What role does timing play in migrants’ transition to marriage? A comparison between endogamous and exogamous marriages

Couple formation and migration are the result of interrelated decision-making processes in the life cycle. Using data from the “Social Condition and Integration of Foreign Citizens (SCIF)” survey, conducted in Italy in 2011-2012 by Istat, we aim to investigate how the timing of migration events affects the type and timing of marriages in the destination country. Time-related models investigate the competing-risk transitions to endogamous and exogamous marriages with Italian spouses. Obtained results provide evidence of the complexity of today’s migrations, and they indicate the coexistence of various patterns among first-generation migrants in Italy, characterised by a plurality of origins, with different projects and behavioural models. The “interrelation of events” hypothesis explains the transitions to both endogamous and exogamous marriages among women, while men usually spend more time finding a partner and achieving economic stability. Despite this general picture, our analysis shows different and original pathways shaping transitions to marriage by reason of migration and considering a number of demographic and migratory characteristics.


Introduction
In Italy, immigration is a more recent phenomenon than in northern European countries, but nevertheless, the first generation of migrants has settled, as highlighted by official sources and literature (Barbiano di Belgiojoso et al. 2008;Ambrosini et al. 2014;Perez 2018). One expression of this change is the increased number of marriages with at least one foreign-born partner, which now constitute a significant percentage of stock data: as of 2017, 14.5% of marriages consisted of couples with at least one foreign-born partner. In particular, before the economic crisis, register data (Gabrielli/Paterno 2016; showed notable growth of marriages performed in Italy between an Italian-born partner and a foreign-born partner (hereinafter defined as exogamous WIT -with an Italian -marriage): exogamous WIT marriages of immigrants increased from 9,900 in 1996 to 24,600 in 2008. They subsequently reduced to less than 18,000 per year during the economic crisis (2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015) and then increased again in the next two observed years (reaching 20,000 in 2017). In 2017, they represented 10.5% of total marriages and 72.1% of marriages with at least one foreign-born partner.
In particular, referring to marriages performed in Italy, official data shows that those between an Italian man and a foreign woman are more predominant than those between an Italian woman and a foreign man (77.2% and 22.8% of total exogamous WIT marriages, respectively). The 2017 ranking of exogamous WIT marriages by area of birth of the foreign partner places marriages with partners from the EU in first place (more than 6,000), followed by marriages with partners from Central and Eastern non-EU Europe (5,700), and those with partners from Central and South America (3,700). As highlighted at the international level (e.g., Kofman 2004), the combination of origin and gender results is crucial to understand family dynamics, and this is also confirmed for Italy. The composition of exogamous WIT marriages in Italy by gender and country of origin of the migrant partner is determined by a complex set of factors, such as: different origins, recent feminisation of the inflows (Impicciatore/Strozza 2015), strategies of integration by gender, characteristics of the "marriage market", and exogamous/endogamous attitudes of immigrant communities (Rossi/Strozza 2007).
In Italy, despite the significant growth of partnerships and marriages with at least one foreign partner, it was only 10 years ago that quantitative research began to focus on this topic (Dalla Zuanna et al. 2009;Conti et al. 2013), with the cross-sectional approach being predominant. Using information obtained from ad hoc surveys (often at regional or local level), scholars have described the different typologies of families and households and their characteristics (Terzera 2006;Simoni/Zucca 2007;Tognetti Bordogna 2011), focusing mainly on partner choice rather than union timing (Lagomarsino 2006;Bonizzoni 2007Bonizzoni , 2015Ambrosini 2008). In particular, following the "status exchange theory", numerous studies looked at the matching of partners in couples formed between Italians and foreigners, identifying exchange forms between partners (Maffioli et al. 2012(Maffioli et al. , 2014Guetto/Azzolini 2015;Gabrielli/Paterno 2016;Azzolini/Guetto 2015, 2017. Focusing on individuals rather than on the couple dynamics and mate-matching choices, our paper aims at providing original results to the existing literature by analysing the role of the timing of the competing transitions to first exogamous WIT marriage and to first endogamous marriage of migrants who arrived to Italy unmarried. As far as we know, it is still a neglected topic in Italy, because the transition to marriage has been studied predominantly at the local level and/or considering marriages between migrant partners with the same origins: endogamous marriages (Bonomi/ Terzera 2003;Terzera 2006;Gabrielli et al. 2019).
In particular, we focus on migrants who were aged 16 over and were unmarried on arrival in Italy, and we analyse the timing of their marriage in Italy by adopting an event history approach. We use data from the multipurpose household sample survey on "Social Condition and Integration of Foreign Citizens (SCIF)", which at the time of writing represents a unique source of data representative at the national level with retrospective in-depth information about migrants and their subsequent transition to first marriage (Perez 2018).
This article is structured as follows. The next section contains the theoretical framework and our research hypotheses. The third section introduces the survey data, the sample characteristics, the method of analysis, and the variables. Subsequent sections present the results of the descriptive and multivariate analyses. The final section discusses the main results.

Theoretical background
International research (Kofman 2004;González-Ferrer 2006;Kulu/Milewski 2007) has focused increasing attention on the relationship between couple formation and migration, as well as the mechanisms that affect mate-matching processes. These topics first attracted interest among researchers in North America and the Asia-Pacific region, and subsequently in Europe, highlighting the important role of different socio-demographic dimensions (Mulder/Wagner 1993;Clark et al. 2009;Kalter/Schroedter 2010;Baykara-Krumme 2017;Azzolini/Guetto 2017;Hannemann et al. 2018;González-Ferrer et al. 2018).
One crucial aspect of marriage among migrants concerns the timing of the migration and marriage events. Elder (1985) is one of the first to use a temporal perspective in understanding demographic events: "life course theory tells how lives are influenced by their life stages in the historical context and … how transitions in life cycle consequences affect subsequent transitions" (Elder 1998:9). Although this approach has become widespread, only recent research has focused on the mechanisms by which migration affects subsequent life events, such as marriage and having children (Kulu/Milewski 2007;Lopez Ramirez 2009;Kalter/Schroedter 2010;Baykara-Krumme 2017;Gabrielli et al. 2019).
In contrast to other hypotheses 1 , the "interrelation of events" hypothesis refers specifically to the timing of events. It relies on the interdependencies of the family and migratory processes in the life course of individuals (Kulu/Milewski 2007). In particular, this hypothesis observes couple formation and migration, focusing on the interrelationship between both events, considered as important "parallel paths" that shape the life trajectories of migrants (Mulder/Wagner 1993;Milewski 2007Milewski , 2010. Observing that migration can be motivated by marriage, although marriage timing and individual "marriageability" can also be affected by migration, this hypothesis assumes that timing of migration and couple formation may be conflated because both decisions are undertaken jointly (Hill 2004;Lindstrom/Giorguli Saucedo 2007;Sánchez-Domínguez et al. 2011). Therefore, migrants have a high probability of getting married soon after their migration when the main aim of migration is forming a family (Stark 1988;Boyle et al. 1998;González-Ferrer 2006). In particular, Niedomysl et al. (2010) defined "marriage migration" as marriages that occur within one calendar year after arrival. This pattern is somewhat reinforced by Italian law, which grants legal status to those individuals that migrate for family reunification or that marry an Italian partner.
In addition, some studies highlight the relationship between migration intentions and family-related events (Erdala/Ezzati 2014; Bettin et al. 2018; Barbiano di Belgiojoso/ Terzera 2018). Therefore, intentions on arrival can affect the timing of family events. For example, an intended short stay in the host country, or an intention to move on to a third country, can discourage family events in the short term.
Similarly, the "disruption hypothesis" suggests that the migration itself, as well as the periods preceding and following it, is a stressful event due to drastic changes in daily life and an interruption of social networks. In other words, in the period immediately after migration, family events might be postponed due to the disruptive factors and difficulties related to the migration itself, or to the new environment (Toulemon 2004).
The disruption hypothesis has been formulated to explain the effect of migration on having children (see, e.g., Milewski 2007Milewski , 2010. However, the disruptive effect of the migration may also impact other relevant life events, such as marriage. On the one hand, migration may entail a longer time finding a partner, especially when the reason for migration is less directly linked to union formation (Carlson 1985;Landale 1994;Parrado 2004;Parrado/Flippen 2005;Kulu/Milewski 2007). On the other hand, according to Italian law, formal marriage is one of the requirements for proceeding with legal reunification of the foreign partner or for naturalisation when it concerns an Italian partner. Postponement of marriage because of migration would presumably only affect couples where both partners can migrate autonomously (i.e., EU nationals).
There is consensus in the international literature on the role played by gender and origin background both in the marital trajectories and in the choice of partners (Kofman 1999;Tognetti Bordogna 2004;González-Ferrer 2007;Clark et al. 2009; Ishizawa/Stevens 2011; Barbiano di Belgiojoso/Terzera 2018).
Many studies (e.g., Stevens/Swicegood 1987;Qian/Lichter 2007) showed that, traditionally, men mostly migrate independently and for work-related reasons, while women are more likely to move for family-related reasons (Mincer 1978;Heering et al. 2004). Notwithstanding, some studies have shown that women also migrate in order to secure a better quality of life, which could include not only access to higher wages but also access to wider "partner" markets and freedom from restrictive social environments (Castles/Miller 2003;Hill 2004;Sánchez-Domínguez et al. 2011;Esteve/Bueno 2012;González-Ferrer et al. 2016). From this perspective, the determinants of the spouse selection among immigrants seem to be strongly gendered (Sassler 2005; Adserà/Chiswick 2007; Sohel/Yahirun 2011).
With a geo-cultural approach, Therborn (2004) conceptualises gender patterns, ranking the countries from traditional to egalitarian. Gender indicators are simple classification tools used to distinguish the gender profile of countries. The Gender Inequality Index (GII), measuring the inequality between the achievements of women and men in three dimensions (reproductive health, empowerment and the labour market), well reflects this geo-cultural approach. The 2012 GII shows that Morocco, the Indian subcontinent and sub-Saharan Africa are the countries and regions with the greatest gender disadvantages (i.e., greater inequalities and less egalitarianism), as compared to advanced-economy countries and some Eastern European countries (Table 1).
This offers an interpretative key referring to migration patterns. For example, on one hand, in contexts characterised by traditional values and less egalitarian gender norms, men are typically the breadwinners, and the pattern of "importing brides" is often prevalent (González-Ferrer 2006). On the other hand, in more egalitarian contexts, women migrate autonomously or are often the forerunner of the migration (Ambrosini 2015). Moreover, in recent analyses (Bijwaard 2010), migrants are distinguished by area of origin, taking "the different cultural and background characteristics" into account, and this distinction often shows diversity in family behaviours (González-Ferrer et al. 2016:3). In this regard, some studies on this topic have restricted analysis to a single country of origin (Milewski/Hamel 2010;Hamel et al. 2012;Esteve/Bueno 2012;Baykara-Krumme 2017), while others considered more than one country of origin (e.g., Guetto/Azzolini 2015). The first approach allows the specificities of the country of origin to be taken into account, but it impedes a generalisation and comparison among family migration models that, conversely, is made possible by the second approach, which is preferred when investigating migrants' life courses (Clark et al. 2009;Vitali/Arpino 2015). The place of birth is particularly significant in the Italian context, as it is characterised by a wide diversity of immigrants' origins (Barbiano di Belgiojoso /Terzera 2016Gabrielli et al. 2019).
The empirical evidence on this topic identified the individual characteristics that tend to be associated with endogamous or exogamous marriage. The literature recognises that three factors affect the choice between endogamous and exogamous marriage: structural opportunities for meeting members of one's own group, the influence of "third parties" (Kalmijn 1998:400) on marriage behaviour, and individual preferences to marry within or outside the group (Kalmijn 1998). Although the distinction between the three effects is theoretically meaningful, in practice, indicators for any of the different dimensions sometimes include elements of the others (Kalmijin/van Tubergen 2010).
Structural factors are significant determinants of exogamous unions and shape the available opportunities for getting married inside or outside the group (Blau/Schwartz 1984). Members of larger groups are more likely to form endogamous unions, since they more often meet members of the same group; skewed sex ratios can constitute a structural force towards out-marriage, and this unbalance seems to matter more for men than for women in shaping their partner choices, both for native and immigrant partners (Pagnini/Morgan 1990; Schroedter/Rössel 2013). A synthetic overview of the percentages of residents and the sex ratio of migrants in Italy by country/area of origin highlights that the overall presence is very fragmented, and that among many countries/areas of departure, the sex ratio of the different collectives is unbalanced (Table 1).
The second factor -the influence of "third parties" on marriage behaviour -is represented mainly by the cultural and religious orientation of the national-origin group and is associated with values, beliefs and practices mainly spread by family and religious community (Kalmijn/van Tubergen 2006. This dimension is discussed using the gender roles and country of birth as a proxy for cultural background (Kofman 1999;González-Ferrer 2007;Clark et al. 2009;Ishizawa/ Stevens 2011).
The third and most important factor comprises people's norms, values and preferences regarding interaction and marriage. It was assumed that an unmarried person searches for a potential spouse who is attractive in terms of socio-economic and cultural resources (Kalmijin/van Tubergen 2010).
Lastly, the crucial role of human capital also emerged in other perspectives. Some studies (Lehrer 1998;van Tubergen/Maas 2007) showed that the most highly educated migrants are often encouraged to search for potential partners outside their own ethnic group.

Research hypotheses
In line with the international literature synthesised so far and the characteristics of the available data, we try to identify the timing of marriage after migration and the main determinants of migrants' competing risks of transition to first exogamous WIT marriage and to first endogamous marriage in Italy.
We expect different time-related patterns by gender (H1). Among women, due to their being less frequently autonomous migrants and more often family-orientated migrants, we expect to find evidence of the interrelation of marriage and migration with the highest probability of marriage shortly after migration. Conversely, among male migrants, we expect to find a low probability of getting married in the first period after migration, followed by a recovery at a later date in the migration experience according to the disruption theory hypothesis. Compared to women, men's migration is more often because of reasons other than marriage, and they therefore spend more time before deciding to form a family and then searching for a partner.
With regard to the fundamental role of the reason of migration in the study of timing of familial events in migration, we expect that when the reason is to form a family, migration and marriage should be very close (H2). In other words, affective reasons at migration (including engagement, affective relationship and informal partnerships) indicate that migration and family formation are part of the same project. Therefore, they foster the transition to marriage, reducing the time between migration and (endogamous and exogamous) marriage regardless of gender. Concerning other reasons for migrating to Italy, we expect them to have the effect of postponing the transition to both endogamous and exogamous WIT marriages, as these migrants have aims other than starting a family. Similarly, the initial migratory reason as well as future migratory intentions should have an additional effect on the transition to marriage: the intention to stay in Italy permanently indicates that migration and family formation are part of the same project; conversely, temporary or undefined initial intentions should indicate the opposite.
We expect that the origin background, expressed by the country of birth, strengthens the impact of gender in the transition to first marriage, as well as the inclination towards endogamous or exogamous marriage. On the one hand, the probability of getting married is strongly determined by norms, culture and gender roles in the country of origin. On the other hand, the combination of gender and origin refers also to the structural characteristics of the population in the destination country (size, gender balance, macro-area of resi-dence). Following the literature, we hypothesise that the "cultural background" dimension prevails among women, while the "structural characteristics" affect mainly men. Women tend to link migratory and family projects more frequently than men, and they are therefore more conditioned by the cultural context of their origin. Among men, this link (family-migration) is less direct, and so the conditions of one's own community in the host country affect the transition to marriage more frequently than for women. In particular, we hypothesise that migrant women coming from countries with more egalitarian gender roles have a higher probability of exogamous marriage (H3). Vice versa, we expect that women coming from contexts with more traditional gender roles have a higher probability of endogamous marriage.
We do not expect similar results among men, but as a consequence of the structural characteristics of the migrant group in the destination country, we expect a higher propensity to endogamous marriage among male members of the most sizeable communities in Italy and to exogamous marriage among male members of the most gender-unbalanced communities (H4).
According to the literature and besides our main hypotheses, we also test if some human capital characteristics (level of first employment in Italy and educational level) play a different role in the transition to endogamous and exogamous WIT marriages. In particular, we expect that migrants with high human capital anticipate exogamous WIT marriage and delay endogamous marriage. Following the literature, migrants with a high education level and/or a highly skilled job on arrival have fewer options to find a partner with the same human capital within the endogamous marriage market and so orientate themselves to the Italian marriage market, which is the largest market and where these characteristics are a "bargaining chip" in the transition to marriage. At the same time, migrants with low human capital are expected to have higher probabilities of marrying, with marriage being a way to increase their economic condition.
Finally, we assume that a family's agreement to allow migration is considered as one of voices of the "third parties": the family. Migration is often a choice negotiated within the family (Milewski/Huschek 2015;Zantvliet et al. 2015;Caarls/Mazzuccato 2016), and therefore we expect that the family's agreement (the parents' agreement in our case) facilitates the transition to marriage when there is a family migration project. However, we hypothesise an overall positive effect of disagreement from the family on the transition to exogamous marriages. If the migration is a family plan, a marriage with someone external to the ethnic group is more likely to trigger family disagreement than marriage to a partner of the same origin, with the disagreement being based on the partner rather than the migration, while if the migration is not a family project, family disagreement could encourage migrants to search for potential partners outside their own ethnic group.

Data and methods
Data was taken from the "Social Condition and Integration of Foreign Citizens (SCIF)" survey, conducted by the Italian National Institute of Statistics (Istat) in 2011-2012. Among the Italian multipurpose household surveys, it is the only one designed to collect data on families with at least one foreign citizen and to provide original information on foreign nationals living in Italy. The cross-sectional survey covers a random sample of about 9,000 households and provides information on the living conditions, behaviours, characteristics, attitudes and opinions of foreign citizens in Italy, including individual retrospective histories on migration, family formation, education and occupation.
In our analyses, we consider a net sub-sample of 6,595 migrant adults (aged 18 and over at interview) who arrived unmarried (thus, never married before migration) in Italy at the age of 16 or over. 2 We analyse, as events, the self-reported first marriage between foreign and native individuals (exogamous WIT marriage), and, alternatively, entry into first marriage between migrant partners of the same country of origin (endogamous marriage) 3 : 882 individuals experienced a first exogamous WIT marriage and 2,612 individuals experienced a first endogamous marriage. In addition, we consider, only for descriptive purposes, 277 immigrants that had a marriage with a migrant partner from a different country of origin. It should be noted that both types of marriages concluded in Italy and elsewhere are included in the analysis: once arrived, the migrant can get married in Italy, or can temporarily return to the country of origin to conclude the marriage there.
In the descriptive analyses, we use weighted data in order to provide results which are representative for the population residing in Italy. After a cross-sectional description of marriage typologies by gender and area of origin, we use an event history approach to observe the timing of occurrence of (endogamous or exogamous WIT) marriages after the migration event (Courgeau 1989;Courgeau/Lelievre 2006;Kulu/Milewski 2007). We emphasise the gender dimension by conducting our analyses separately for men and women.
In this context, we estimated Kaplan-Meier survivor functions as well as piecewise constant exponential models to study the competing risk transitions to exogamous WIT marriage and endogamous marriage. The duration until the first marriage is operationalised in terms of time passed since arrival in Italy, and respondents were right censored if they had not yet been married at the time of the survey. We also right censored (at the time of their marriage) those individuals whose first exogamous marriage was with a non-Italian partner.
The basic idea behind employing a piecewise constant exponential model is to split the time axis into intervals, and to assume the transition rates to marriage to be constant in each of them, but potentially changing across them (Blossfeld/Golsch/Rohwer 2007). In particular, we split the first two years after arrival into four intervals (six months each) in order to analyse marriages that occurred immediately after migration (less than one year).
We have chosen this approach as it enables us to account for the intrinsic nonmonotonic shape of the rate along time (recurrent in many social phenomena) and for the impossibility of measuring time-varying covariates. This model's virtue thus lies in its flexibility as an instrument of analysis keeping the mathematical and computational features of the statistical exercise at a reasonable level.
Regarding the predictors, we include the following respondents' characteristics (see Appendix Table 1 for further details) in the models: Country/area of birth -grouped into 7 and 14 categories, respectively, in the Kaplan-Meier survivor estimates and in the parametric analyses (reference Romania) 4 . We define these groups according to the sample size and the similarities of their family dynamics and gender roles.
Educational level -four categories: primary or less (reference, 3-10 years of age), lower secondary level (11-13 years of age), upper secondary level (14-18 years of age), tertiary level. We assume that respondents completed their education (mostly abroad) and reached their highest educational level before marriage.
First job in Italy -four categories: unemployed, unskilled or low level (reference), skilled or middle level (groups 5 and 6 of ISCO), highly skilled or high level (groups 1, 2, 3 and 4 of ISCO). We consider the first job after arrival in Italy and before marriage (if it has occurred). In particular, we defined as "unemployed" the respondents who started their first job after marriage or who have never had a stable occupation in Italy.
Reasons for migration -we include three dummy variables (reference no) regarding the main push factors of migration to Italy, which are not mutually exclusive. Labour reason: includes the "difficulty in finding an adequate job" in the origin country and the "desire to improve income and life wellbeing"; Affective reason: includes "engagement", "marriage", "reunification with relatives" and "escape from family problems" 5 ; Other reasons: includes the remaining reasons (new experiences, study, health, war, persecution, not my decision, other).

Dummy variable for parents' agreement about migration (reference yes).
Migratory project on arrival in Italy -this is separated into three categories: to live in Italy forever (reference permanent intention), to live in Italy for a limited period (temporary intention) and "do not know" (indefinite intention).
Two further control variables are included in the model. They represent, respectively, the age-structure and geographical-settlement characteristics of respondents that we take into account in the models: Age on arrival in Italy -5-year classes (reference 18-22 years). We expect a reversed Ushape pattern along the age classes.
Macro-area of residence on arrival in Italy -four categories: North-East (reference), North-West, Centre, South & Islands. Internal mobility within the macro-area is not significant (only 14% of interviewees changed their residential macro-areas between arrival in Italy and the interview). We expect a significant negative impact of the southern areas of residence on transition to marriage (men and women), with the southern areas being less populated by migrants and a proxy for economic conditions that are usually worse in the south compared to the north.

Descriptive results
Before proceeding with the analysis, it should be noted that in Italy, in 2012, 40% of the over 4 million foreign residents came from three countries: Romania, Albania and Morocco (Table 1). The remaining foreign residents had a variety of origins that are strongly characterised by gender. The main communities have a more balanced sex ratio than the smaller ones (Table 1).
A brief description of the immigrants' matrimonial conditions at the time of the survey appears in Table 2. Among the 6,595 foreign nationals included in our subsample, the majority are married (60.6% vs 39.4% single) and most are married to a spouse from the same origin country (39.4%). Four out of five foreigners in exogamous marriages have an Italian spouse. Women are more often involved than men in exogamous marriages (especially exogamous WIT marriages; 28.6% and 7.0%, respectively). Men are characterised by higher percentages of singles compared to women (42.6% and 35.4%, respectively), as well as a higher percentage of endogamous marriage (44.9% and 32.7%, respectively).  Source: Authors' elaboration using SCIF survey (2011)(2012) A second result concerns the timing of the first marriage after migration, analysed by means of the Kaplan-Meier survival estimator. It shows, along the migration period, the probability of experiencing a first marriage by typology: endogamous and exogamous WIT marriages (Figure 1). Hereinafter, due to the small sample size, we do not consider exogamous marriages between migrants from different countries. The results highlight relevant gendered differences in the timing of the first marriage. More specifically, the transition to endogamous marriage is more rapid among women in the first period of migration (in particular on arrival in Italy and in the following 6 years), and afterwards they become less frequent, while for men, the survivor function continuously decreases with time. Furthermore, after 15 years in emigration, over 70% of men have recorded a first endogamous marriage, compared to 50% of women. Interestingly, the Kaplan-Meier survival estimator shows a high number of women that have formed an endogamous marriage immediately upon arrival (less than one year). The percentage of "marriage migrants" (Niedomysl et al. 2010) among women is 46.8% (of the total number of marriages), while it is only 12.5% among migrant men.
Focusing on the areas of birth 6 , the lowest probability in the transition to endogamous marriage in Italy is among migrants from advanced economies, followed by Latin Americans and migrants from Eastern European countries of the EU. The other areas show higher probabilities and more similar patterns. In the transition to exogamous marriage in Italy, the highest percentages of "marriage migrants" are among migrants from advanced economies (40.0%) and non-EU Eastern Europe (34.2%); in contrast, Asia and sub-Saharan Africa have the lowest percentages of "marriage migrants" (respectively, 19.0% and 18.1%).
First exogamous WIT marriage is observed (as in Figure 1) more among women than men: the Kaplan-Meier survival function of women shows a continuous decrease with time, and in particular, within the first years after arrival in Italy (32.9%, while men represent only 12.9%).
The analyses by area of birth show reversed patterns in respect to endogamous marriage: the transition is observed mostly among migrants from advanced economies, Latin America and Eastern Europe. Latin Americans also have the largest percentages of endogamous marriage migration (44.2%), while migrants from Eastern European countries of the EU have the lowest (22.6%).

Parametric results
The results of the piecewise constant exponential model (Table 3) shed light on some understudied aspects of the transition to marriage (endogamous marriages vs exogamous WIT marriages). The baseline hazard ratios are not shown here as they have very similar patterns to those reported in Figure 2.
Overall, we observe, both for men and women, a reverse U-shaped effect of age on arrival for both endogamous marriage and exogamous WIT marriage. Moreover, the peaks are at ages 23-27, with the only two exceptions being women in the transition to endogamous marriage (peaking at ages 18-22) and men in the transition to exogamous marriage (peaking at ages 28-32). Unsurprisingly, and according to descriptive analyses, relevant differences emerge by gender and country of birth. Migrant women from Albania, the Balkans, the Indian Subcontinent and North Africa have the highest propensity for endogamous marriage, while those from Latin America and advanced economies have the highest ratios of marrying an Italian partner compared to Romanians.
Among men, we have similar patterns despite some exceptions. Migrants from Poland, the Philippines, Africa, Latin America and advanced economies have the lowest probability ratio of endogamous marriage. Migrant men coming from the same countries, such as Poland, Latin America and advanced economies, have the highest probability ratios of exogamous WIT marriage. In addition, Albanian and North African (in particular, Moroccan) men have a higher probability of an exogamous WIT marriage than those from the reference country.
The favourable economic conditions in the area of settlement, approximated by the macro-area of residence, positively affect the transition to endogamous marriage, especially among men: the lowest hazard ratios are in the south and the islands, and in central Italy. Conversely, we do not have a clear geographical pattern when looking at the macroarea effect on the transition to exogamous WIT marriage.
Education level has a different impact by gender. We found a greater probability of forming an endogamous marriage among educated men (medium/high level of education), while among women, the effect of education is not statistically significant for the same typology of marriage. The opposite is true for exogamous WIT marriage: highly educated women have a greater probability of forming a union, while education has no significant effect on exogamous marriage of men.
The results concerning the first job after migration and (possibly) before marriage indicate a double effect: unemployment increases the probability of marriage regardless of gender and the origin background of the partner, while a high-skilled occupation seems to be a positive determinant of exogamous WIT marriage, while it is negatively associated to endogamous marriage.
Information about the reasons for migration is equally interesting. Among men, labour-based migration increases the probability of forming an endogamous couple, while among women, it assumes a significant and negative effect in the transition to exogamous WIT marriage. Among women, affective motives foster the transition to marriage regardless of the origin background of the partner. Surprisingly, among men, these motives considerably reduce the probability of marrying a co-national. In addition, migration for other reasons (study etc.) reduces the probability of endogamous marriage, but increases the hazard ratio of exogamous WIT marriage among men. Note: (a) the baselines are here not shown as they result very similar to the ones reported in Figure 2; (b) too few cases. Legend: * p<0.05; **p< 0.01; ***p<0.001 Source: Authors' elaboration using SCIF survey (2011)(2012) Parents' disagreement with the decision to migrate significantly and positively affects exogamous WIT marriage. Temporal or indefinite intentions about the migratory project in Italy reduce the transition to first marriage, but the initial migratory project has an impact among women: temporary migrants are less likely to marry an Italian. As far as the timing after arrival in Italy is concerned (Figure 2), two different patterns emerge by gender, confirming the descriptive analyses: regardless of the type of marriage (endogamous or exogamous), the hazard ratios for women are the highest on and soon after (around 6 months) arrival in Italy. They decrease consistently in the following years.
Among men, the same peak in the 6 months after arrival is observed only for endogamous marriage, but at a reduced level of hazard ratio. After that, the hazard ratio of endogamous marriage assumes a reverse U-shaped pattern and peaks at 6-8 years after arrival. Men have very low hazard ratios of exogamous WIT marriage, which slightly increase with their permanence in Italy. The differential patterns by gender are also evident when looking at reasons for migration ( Figure 3). Men and women present similar patterns to those shown above, regardless of the reason for migration, but at different levels: for women, the highest hazard ratios on arrival in Italy occurred if they migrated for family reasons; for men, the most clear reverse U-shaped pattern, which peaks at 6-8 years after arrival, occurs in the transition to endogamous marriage if they migrate for labour reasons. Interestingly, additional peculiarities emerge. For men who migrated for family reasons, the peak of the hazard ratio in the six months after arrival is not as high as that of women. At the same time, women who migrate for labour reasons do not have the same reverse U-shaped pattern as that of men.

Discussion and conclusion
The results provide evidence for the complexity of today's migrations and indicate the coexistence of various patterns among first-generation migrants in Italy characterised by a plurality of origins with different projects and behavioural models. More specifically, the results illustrate the way in which the Italian marriage market operates by outlining gendered differentials in the timing of forming exogamous and endogamous marriages.
The "interrelation of events" hypothesis seems to better explain the transitions to both endogamous and exogamous WIT marriages among women (H1), who have the highest probability of being "marriage migrants" (Niedomysl et al. 2010). Conversely, men usually spend more time achieving economic stability before finding a partner. However, the postponement and recovery patterns among them (typical of disruption hypotheses) are much more evident in the transition to endogamous marriages than exogamous ones, where we found constant and low transition rates.
This differentiation along gender lines is particularly evident when looking at reasons for migration (H2). Affective reasons for migration are of crucial importance for women who transit to an endogamous marriage. Conversely, men who migrate for labour reasons postpone the transition to endogamous marriages.
Interestingly, despite this general picture, our analysis shows different and original paths as well. Among men, the pattern of marriage migration does not significantly differ according to the reason for migration. At the same time, female labour migrants do not have the same time-related pattern in the transition to marriage that is typical for male labour migrants. Thus, in our analyses, the gender role seems to be more important than the reason for migration in defining the timing of endogamous marriage.
A further and different picture emerges for exogamous WIT marriages. The general patterns differ more according to the reason for migration than by gender: affective reasons stimulate the transition to exogamous WIT marriage regardless of gender, while labour reasons postpone it. This result could be related to two factors: exogamous WIT marriage is frequently a strategy for gaining integration in the host country (by achieving naturalisation); in addition, this could indicate that an individual could already be in contact with his/her future partner before migration. Unfortunately, the data do not allow us to define the legal status of migrants on arrival in Italy, or the process of regularisation.
In addition, our results confirm the important role of the context of origin, by gender, as an expression of the system of values and norms, and of the culture, in shaping the transition to marriage (Kalmijin 1998; Kalmijin/van Tubergen 2010). Women from countries with more egalitarian gender roles have a higher propensity for exogamous marriage, while women from countries with more traditional gender roles have a higher propensity for endogamous marriage (H3). Conversely, and confirming previous results (Schroedter/ Rössel 2013), among men, a structural perspective prevails: men of small/ medium-sized communities in Italy with higher gender imbalances show lower transition rates to endogamous marriages and higher transition rates to exogamous marriages (H4).
The results support our hypotheses about the impact of human capital on marriage transition for women, but only for the transition to exogamous WIT marriage among men. In addition, having a high level of employment seems to be more powerful as a "bargaining chip" among both women and men. Unemployment fosters the transition to marriage not only for exogamous marriage, as expected, but also for endogamous marriage.
It should be noted that this study has three main limitations that should be taken into account. First, the data do not allow us to consider affective relationships and informal partnerships from a longitudinal perspective. Thus, we cannot know if the respondent had already met his/her future spouse before migration. This could reduce the time before the transition to first marriage after the migration event. However, the dummy "affective reason" for migration and the variable "migration project on arrival in Italy" should control for this bias. Moreover, Italian law allows legal reunification only for those in formal or marital union, and thus the risk of wrongly assuming the partner's migration as "independent" should be low.
Second, the use of event-history methods is limited, since most variables included in the models are observed as time-invariant. Moreover, little information is related to the premigration period or concerns the migration event itself. For example, we cannot consider an irregular status at migration that will undoubtedly postpone endogamous marriage as well as enhancing the probability of an exogamous WIT marriage in the destination country (to reduce the time required to achieve legal status). Unquestionably, longitudinal data would be the most appropriate source, since it would allow us to follow migrants throughout their migration experience, using time-varying variables and determining the casual nexus between migration and marriage (Elder 1985;Terzera 2011;Mussino et al. 2015). Unfortunately, due to the lack of longitudinal data for Italy, such an approach is not feasible. Despite this limitation, the cross-sectional SCIF data provide retrospective information about the timing of specific events (such as marriage or migration), which represent an original contribution to the general knowledge by applying time-related analyses.
Third, we have retrospective information due to the cross-sectional nature of the data. Migration and marriage are considered major life events, and thus we do not expect any memory problems to have affected the data. However, we should take into account the possibility of bias due to re-emigration (Ortensi 2015; Ortensi/Barbiano di Belgiojoso 2018). We could have missed all the migrants who have already left Italy (such as return migrations or onward migrations). Few data are available for re-emigration in Italy; however, the empirical evidence (Barbiano di Belgiojoso /Ortensi 2013) shows that at the time of the survey, less than 10% of migrants declared their intention to leave Italy. In addition, the selections generated by returnees and onward migration are sometimes complementary (ibid.): onward migration seems to be more related to both a search for better opportunities on the part of educated migrants, favoured by legal conditions and a coping strategy against unemployment, while flows toward the countries of origin are less linked to unemployment, and thus the migration experience seems to be related to withdrawal on the part of those individuals with more limited capabilities.
Notwithstanding these limitations, the obtained results meaningfully and originally contribute to the international debate related to the timing of union formation of single migrants in the destination country, which also depends on the settlement process, such as integration into the labour market and the socio-economic context, and/or on discrimination faced in Italian society.
The current Italian legislation (introduced in 2002 with the "Bossi-Fini Law") requires that third-country citizens applying for or renewing their residence permit must have regular employment. Thus, it reduces the chance of permanent regular settlement in Italy in cases of unemployment or irregular work, and consequently, it negatively affects couple formation. In particular, according to our results, male migrants are the most disadvantaged in the transition to union, as they remain single in many cases or need long periods of settlement in the host country before forming a couple.
The recent economic crisis negatively affected employment conditions and increased the unemployment rate more among immigrants (in particular, men) than among natives (Bonifazi/Marini 2014). At an individual level, the financial uncertainty deeply influenced demographic behaviour, delaying couple formation and childbearing in early adulthood. In the last few years, marriages with at least one foreign partner (in particular, the ones with both partners being foreign) significantly decreased in Italy (Gabrielli et al. 2017).
These matters require reducing or removing the link between having a regular job and obtaining a residence permit in order to favour migrants' settlement/integration and couple-formation processes. Moreover, it is important to provide more regularisation channels and legal entry routes in order to avoid ex-ante (and not only ex-post through periodical regularisations) negative consequences of irregular status for migrants, in particular with respect to handling family-related issues.
In the end, policies supporting the reconciliation of work, family, and private life (such as employment protection and non-discrimination measures, cash benefits, health protection, maternal leave at work, and childcare provisions) may encourage couple formation among migrants.