;Faggian et al., 2017a;Venhorst et al., 2010). However, the number of contributions on the reasons underlying the choice to remain in a place is more limited (Abel et al., 2014;Mellander et al., 2011;Rowe et al., 2015) especially when dealing with peripheral areas.
Our paper analyses young people's migration propensity and the degree of 'voluntariness' of their choice by differentiating between 'opportunity' and 'necessity' behaviours (i.e., people who are firmly convinced versus reluctantly forced to migrate/stay). We thus contribute to fill the current gap on the underexplored motivations behind the 'immobility' option by providing evidence on young people's reasons to stay in the so-called Italian 'inner areas' (IAs), which are peripheral areas suffering from long-term slow-burning processes (de Renzis et al., 2022;Pendall et al., 2010;Pike et al., 2010). 1 At the same time, we deepen the understanding of the factors behind the decision to leave these places.
By using a unique database built on 948 interviews of young people residing in the Italian IAs, we develop a new classification for peripheral young people based on their propensity to leave or stay and on the voluntariness of their choice, thereby outlying four profiles and thus uncovering both the individual characteristics and contextual conditions systematically associated with each of them.
We selected young people living in the Italian peripheral areas, as identified within the Italian National Strategy for Inner Areas (SNAI), one of the first examples of a place-based governmental intervention to actively act on population decline in low-density territories. Exploiting the SNAI 2 classification allows us to rely on a widely accepted and officially recognized definition of peripheral municipalities.
Overall, our paper makes a three-fold innovative contribution to the current policy and scholarly debate. First, it adds to the migration literature by providing a new conceptualization that can be used in future theoretical and empirical research based on four migration/stay categories with reference to the different degree of voluntariness. It brings into this field concepts and ideas from the literature on entrepreneurship with specific reference to opportunity and necessity behaviour in making decisions (Calderon et al., 2017;Giacomin et al., 2011;Wennekers et al., 2005). Second, it develops a new, highly replicable, empirical approach based on an original dataset for the analysis of peripheral youth. Third, these novel results may offer relevant policy guidance on how to address issues concerning Italian and, more generally, European lagging-behind and depopulating places by attracting and retaining young people. The implications of our findings are also relevant in light of some recent post Covid-19 phenomena, such as the change in lifestyles and the related debated issue of the 'Great Resignation'.
The remainder of the paper is structured as follows. Section 2 provides a brief theoretical discussion on the factors that influence migration behaviour and on peripheral youth dynamics. Section 3 presents our unique dataset and describes the methodology. Section 4 reports and discusses the main results. Section 5 concludes.

UNDERSTANDING MIGRATION PATTERNS OF PERIPHERAL YOUTH
2.1. The main determinants of migration In an increasingly globalized world, migration has become more frequent, guided by the aspiration to improve one's position in society (Faggian et al., 2017b) and 'selective' of specific types of individuals, that is, younger and more educated. In this context, youth migration has increasingly attracted attention of both academics and policymakers (Cairns, 2010;United Nations, 2016). What is clear is that these migration patterns are highly uneven across space. Regions, cities and countries have different retaining, attracting, and losing current and future trajectories. A wide and consolidated literature on migration has extensively investigated a diverse set of potential drivers shaping migration decisions. Empirical evidence suggests that migrants differ from non-migrants across a range of observable characteristics. In particular, drawing on the existing scientific knowledge on the topic, it is possible to identify some key personal determinants influencing migration outcomes (Faggian et al., 2017b).
In regional studies literature, it is well known that the level of education (i.e., human capital) is an element playing a key role in increasing the probability to migrate. Highly educated individuals tend to migrate more, for example, people with higher education are more mobile than individuals with only secondary education (Haapanen & Böckerman, 2017;Venhorst et al., 2010). Highly skilled young people are also more prone to relocate multiple times in their lifetime (Faggian & McCann, 2009;Faggian et al., 2017a).
Age also influences migration propensity. Migration is seen as an age-selective process in which the youngest segment of the population tends to migrate more than older counterparts (Faggian et al., 2007). In fact, the peak is reached in young adulthood stage (Bernard et al., 2014). Following human capital theory, young people are more prone to migrate because they have a longer payback period to perceive the return on this investment. In particular, life-course transitions such as entry in education or in labour force, union formation and childbearing, divorce and retirement, are linked to migration age profile, especially when looking at the passage to adulthood (Venhorst et al., 2011). Even gender is assumed to affect migration behaviour. Broadly, men appear to be more mobile than women (Dixon, 2003;Robinson, 1993) despite some exceptions found for female graduates in Great Britain by Faggian et al. (2007) and in Italy by Coniglio and Prota (2008). For women, migration can also be a way to partially reward gender labour market discriminations and to get an earning premium, as founded by Jacobsen and Levin (2000) in the United States.
When looking at the migration history, it was found that previous migration experience, as said, increases the probability to migrate (e.g., only few of those who migrate after graduation return to pre-university location) (Faggian et al., 2007). What is more, the diverse combinations of these featuresceteris paribuscan also influence how people consider and assess regional differences. Finally, Crown et al. (2020) demonstrated that even personality traits (e.g., extraversion, openness to experience, conscientiousness, agreeableness and emotional stability) are significant factors in migration decisions. In sum, the factors that shape migration decisions are multiple and extensively debated. However, migration could be reasonably considered, as led by the interrelationship between context-specific features and individual characteristics (Faggian et al., 2017b).

Mobility and young people in peripheral areas
Addressing depopulation dynamics, ageing population and human capital shortage in marginal territories is of pivotal importance to strengthen cohesion, reduce territorial inequalities and promote sustainable development across the European Union (EU). Because of the decline of the agricultural economy and of the essential living conditions in these territories, demographic shrinkage has affected Italian peripheral (i.e., inner) areas since the late 1930's (Bevilacqua, 2018). This process was accelerated by the economic and industrial growth around the 1950's and 1960's which led to a massive out-migration from the Apennines and from the Mezzogiorno (Bonifazi et al., 2020) and has been persisting till today (ISTAT, 2019(ISTAT, , 2020. More generally, Europe is characterized by a great heterogeneity between countries which may affect migration patterns and the weights associated with the (economic versus non-economic) factors at stake in the decision to migrate (Alvarez et al., 2021). Biagi et al. (2011) showed, for instance, that long-distance movements follow the EU trend in Italy (i.e., from the poorer rural South to the richer and industrialized North), whereas amenities matter in the short distance.
In peripheral areas, poor essential services, combined with the lack of opportunities, often act as push factors: people are encouraged to leave and follow the tide, thereby generating negative underdevelopment spirals. The brain-drain represents the well-known and widely studied consequence of the selective out-migration especially of high-skilled individuals which is part of the wider phenomenon of exodus from peripheral regions (McCann, 2001;Glaeser, 2011). Given the relevant impacts of such a phenomenon, recent studies have focused specifically on high skilled migrants (Faggian et al., 2007;Faggian & McCann, 2009). Rauhut and Johansson (2012) highlight the gender aspect of peripheral migration focusing on rural Sweden: young out-migrants are more frequently women and the lack of employment opportunities in their profession seems to be the main reason for not returning.
The 'stay option' is a phenomenon that has attracted comparatively less attention, especially in rural areas as a result of a prevalent mobility perspective (Nì Laoire, 2001;Teti, 2022). The act of 'staying' is not rarely imbued with negativity, devaluated and perceived as a failure to leave, which is favourably seen as a need to move to proceed in one's own life (Nugin, 2014;Tucker et al., 2013). More recently, there is a call to focus on immobility in order to bring new perspectives shedding light on the complementarity of the choices to stay/migrate which are frequently connected and interrelated, as well as on the factors influencing this decision. Deepening the motivations of those who stay, such as a conscious and positive experience (Schewel & Fransen, 2022;, is thus potentially enlightening. In this regard, some interesting models classifying and identifying different types of immobility have been introduced in literature. For instance, the model of the aspiration-capability (Carling, 2002;Schewel, 2015) is a stepping-stone in immobility theory (Membretti, 2021). It relies on a two-step process: first, the aspiration to migrate; and second, the capacity of the subject to realize it. Consequently, the involuntary immobility identifies the condition of those who wish to migrate (Carling, 2002) whereas the acquiescent immobility that of a lack of clear aspirations to move or stay (Schewel, 2015). Building on this, Mata-Codesal (2018) includes the category of desired immobility to describe those subjects with both the desire and ability to stay who have also acquired the capacity to remain.
As mentioned before, the taxonomy stay/migrate, commonly adopted in migration literature, fails to take into account the nuanced reality behind the act of making this decision. In an increasingly connected and mobile world, stay/migrate attitudes are extremely blurred and dynamic processes : migration is not a one-off decision, people are currently more and more living across lagging and core areas (Crescenzi et al., 2017). Multi-locality is a growing social and spatial phenomenon that combined with daily commuting and the spread of working-from-home possibilities. This might have relevant implications for peripheral areas (Greinke & Lange, 2022).
It has been increasingly acknowledged that in the choice to migrate or to stay non-economic factors play a key role (Rérat, 2014). This decision is far from being guided only by labour market conditions: family still represents one of the most important drivers in location decisions after graduation (Bjerke & Mellander, 2017) and place attachment plays a role in the active choice of staying . Although the offer of job opportunities is at the root of the choice of leaving or returning, Ferrario and Prince (2014) showed that demotivation on future prospects has played a significant role in the case of a peripheral mountain Italian area.
However, against a state of the art of the international empirical literature showing a paucity of studies directly investigating young people's motivations to stay in more peripheral areas or to leave them, the same holds in Italy, with no comprehensive analyses widely exploring the phenomenon.
To date, studies addressing the determinants to stay/migrate in peripheral areas mostly relate on site-specific qualitative case studies mostly concentrated in mountainous areas which have, for example, shed light on the desire of relocating from urban areas to mountains to develop innovative initiatives (Barbera et al., 2019;Membretti, 2021).
Our paper will significantly contribute to produce such missing empirical knowledge, also highlighting within-country dynamics and hence allowing, for the first time, some evidencebased spatial considerations on the geography of the propensity to migrate of Italian young people.

DATA AND METHODOLOGY
The paper draws on an ad hoc, newly developed representative survey on young people living in Italian IAs. The study has been developed within the framework of an interdisciplinary research-action project ('Giovani Dentro' 3 ). Its aim was to investigate the needs and aspirations of the peripheral young generation. This dataset allowed us to capture for the first time the profile of this specific segment of population living in these territories by the degree of voluntariness within the migration/stay choice. We propose a new conceptualization of migration propensity, encompassing different degree of voluntariness. Empirically, this is based on a two-stage approach. First, the basic distinction concerns the attitude to migrate from IAs (migrate option) or to remain there (stay option). Second, for each of these two categories, we identify 'opportunity' and 'necessity' attitudes freely borrowing from the entrepreneurship literature (Calderon et al., 2017;Giacomin et al., 2011;Wennekers et al., 2005).
'Opportunity entrepreneurship' defines those entrepreneurs who create businesses when they see a market opportunity. 'Necessity entrepreneurship' identifies entrepreneurs who start a new business because of the lack of other options in the labour market. In our analysis, we refer to 'opportunity' (migrants or stayers) when people are firmly convinced to migrate/stay because they see future prospects, whereas with the term 'necessity' we identify people who are reluctantly forced to migrate/stay because they feel they have no other option.
Several context-dependent structural variables should be taken into account in the analysis of the determinants of young people's choice to stay versus migrate. In fact, place-based aspects appear to be crucially important and further analysis is needed in this direction to deepen the understanding of these issues. In our work, we develop some first reflections on the distribution of young people's profiles based on their migration propensity at the macro-regional level.
The main aim of this study is therefore to contribute to the ongoing debate by understanding the personal observables characteristics and experiences that underlie young people's attitude to stay in or to migrate from IAs. Once individual traits are scrutinized to investigate the determinants of this choice, personal features and life stages will be examined to gain a full understanding of the issue at stake. The insights of our research might be virtually applicable to other countries coping with depopulation and young out-migration in peripheral areas, especially in the EU. What is more, the high replicability of our survey may pave the way for cross-countries comparative analyses.
Our sample consists of 948 observations collected via an ad hoc survey 4 given in December 2020 to young people aged between 18 and 39 years living in the municipalities defined as IAs (as for SNAI classification 5 ). For each young inhabitant of IAs in our sample, a wide range of variables were collected, including personal features, the region where they are located and, crucially for the scope of this paper, information on their migration history and intentions.
Our classification of migration propensity is defined as follows: . Necessity migrants: all interviewees who, although wishing to stay, are prone to migrate. . Necessity stayers: all interviewees who, although wishing to migrate, are prone to stay. . Opportunity stayers: all interviewees who are convinced to stay and plan their life and work in IAs. . Opportunity migrants: all interviewees who are convinced to leave IAs and are planning to migrate.
This classification has the advantage of being easily replicable in other contexts for further analyses.
After defining our four categories, we proceed as follows: . Descriptive statistics draw a picture of the youth population living in Italian IAs and the geographical patterns of their profiles. . Two multinomial logit models identify the factors affecting their migration propensity, allowing us to compare the characteristics of the different subgroups.
We estimate the following multinomial logit (MNL) model: Equation (1) provides the estimated probability of a young resident in IAs to belong to a certain 'migration propensity category' m compared with the 'base category' b (which in our case is Opportunity stayers) as a function of a series of explanatory variables (vector x). Our four different alternative categories of migration propensities (1. Necessity migrants; 2. Necessity stayers; 3. Opportunity stayers; and 4. Opportunity migrants) are the dependent variables in our model. A wide range of explanatory variables is included in the models accounting for personal characteristics, labour experiences, migration and lifestyle, social capital and EU planning. We also include a set of controls for key geographical features of the residential area of our interviewees so as to ensure that our variables of interest compare young people who are as more similar as possible. Hence, we include macro-region fixed effects (i.e., North-west, North-east, Center, South and Islands).

RESULTS AND DISCUSSION
This section presents the findings of our analysis on the migration propensity of young people living in Italian IAs. First, we start by discussing some descriptive statistics to get a better picture of our sample and detect differences across the four migration propensity categories. Second, we present two multinomial logit models estimating the probability of young people residing in IAs to belong to our defined categories. Our base model controls for key personal characteristics which have been found to influence individuals' migration propensity in the literature. We then develop a full model by adding further variables to catch how young people live and are attached to IAs.

4.1.
Profile and geography of migration propensity: descriptive statistics Thanks to the rich information included in the dataset, descriptive statistics 6 already make an innovative contribution. What emerges when looking at the distribution of the migration propensity categories is far from conventional wisdom. As Table 1 shows, the most common attitude of young people living in IAs is non-migration. More precisely, 53% of interviewed are firmly planning to stay in peripheral areas (Opportunity stayers), whereas only 12% would prefer to leave and live elsewhere (Opportunity migrants). Even more interesting is the figure for Necessity migrants: 16% are planning to leave their peripheral municipality because it does not offer job and life opportunities. However, it would be interesting to investigate if remote working or working from home may nourish the chances to stay for the subjects within this group who now feel forced to migrate. Finally, about 19% will stay due to the lack of alternatives (i.e., Necessity stayers). Taken together these results demonstrate that the majority of Italian young people wish to stay in IAs or they would stay if there were better future prospects. Table 2 shows the profiles associated with each of the four categories. Female interviewees are slightly more prone to remain in IAs, both out of necessity and opportunity, despite some previous studies demonstrating a lower migration propensity of women (Dixon, 2003;Robinson, 1993). Women's behaviour, life and opportunities in non-core areas represent a high priority for future research. The role played by age in migration decisions across a lifetime is confirmed by our data. The choice to migrate (both for necessity and opportunity) is more common among people aged between 18 and 29 (around 66%) vis-à-vis those aged between 29 and 39 years, confirming the highly mobility of the youngest segment (Faggian & McCann, 2009). Family stages and ties are crucial in influencing the propensity to stay as well: among young families 'necessity/ opportunity stay' behaviours are the most common choices. On average, being a graduate is a characteristic with a more homogeneous distribution among our four migration categories. Conversely, when controlling for work status and experience, we get, somewhat unsurprisingly, a heterogeneous picture of young people's migration propensity. In fact, 76.5% out of Opportunity stayers are actually workers. What is worth noting is that even previous work experience abroad seems to influence the propensity to migrate despite the result being different depending on the place this was carried out. On the one hand, having worked in other Italian municipalities is more common among Necessity migrants and Opportunity stayers. On the other hand, around 26% have made their work experience abroad among Opportunity migrants versus 15% of Opportunity stayers. About 18% of Necessity migrants have attended university out of their region of origin. The fact that migration guided by opportunity or necessity is the most common choice among people who have done previous work or university experiences far from their home is consistent with existing literature, confirming that previous migration is crucial when analysing migration propensity and behaviour: people who have already migrated are more efficient at finding information to move again (Faggian et al., 2007). To further investigate what influences the link between the way young people really experience IAs and the propensity to migrate we add further variables in our analysis accounting for their place attachment. On the one hand, we found a greater openness to commute between Necessity stayers, showing either a strong attachment to their IA (also due to family ties) or also the impossibility to permanently live in the city of their workplace because of a higher cost of living. Among Opportunity migrants a higher percentage of individuals is living between two places (e.g., multi-residence). This might be only a temporary choice, since young people when living at the same time two places generally prefer to abandon the IA in which they are currently living the moment they can afford it. On the other hand, the higher participation in social initiatives and volunteering between Stayers clearly point to their greater attachment to place compared with Migrants who participate to a lesser extent.
Looking at the geography of migration propensity, Table 3 does not portray an encouraging picture for some macro-areas across Italy. Although the percentage of Opportunity migrants is homogeneous across macro-areas, when we look at the category of Necessity migrants distributions change. On average, young people living in IAs of the North-east and South are more likely to be Necessity migrants. Young and skilled migration is a well-established phenomenon that affects South of Italy: in 2017, 50.4% of migrants were young and 33.0% of them were graduated, and migration from mountain and rural areas was above the national mean (SVI-MEZ, 2019). Our findings show an interesting picture which goes beyond the well-known North-South divide highlighting a lower propensity to stay in the North-east. On the opposite, in North-west and Center Opportunity stayers prevail.

Factors affecting peripheral young people's migration propensity: empirical model
Drawing on the descriptive statistics presented above, we provided an overview of the general characteristics of young people falling within the same migration category that we have identified to classify their attitude toward migration. Moving onto the empirical results, we shed light on the factors behind the decision to stay in IAs or vice versa to migrate by using a multinomial logit (MNL) model.
First, in our 'base model', we identify to what extent the likelihood for an individual to belong to one of our four categories is related to key general personal features which have been found to affect individual migration propensity in the academic literature, that is, gender, age, marital status, children, education, labour market-related characteristics, namely employment status and work experience (in Italy or abroad). Second, in the 'full model', to seize features that draw from lifestyle and experiences of young people in IA, we exploit data on university migration, commuting willingness and multi-locality while at the same time controlling for social capital and participation in EU opportunities. The description of all the variables used is provided in Table A1 in Appendix A in the supplemental data online.
Results are presented as odds ratio, being more intuitively interpretable. An odds ratio > 1 means that the associated explanatory variable increases the chances of being in that particular category.
The empirical results (Table 4) offer a very clear and consistent diagnosis of the possible determinants affecting young peripheral people's migration propensity.
Starting from the 'base model', we find that gender does not determine migration propensity since the coefficient associated is not significant across our categories. Age instead does. Younger individuals (aged between 18 and 29 years) are 90% more likely to belong to the Opportunity migrant category. This result is in line with the human capital migration theory and previous empirical findings for the UK (Faggian & McCann, 2009). However, they are more likely to show Opportunity stayer attitude over Necessity stayer than their older counterparts.
The odds ratios are shown in Figure A1 in Appendix A in the supplemental data online, 7 where it is possible to rapidly look at the magnitude and significance of the effects. For example, looking at the second line for 'age 18-29', we can see that the second and fourth points above are the two extremes with no connecting lines. This means that Opportunity migrants are in general younger than all other categories and particularly than Necessity stayers.
This result calls for a more detailed analysis able to grasp attitudes that can differ even across small age population's segments.
The role of family ties is detected through two variables: the positive and significant coefficients of the marital status suggest that young people that are married are more likely to belong to the Opportunity stayer category whereas those who have children are 71% more likely to be  Note: Coefficients are statistically significant at the *p < 0.05, **p < 0.01 and ***p < 0.001 levels.

718
Giulia Valeria Sonzogno et al. REGIONAL STUDIES, REGIONAL SCIENCE Necessity stayers. The necessity attitude may indicate that young parents may suffer from the lack of schools, assistance and healthcare revealing the well-known weaknesses of peripheral areas in terms of essential services provision and parenting facilities. However, these results point to a higher propensity to stay in IAs among young people who have already started their own family which is guided by both necessity and opportunity. Conversely, having completed tertiary education does not seem to affect migration propensity. Surprisingly, this result is in contrast with the well-established literature on migration (see Section 2). This could be partially because our sample is rather homogeneous, since it is mostly composed of high-skilled young people. On the other hand, this result might also highlight the relative importance of other factors at stake when looking at the decision to stay/migrate of young people living in IAs.
When controlling for labour market characteristics, the significant and positive coefficient associated with the employment status indicates that young people, who are employed, are more likely to be non-migrant. Therefore, being employed, and hence having positive prospects, increases the probability to stay with conviction. Figure A2 in Appendix A in the supplemental data online makes the magnitude and significance of this effect immediately visible, since the absence of connecting lines between adjacent categories clearly shows that being employed has a significantly different effect on them. In particular, being employed decreases the odds of being a Necessity stayer, a Necessity migrant or an Opportunity migrant relative to being an Opportunity stayer.
Once acquired that being employed increases the probability to stay, further research may deepen the implications of a systemic and long-term diffusion of working from home and remote working in terms of population growth and retention for peripheral areas.
In contrast, previous work experience outside the place of residence seems to have different impacts on the migration propensity depending on the location. The positive and significant coefficient > 1 suggests that young people who have done work experience in other Italian municipalities are 60% more likely to be Necessity migrants while being at the same time 49% more likely to be Necessity stayers. In contrast, those who have an international work experience in their career path are more than twice likely to be Opportunity migrants. Figure A2 in Appendix A in the supplemental data online shows that having undertaken international work experience increases the odds of being an Opportunity migrant relative to the other three categories. These results, taken together, are overall consistent with existing literature (Faggian et al., 2007) demonstrating that having done some experience outside one's place of origin/residence clearly influences the propensity to migrate. Those who have 'returned' are in fact always more prone to migrate and see less opportunities in staying.
Finally, some results from our geographical control variables are worth mentioning (see Figure A3 in Appendix A in the supplemental data online). Very interestingly, we find that young people who reside in IAs in the Central macro-area are more prone to be non-migrants. In fact, in this macro-area young people are more likely to stay and develop their lives there compared with people living in the other macro-areas. IAs of the Centre of Italy are therefore more able to retain young population. This result could be because young people residing in Central Italy might benefit from a more connected and polycentric environment (i.e., Rome, the capital, and other important medium-sized cities) compared with other macro-areas, especially Southern regions and the islands (Viesti, 2021). However, in order to better understand these geographical patterns, further investigation is needed to provide specific evidence on the local favourable conditions influencing this positive outcome.
Although our 'base model' demonstrated the role of individual features in migration propensity, the following results shed light on the relevance of individuals' relationship with 'place', providing evidence on the need to account for it in regional development (MacKinnon et al., 2022). In our 'full model', the coefficients associated with individual features confirm previous findings getting, however, generally less significant when including the new variables.
Interestingly, the coefficient associated with human capital is though now significant revealing that the decision to 'stay' made by high-skilled individuals is a fully convinced one. In particular, those who have chosen and attended university out of the region of residence are about 95% more likely to be Necessity migrants and Necessity stayers. This result is consistent with previous findings, which highlight that those who have returned after an experience outside of the place where they now live (be it for study/university or working reasons) are generally more prone to migrate (or they would have preferred it). We found a similar trend also among those who are more prone to commute in the future. In particular, individuals who are open to commute are nearly three times likely to be Necessity stayers. In other words, when there is no chance to migrate, young people are disposed to commute. It could be possible that Necessity stayers are more prone to commute because firmly convinced (or resigned) that there are no opportunities and services in their place of origin. The result on multi-locality is extremely interesting: those who already live between two municipalities are more than twice as likely to be Opportunity migrants. Figure A4 in Appendix A in the supplemental data online clearly shows that category 4 (Opportunity migrant) is the furthest on the right, with no connecting lines to the other three categories, showing that living between two municipalities increases the probability of being Opportunity migrants. The significance of the coefficients associated with the variables accounting for social capital (Putnam et al., 1993) and the participation to EU tenders and opportunities calls for a more in-depth analysis and a general greater consideration of these issues. Young people who are already active in local and national social organizations are more than twice likely to be Necessity stayers. On the one hand, social activism may produce the 'attachment and belonging' (or is the consequence of) which increases this propensity. On the other hand, the necessity attitude may reveal the perceived difficulties connected to this choice. Contrary to expectations, having already participated in EU initiatives (e.g., Erasmus programmes or tenders) increases the probability of migrating by necessity by 95%. These individuals may potentially have experienced problems in applying to EU projects in IA, or they were not able to find locally technical information or assistance, or these opportunities are not perceived as ensuring long-term prospects.
Our analysis identified the personal characteristics determining the migration propensity of young people living in peripheral areas, also revealing the importance of properly taking into account the broad and complex dimensions linked to place attachment and to the way individuals actually 'live the place'.

CONCLUSIONS
Decades of depopulation and ageing, negative spirals of underdevelopment and the perception of a neglected future that characterize peripheral areas have put the stability and prosperity of Europe at stake. Despite the increasing awareness of the importance of developing specific policies to tackle, and possibly invert, the actual trends, the challenges for these places are still enormous. Traditional policies had limited returns in bringing solid opportunities to develop a better future for these areas. An adequate provision of basic services to revert population decline is part of the story, but not the whole story. What needs to be done urgently is finding ways to stop the continuous draining of the youngest and brightest population segment from these peripheral areas. However, the scholarly and policy debates have so far failed to provide timely and concrete answers to face this urgent need. Also as a result of Covid-19, new dynamics and ways of living, involving especially the youngest segment of the population, are currently spreading across peripheral areas with no evidence to guide practical policy interventions to match their needs and assist their aspiration to stay or return. More generally, a conceptual and empirical framework guiding policy answers especially targeted at young people is still missing. This paper, by leveraging novel and fresh data on young people residing in Italian IAs, addresses a primary literature gap by looking at the determinants of migration propensity of peripheral youth (guided by opportunity or necessity) as a fundamental, even if not sufficient, element to be aware of. Although further research is needed to investigate in depth the dynamics emerging from our preliminary analysis, our findings shed light on a series of policy implications which are valid for Italian IAs, but also, more generally, for all peripheral territories in the EU.
First, since the majority of young people are prone to stay (even the high-skilled ones), the implementation of a place-sensitive intervention targeted to this segment of the population in peripheral areas is an urgent matter. Grasping this desire by leveraging young people's full potential and expertise, providing tangible opportunities, is crucial. Second, interventions able to stimulate opportunities to return after experiences outside could reconcile the need of thinking 'out of the box' with the desire to come back.
Ideally, measures should be coupled with labour market related facilities and specific education on the ground, even to attract new inhabitants and retain those who are now living between two places.
In addition, in light of the potentially strong effects of Covid-19 on labour market dynamics (i.e., increasing remote working or working-from-home options) (INAPP, 2022), our evidence sheds light on the potential positive implications of the above-mentioned working-related measuresif coupled with specific interventions to regulate these new trendson nourishing the chances to stay for those who now feel forced to migrate (i.e., Necessity migrants). Third, it is worth mentioning that, as for our survey, young parents are staying though wishing to migrate. Therefore, parenting support facilities may nourish their propensity to stay also attracting new family units. Finally, a bit unexpectedly, EU opportunities/tenders seem to fail in providing sustainable long-term chances to stay, undermining cohesion and future prospects, thereby potentially fuelling discontent in peripheral regions. In order to enhance tangible impacts of the EU instruments, our findings call for an urgent and thorough understanding of the reasons behind their apparent ineffectiveness among young people and of the current main hindrances to their full exploitation, so as to more efficiently address the mismatch between policy supply and demand that young people are likely to face in their attempts to seize the opportunities provided by the EU.
Finally, despite we still lack understanding of all forces at stake in shaping migration propensity, our results point to an unquestionable role of the relationship with, and attachment to, places.
Broadly speaking, a key starting point in addressing this aspect would imply to expand the policy approach to regional development in order to stimulate the liveability of these areas and the sense of belonging to them. In other words, place-based policy interventions to improve essential services and economic growth must be coupled with initiatives aimed at rebuilding and nurturing a positive relationship with the place.
Empowering and engaging local communitiesespecially the youngest and dynamic segment of the populationin policy design and implementation must be at the root of new and better policies. Giving new generations the keys of their future by leveraging their strategic involvement in the processes of economic development and (green and digital) transitions, especially in lagging behind areas, is the steppingstone for the EU of tomorrow.

NOTES
1 The Italian definition of 'inner areas' recalls the one of 'inner peripheries', which is widely used at the European level. These are identified, following the European Observation Network for Territorial Development and Cohesion (ESPON) (2017), as: (1) enclaves of low economic potential, (2) areas with poor access to services of general interest or (3) areas experiencing a lack of relational proximity. As most peripheral areas in Europe, Italian inner ones are challenged by slow-burning processes (e.g., depopulation, high-skilled outmigration, deindustrialization, prolonged recession, etc.) as a result of their marginal condition. Slow-burning processes are phenomena that differ from acute shocks (e.g., earthquakes) because they are chronic, corrosive and persistent pressures that undermine the future prospects of places. More generally, in the resilience literature, these processes are assumed to result from long-term depletion and deterioration of endogenous resources. 2 In line with the ESPON (2017) classification, SNAI has developed an innovative framework to classify Italian municipalities by relying on their level of peripherality, measured as travel time distance from (essential) service provision centres (healthcare, education, mobility). Since its launch (2014), it has involved 1077 municipalities (corresponding to 72 project areas) to promote development and territorial cohesion, by tackling depopulation and marginalization dynamics. 3 Giovani Dentro is a research-action project promoted and coordinated by the association Riabitare l'Italia, financed by the Fondazione Peppino Vismara and CoopFond and other cofinancing partners (Gran Sasso Science Institute, CREA, Osservatorio Giovani -Università di Salerno, CPS -Università di Torino, Eurach Research). 4 The survey was carried out by SWG S.p.A. by leveraging a mixed technique, that is, computer-assisted telephone/mobile interview (CATI/CAMI) and online via computer-assisted web interview (CAWI). Our sample is composed of 948 young people stratified by representative shares of the population aged 18-39 years with respect to gender (male, female), age (18-29, 30-39 years) and residence within the Italian macro-regions (North-west, North-east, Center, South and Islands). 5 SNAI classified all Italian municipalities according to their provision of essential services (i.e., healthcare, education, mobility) into six categories (Agenzia per la Coesione Territoriale (ACT), 2014). First, municipalities serving as service provision centres (A) or a group of neighbouring municipalities (B) were identified as capable of guaranteeing: (i) a full range of options for secondary education, (ii) at least one emergency care hospital and (iii) one railway station. Second, the remaining municipalities were classified in four categories according to traveltime distance to the above-mentioned service provision municipalities (i.e., A, B): outlying areas within 20 min (C); intermediate areas between 20 and 40 min (D), peripheral areas (between 40 and 75 min) and ultra-peripheral areas at over 75 min distance (E). Hence, by relying on this classification, IAs are those municipalities that are labelled as intermediate (D), peripheral (E) and ultra-peripheral (F). 6 Our descriptive statistics are based on the observations for which the information on migration propensity was available (amounting to 886 out of 948 observations, i.e., the full sample). 7 An odds ratio plot allows one to uncover patterns in results for MNL model. In an odds ratio plot the independent variables are each represented on a separate row, and the horizontal axis indicates the relative magnitude of the coefficients associated with each outcome. These plots reveal a great deal of information (for more details, see Long & Freese, 2006). To begin, if a category is to the right of another category, it indicates that increases in the independent variable make the outcome to the right more likely. Also, the distance between each pair of categories indicates the magnitude of the effect. When a line connects a pair of categories, this indicates a lack of statistical significance for this particular coefficient, suggesting that these two outcomes are connected.
partners Fondazione Peppino Vismara and CoopFond for their economic support, and to all other participants in it (Gran Sasso Science Institute, CREA, Osservatorio Giovani -Università di Salerno, CPS -Università di Torino, Eurach Research) for co-financing and the precious reflections and comments developed within the scientific committee. A preliminary version of this paper was published in the GSSI Discussion Paper series in Regional Science & Economic Geography (RSEG). All errors and omissions are the authors' own.

DISCLOSURE STATEMENT
No potential conflict of interest was reported by the authors.

FUNDING
This work was supported by Coopfond -Fondo Mutualistico Legacoop and Fondazione Peppino Vismara.