Influence of Culture and Economy on Immigrant Couples' Female Labor Force Participation, Summaries of Decision Making

Download Influence of Culture and Economy on Immigrant Couples' Female Labor Force Participation and more Summaries Decision Making in PDF only on Docsity! Essays on International Family Migration Inaugural-Dissertation zur Erlangung des Grades Doctor oeconomiae publicae (Dr. oec. publ.) an der Ludwig–Maximilians–Universität München 2016 vorgelegt von Till Nikolka Referent: Prof. Panu Poutvaara, Ph.D. Korreferent: Prof. Dr. Joachim Winter Promotionsabschlussberatung: 16.11.2016 Datum der mündlichen Prüfung: 31. Oktober 2016 Namen der Berichterstatter: Prof. Panu Poutvaara, Ph.D., Prof. Dr. Joachim Winter, Prof. Helmut Rainer, Ph.D. Contents Introduction 1 1 International Family Migration and the Dual-Earner Model 7 1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 1.2 Theoretical Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 1.3 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 1.4 Big Picture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 1.4.1 Main Motivation to Emigrate . . . . . . . . . . . . . . . . . . . . . . . . . . 25 1.4.2 Actual and Preferred Labor Force Participation . . . . . . . . . . . . . . . . 28 1.4.3 Household Economies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 1.5 Econometric Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 1.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 2 Family Decision-Making on International Migration 49 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 2.2 The Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 2.2.1 Utility of a Single . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 2.2.2 Utilities in a Couple . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 2.2.3 Migration Probabilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 2.2.4 Comparative Statics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 2.3 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 2.4 Descriptive Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 vi Table of Contents 2.5 Econometric Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 2.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 3 Family Return Migration 77 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 3.2 Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 3.3 Descriptive Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 3.4 Econometric Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 3.4.1 Children and Return Migration . . . . . . . . . . . . . . . . . . . . . . . . . 95 3.4.2 Earnings, Family Ties and Return Migration . . . . . . . . . . . . . . . . . . 105 3.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 A International Family Migration and the Dual-Earner Model 113 B Family Decision-Making on International Migration 129 C Family Return Migration 137 Bibliography 141 List of Figures 1.1 Required child care and min. relative wage for labor force participation of partner b. 16 2.1 Conditions for joint emigration of a couple (illustration for wa = wb and h > c). . . 61 2.2 Conditions for joint emigration of a couple (illustration for wa > wb and h < c). . . 62 3.1 Return migration propensities in percent according to age of oldest child. . . . . . . 89 3.2 Cumulative distribution functions for log-standardized earnings. . . . . . . . . . . . 92 x List of Tables 2.3 Income and migration preferences. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 2.4 Migration preferences by destination country. . . . . . . . . . . . . . . . . . . . . . 71 2.5 Linear probability regressions: Stronger male preferences towards joint emigration. . 73 2.6 Linear probability regressions: Stronger male preferences towards joint emigration, dual-earner couples. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 3.1 Immigrant population in Denmark, 2005. . . . . . . . . . . . . . . . . . . . . . . . . 84 3.2 Origin countries. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 3.3 Descriptive statistics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 3.4 Linear probability regressions: Family return to origin country. . . . . . . . . . . . . 97 3.5 Linear probability regressions: Children and family return to origin country. . . . . 101 3.6 Linear probability regressions: Schooling considerations and family return to origin. 103 3.7 Linear probability regressions: Year dummy coefficient estimates for return migration.104 3.8 Linear probability regressions: Earnings and return migration, all countries, male singles and primary earners. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 3.9 Linear probability regressions: Earnings and return migration, all countries, female singles and primary earners. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 3.10 Linear probability regressions: Earnings, children and return migration. . . . . . . . 108 Introduction Most of the research studying international migration takes an individual level approach abstract- ing from family ties and the household perspective. However, according to data from a representa- tive survey among long-term emigrants from Denmark (Poutvaara, Munk, and Junge, 2009) 33% of respondents stated to have migrated mainly for family reasons. Among all respondents this is the second largest fraction after own work as a migration motive with 37%. In particular, among female migrants the share of individuals stating that the family was the main reason to migrate is large, with 47%. The importance of family ties for migration decisions has been pointed out early in the literature (Mincer, 1978; Stark, 1985). There is little evidence, though, on the role of family considerations in the context of international migration. Most of the existing empirical work analyzing migration of families is restricted to within country migration (e.g. Costa and Kahn, 2000; Nivalainen, 2004; Tenn, 2010; Rabe, 2011). One reason for this is a lack of data on family ties of international migrants. On top of this, the understanding of mechanisms which determine household migration decisions in the international context is still incomplete. To this end, micro level analysis that takes a more comprehensive household perspective not only helps to better understand the complex nature of individual mi- gration decisions. It eventually provides insights related to migration flows on the macro level and can establish a basis for policy design also taking into account this perspective. This dissertation contributes both theoretically as well as empirically to this goal. The presented work provides new insights on the role of family considerations for emigration and return decisions of migrants. It studies how partners’ motivations, preferences and eventual decisions to migrate are related to individual and family characteristics as well as labor market outcomes in the context of interna- 2 Introduction tional migration. Previous research has already shown that family ties are an important impediment to migration within national borders. Mincer (1978) finds low migration rates of individuals with family ties, in particular if both partners are employed. Families are most likely to jointly relocate if the wife has weak labor market attachment or low earnings. Thus, family migration often reflects incentives and gains of one partner while the other one is a tied mover. In dual career households the co-location problem can be expected to be most severe and joint migration might go along with one partner sacrificing his or her job opportunities, and becoming a tied mover. In the light of increasing female labor force participation in many countries there is an extensive literature looking at the relationship between female labor market outcomes and family migration decisions. Costa and Kahn (2000) show that couples in which both partners hold a college degree tend to reside in large metropolitan areas to reduce their co-location problems. On the other hand, Tenn (2010) finds that female education and earnings are still a weak determinant for overall migration flows of couples and families in the US. He suspects that it is too difficult to balance two careers for a couple and concludes that women are in the weaker position when it comes to joint labor market decisions. Gemici (2011) also argues that family migration inside the US is primarily a response to male labor market options. Gemici finds that the correlation of married partners’ earnings gains between locations is low. Despite increasing female labor force participation these more recent empirical evidence confirm older studies for migration within the US finding significant over-qualification of women after family relocations (Frank, 1978) and low labor force participation rates of married women after migration (Lichter, 1980). Confirming empirical evidence on family migration within the US, Smits et al. (2003) find that for family relocations in the Netherlands between 1977 and 1996 females are persistently tied movers, too. Potentially conflicting preferences on migration can be expected to be even more important for international migration: Partners do not necessarily share the same language skills, and different types of education vary in the extent to which they are internationally transferable. However, little Introduction 5 in Denmark is studied which tightened family reunification rules for foreigners in 2002. Results indicate that return propensities for couples from non-Western countries increase after the reform. The increase being statistically significant only for couples with children indicates that marriage considerations for the children might play a role in this context. Furthermore, the chapter studies selection into return migration on primary earners’ income among couples with different family characteristics. Primary earner incomes of those returning to the home country are higher than among those who stay in Denmark, in particular for immigrant couples from non-Western coun- tries. This selection pattern is weaker among dual-earner couples and families with children. 6 Introduction Chapter 1 International Family Migration and the Dual-Earner Model∗ 1.1 Introduction A couple considering migration might face a difficult trade-off in whose career to prioritize. An important question from a societal perspective is hence whether post-migration changes in labor force participation and and intra-household resource allocation reflect both partners’ preferences, or whether migration hurts one of the partners. Previous research on internal migration has found that couple migration is typically associated with career gains for men, with women often leaving the labor market, or at least reducing hours worked (Mincer, 1978; Frank, 1978; Costa and Kahn, 2000; Compton and Pollak, 2007). As partners may differ in the international transferability of their education and in their language skills, international migration is likely to impose even more difficult trade-offs. This is of increasing importance in globalized economies, affecting both inter- national allocation of talent and firm competitiveness. However, so far there is no research linking pre-migration and post-migration outcomes in the context of international family migration. This chapter presents first a theoretical model on how couples with and without children decide on their labor supply. Depending on wage rates, gender identity and, in the presence of children, the ∗This chapter is based on joint work with Martin D. Munk and Panu Poutvaara. 10 1. International Family Migration and the Dual-Earner Model hypothesis that female emigrants’ labor force participation rates would converge to female labor force participation rates in the destination country. While most of migration research has focused on migration from poor to rich countries, migration flows between rich countries are also substantial. In 2013, 22 million persons born in one of the EU15 countries2 live outside their country of origin. Of them, 42% live in another EU15 country and an additional 13% in the United States (United Nations, 2013). The pattern of emigration from Denmark is rather similar. In 2013, over a quarter million Danes lived outside Denmark (corresponding to about 5% of the Danish-born population), with 50% of the migrants living in other EU15 countries and 13% in the United States (United Nations, Department of Economic and Social Affairs 2013). Emigration rates from Denmark are neither exceptionally low nor excep- tionally high compared to other European countries. In 2012, the emigration rate was in the age group 25 to 54 among the native-born 0.33% in Denmark, 0.18% in Germany, 1.19% in Ireland, 0.37% in the Netherlands, 0.13% in Spain, and 0.29% in Sweden (Eurostat 2016). We make three main contributions to the previous literature on couple migration. First, we obtain direct evidence on partners’ stated motivations to migrate and on whether migration was a joint preference, or if one partner was a tied mover who would have preferred not to migrate. Second, we have collected data on labor force participation and household economies before and after mi- gration. Third, we analyze international migration. Due to data restrictions, almost all literature on family migration has focused on internal migration. The few papers that have analyzed in- ternational migration in the family context have looked at labor force participation or earnings only in the destination (Borjas and Bronars 1991; Cobb-Clark 1993), or only in the country of origin (Junge et al. 2014).3 Junge et al. (2014) show that the likelihood that a dual-earner couple emigrates increases strongly in the earnings of the primary earner. Our main focus is not on the self-selection into emigration, but on partners’ motivations to emigrate and preferred and actual 2This country group includes: Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Portugal, Spain, Sweden, and the United Kingdom. 3There is an extensive literature on the effects of temporary migration of family members on those left behind in the home country for example through remittances. For a survey see Docquier and Rapoport (2006) 1.1 Introduction 11 labor force participation patterns abroad. We analyze couples that migrated together between 1987 and 2002, stayed together after migration and had not returned to Denmark by the time of the survey in 2008. We find that most male respondents migrated for own work reasons and most females for family reasons. Changes in labor force participation after emigration are in line with stated main motivations. Female labor force participation declined from 86.2% in Denmark to 68.0% abroad in 2008. Male labor force partici- pation, instead, increased from 94.8% to 97.5%. Migration was almost always a shared preference: 6.4% of women and only 0.6% of men are reluctant tied movers in the sense that they would have preferred to stay in Denmark, but migrated due to their partner’s superior job opportunities. There are four potential explanations for lower female labor force participation rates outside Nordic countries: wider wage differences, relatively higher price of child care services than in Denmark, assimilation of emigrants towards prevailing social norms in the destination country, and self- selection of couples in terms of their preference towards male breadwinner model. Empirical analysis confirms that female labor force participation is lower among couples living in countries where incomes are more dispersed, as measured by the GINI coefficient, as well as in countries with less affordable child care. Prevailing female labor force participation rates in the countries of residence, instead, fail to explain the observed differences in female labor force participation among the migrant couples. Our survey also provides tentative evidence supporting self-selection in terms of labor force participation preferences: a sizable minority of women viewed less stressful working life as an argument in favor of emigrating, and a much smaller group against, while the opposite holds among men migrating outside Nordic countries. Furthermore, considerations in favor of children were viewed as an argument in favor of emigration clearly more often by those migrating outside Nordic countries than by those migrating to other Nordic countries. Our findings raise intriguing questions about family preferences towards the dual-earner and the male breadwinner model, also outside the context of migration. One third of couples that have 12 1. International Family Migration and the Dual-Earner Model emigrated outside the Nordic countries prefer that only the male works, in a stark contrast to the Nordic countries in which only one in sixteen couples prefers that the male works and the female stays at home. Nonetheless, there are no pre-migration differences in female labor participation between couples that subsequently migrated to other Nordic countries and those migrating else- where. Furthermore, especially women migrating to the United States and non-Western countries viewed escaping stressful working life as an argument in favor of emigration, and very few against. This suggests that a large fraction of couples emigrating outside the Nordic countries preferred to switch from a dual-earner model to a male breadwinner model. Those preferring a dual-earner model might choose to stay in Denmark or to go to another Nordic country in which generous welfare services, especially child care provision, make combining work and family easy, in order to avoid the female being pushed out of the labor force. Tiebout (1956) already suggests that different policy preferences play a role in the decision where to live. As it is unlikely that all couples preferring a male breadwinner model would go as far as to emigrate, an open question is how many of the dual-earner couples living in egalitarian Nordic countries would actually prefer the traditional male breadwinner model, if wider income differences would allow a high living stan- dard with only one partner working. The big differences in labor force participation and household economies between emigrant couples in different destinations are likely to reflect a joint effect of self-selection and labor supply adjustments as a reaction to different relative wage rates and price and availability of child care services. However, addressing self-selection based on revealed pre- migration preferences we do not find big differences in female labor supply in Denmark between couples subsequently migrating to different destinations. Some studies, e.g. Antecol (2000), Fernández and Fogli (2009) and Blau et al. (2011), conclude that cultural background plays a major role for migrant families’ labor supply decisions in the des- tination country. In particular, female labor force participation tends to be higher among families coming from countries with historically stronger female labor force attachment. Our analysis in- dicates that female labor force participation rates among emigrant couples from the same cultural background differ substantially according to the country of residence. Our findings underline the 1.2 Theoretical Framework 15 We separately analyze household utility maximization under two mutually exclusive conditions. First, if wb < pD, the full amount of required child care D̄ is produced privately. Second, if wb ≥ pD all child care D = D̄ is bought on the market. For couples without children child care does not enter in the labor supply decision as D̄ = D = 0. In both cases we proceed similarly to obtain the utility maximizing time allocation of a and b within the household. Solving the model for optimal time allocation of partners a and b always yields at least one corner solution for either H∗ a or H∗ b . It depends on the parameter values whether the model yields an interior solution for one partner or whether both partners fully specialize in different activities. In case there is an interior solution one partner divides time between working in the labor market and in the household while the other partner fully specializes in one of the two activities. If there is no interior solution to the utility maximization problem, both partners fully specialize working either in the labor market or in household production. Below we present the parameter values that have to hold for each of these solutions (0 < H∗ a ≤ 1, H∗ b = 1), (H∗ a = 0, H∗ b = 1) or (H∗ a = 0, 0 ≤ H∗ b < 1). The appendix provides a consistency check that the presented time allocations and labor supply decisions yield indeed optimal household utility levels. In case wb < pD  H∗1 a = 1− 2α + αD̄,H∗1 b = 1 if α < 1 2−D̄ H∗1 a = 0, H∗1 b = 1 if 1 2−D̄ ≤ α ≤ wa wa+wb−wbD̄ H∗1 a = 0, H∗1 b = (1− α) ( wa wb + 1 ) + αD̄ if wa wa+wb−wbD̄ < α. In case wb ≥ pD  H∗2 a = 1− 2α− (1− α)pD wa D̄,H∗2 b = 1 if α < wa−pDD̄ 2wa−pDD̄ H∗2 a = 0, H∗2 b = 1 if wa−pDD̄ 2wa−pDD̄ ≤ α ≤ wa−pDD̄ wa+wb−pDD̄ H∗2 a = 0, H∗2 b = (1− α) ( wa wb + 1− pD wb D̄ ) if wa−pDD̄ wa+wb−pDD̄ < α. 16 1. International Family Migration and the Dual-Earner Model As can be seen from H∗1 b and H∗2 b , individual labor supply of partner b increases along the intensive margin with wb and decreases with wa. Additionally, the labor market participation threshold for b in terms of the couple’s utility weight for consumption α decreases with wb and increases with wa. For couples with children, the threshold for which b’s labor supply equals zero also depends on the amount of required child care D̄. Among parents providing D = D̄ of own time for child care (wb < pD), the participation threshold value increases with D̄. Less time is available for household production generating relatively high marginal utility and the minimum relative wage for b to participate in the labor market increases. In families buying D = D̄ on the market (wb ≥ pD), the threshold for labor market participation of b decreases with D̄. The relative wage of b needed for labor market participation is lower if required child care increases because financing more expenses for child care makes additional labor supply necessary. These results also hold for the case that wa < wb while λ >> 0 (Appendix). Figure 1.1 illustrates the relationship between the minimum relative wage for labor market participation of b and the required time D̄ for different values of α. The values on the vertical axis (D̄ = 0) reveal the required relative wage for b to participate in the labor force if the couple has no children. Figure 1.1: Required child care and min. relative wage for labor force participation of partner b. The theory presents a mechanism relating each partner’s earnings potential and required child care to labor supply decisions in the household. The decision on which of the partners specializes in the labor market could either be due to comparative advantage (λ = 0) or due to preferences for 1.2 Theoretical Framework 17 partner b to specialize in household production (λ >> 0). In the data we observe that almost all male partners work full-time before as well as after migration. Female labor supply and labor market participation vary to a larger extent. Given our data the subsequent analysis is going to study the implications of the model for the case that households adjust female labor supply and the male partner works full time on the labor market. Hence, we refer to partner a as the male and partner b as the female partner. This chapter focuses on household economies and labor supply decisions in the context of inter- national migration. Migration goes along with changes in employment and the partners’ income situation. In the light of our theoretical model, we relate potential explanatory factors like the partners’ education - as an indicator for their earnings potential - as well as family characteristics to household economies and labor supply abroad. We can summarize the hypotheses we test as follows: Hypothesis 1 (tied mover effect). Female labor force participation is lower abroad if the female partner is a tied mover. Hypothesis 1 follows as if a couple migrates mainly for job opportunities of one partner, we expect by the revealed preference argumentation that the wage rate of that partner increases. This, in turn, makes it likely that the other partner works less, even if the wage rate of the other partner would be the same abroad. If the wage rate of the accompanying spouse is lower abroad, this further reduces his or her expected labor force participation. Hypothesis 2 (the effects of children). Having children, especially young children, reduces female labor force participation in countries with expensive child care services. Hypothesis 2 follows from our model if the female partner is more often the one taking care of chil- 20 1. International Family Migration and the Dual-Earner Model is less easily available and more costly than in Denmark.6 Our prior is that female labor force participation decreases abroad if the couple has children, in particular for couples children below school age, and that this relationship is less strong for the Nordic countries. Although we presented our model in a unitary framework, it would be straightforward to extend it to a collective framework. Instead of maximizing household utility function, we would write the two partners’ utility functions separately. When the household resource allocation is Pareto effi- cient, the bargaining outcome is as if the household would maximize the sum of the two partners’ utility functions, weighted to reflect their relative bargaining power (Chiappori 1988; Apps and Rees 1988). Cherchye et al. (2012) extended the model to account for more than one domestic good (like parental time and household activities like cleaning), and brought the model to data. We use the simpler unitary model as our data does not allow testing the richer collective model. As Cherchye et al. (2012), we separate between time spent on child care and other household production, but our setting is simpler as we take the time required for child care as exogenous and do not model leisure choice. Instead, we allow child care to be produced either inside the household, or bought from the market, and show that the effects of time required for child care on female labor supply depend crucially on whether the child care cost is more or less than the female wage rate. As a result, female labor force participation is expected to be much lower among couples requiring child care in countries with more expensive child care, but not in the Nordic countries where public day care centers are available and affordable for most parents. 1.3 Data We study household migration using own survey data on Danes who had emigrated in 1987, 1988, 1992, 1993, 1997, 1998, 2001 or 2002, and had not returned to Denmark by 2007. The survey was 6Previous literature has shown that the quality, availability and affordability of child care services has a consid- erable impact on couples’ labor supply decisions. For example, Berger and Black (1992) show that receiving child care subsidies increased employment of mothers in the US. In general, public child care is more easily available and more heavily subsidized in the Nordic countries than in most other countries. This makes it easier and more attractive for both partners to participate in the labor market when having children. 1.3 Data 21 planned by Munk and Poutvaara within the project "Danes Abroad: Economic and Social Mo- tivations for Emigration and Return Migration", financed by the Danish Social Science Research Council. The survey was carried out by Statistics Denmark. Statistics Denmark used the full population register to identify all Danish citizens who had em- igrated, and had not returned by the end of 2007. Emigrants had to be aged 18 or more when they emigrated, and at most 59 in 2007. These restrictions yield a population of 17,309 emi- grants during the selected years. For 55% of those emigrants Statistics Denmark found contact information on relatives living in Denmark. They were asked to provide address, telephone num- ber and email address of the person abroad. Statistics Denmark then tried to contact the target persons using this information. The link to the web-based questionnaire could finally be sent out by email to 6,984 emigrants with validated email addresses. When data collection was closed in August 2008 4,257 persons had answered the survey. This makes an overall response rate of 61%. We can link respondents with the population register data using a unique personal identifier, the social security number. Like other Nordic countries, Denmark collects comprehensive data on residence, wages, tax payments, education and household composition for every individual in the country. Having a social security number is compulsory and necessary for everyday life activities, like visiting a doctor, enrolling at school or opening a bank account. Our main data sources are the individual administrative register and the migration register containing information on age, gender, date of emigration and destination and eventual return migration, educational attainment and household composition in Denmark. Our survey data contains information on several characteristics like education, household compo- sition and work situation for the respondent and his or her partner before migration in Denmark as well as in 2008 abroad. Survey respondents were also asked about their main motivation to em- igrate and their preferences for their own and for their partner’s labor force participation. On top of this, they were asked to provide information about their household economies before migration 22 1. International Family Migration and the Dual-Earner Model in Denmark and in 2008 abroad. This analysis focuses on long-term emigration of couples and we consider only partners who stayed together abroad at least until 2008.7 We restrict the sample to individuals who emigrated with their partner and whose partner is a Danish citizen, born in Denmark. The reason for this restriction is that in international couples, emigration from Denmark might imply returning to the home country of one partner, making a migration decision qualitatively different. Our analysis revealed that results on the subsample of respondents with a foreign partner are qualitatively similar but statistically weaker and thus will not be reported here. We furthermore require that the partners lived together in Denmark. In our survey 632 respondents fulfill the above restrictions. Using migration events from the administrative registers we are able to construct the population of individuals who emigrated with their partner during the same years as those considered for the survey. Using the register data restricts attention to joint emigrations where partners migrated to the same destination country in the same year. This is more restrictive than our survey sample as it does not include sequential migrations. However, we cannot distinguish between sequential emigration and split-ups using only the administrative data. The register data yields an overall sample of 1250 couples in which partners were at least 18 when emigrating and at most 59 in 2007, who emigrated jointly and did not return until 2008. As in the survey, both partners are Danes, born in Denmark. The descriptive statistics for migrations according to register data and our survey are reported in columns 1 and 2 of Table 1. For the sample of survey respondents which are going to be analyzed in the remainder of this chapter attention is restricted to partners who reported to have emigrated together and who still live together abroad in 2008, which yields a sample of 522 respondents referred to in column 3. Table 1 shows the distribution of initial destinations and emigration years for the three samples. Furthermore, it reports female labor force participation rates in Denmark before migration. Data indicates that a slightly lower share of survey respondents emigrated in 1987/88 compared with the register data emigrations. Overall, the shares of migrant couples having emigrated in 7Both married and cohabiting couples are included. Cohabitation before or instead of marriage is a common phenomenon in the Scandinavian countries (Kiernan, 2004). 1.4 Big Picture 25 tertiary education was 24% in 2008 (Statistics Denmark 2009). Thus compared with the overall Danish population the average level of education among our respondents is high. This is in line with results from Borjas, Kauppinen and Poutvaara (2015) who find that emigrants from Denmark are strongly positively self-selected according to their pre-migration education and earnings. Table 2.1 reports the country of residence of the respondents in 2008. A large fraction of the couples in our sample (20.9%) live in one of the other Nordic countries, Sweden, Norway, Finland or Iceland. 16.1% of the couples live in the US, and 13.0% in the UK. In total 65.7% percent of the couples live in another Western European country.9 In 2008, 78.2% of couples have children in the household, with 33.0% of the couples having children below school age. Female labor force participation has decreased considerably in 2008 compared to the situation in Denmark: only 68.0% of females par- ticipate in the labor force. Male labor force participation, on the other hand, has even increased to 97.5%. Characteristics presented in Table 2.1 do not reveal any response bias for whether the survey respondent was male or female, with the exception that there are slightly more female than male respondents in other Nordic destination countries. This is the only difference among the reported characteristics which is statistically significant on the 5% level. 1.4 Big Picture 1.4.1 Main Motivation to Emigrate Our survey provides information about the respondents’ main motivation to emigrate. Table 1.3 presents information separately for men and women who emigrated with a Danish partner and live together abroad in 2008.10 For all residence country groups, own work was the most important motivation to emigrate for a clear majority of men. The majority of women in all non-Nordic 9The differences in the shares of country groups between Table 2.1 and 1.1 are due to the fact that some couples migrated from the initial destination country to another country. 10Due to small numbers of observations when reporting results separately for female and male respondents we group together other Western destination countries apart from the US and the Nordic countries in one category. We keep these country groups throughout the remainder of the analysis and show in the appendix that our findings we describe later do not differ much between the destination groups Australia, Canada and New Zealand, the UK and other Western European countries. 26 1. International Family Migration and the Dual-Earner Model countries migrated due to family reasons. Work reasons dominate especially strongly among men and family reasons among women who live in the US and in non-Western countries. Among women who live in other Nordic countries, about one third emigrated for work, almost one third for family and a little more than one third for other reasons, like studies. Other Non- Nordic United Western Western countries States countries countries Total Male partners Own work 66.0 81.0 71.1 92.1 74.7 Family 10.6 0.0 7.8 0.0 6.0 Other reasons 23.4 19.0 21.1 7.9 19.3 Female partners Own work 33.9 11.9 20.7 14.2 21.7 Family 30.7 76.2 52.9 67.9 53.0 Other reasons 35.4 11.9 26.4 17.9 25.3 Source: Survey data Table 1.3: Main motivations to emigrate. Table 1.4 presents evaluations of further factors which had an impact on the decision to emigrate. Two out of five women living outside Nordic countries viewed a less stressful working life as an important or very important reason to emigrate, and one in ten as an argument for staying in Denmark. Among men migrating outside other Nordic countries, a less stressful working life was more often an argument for staying in Denmark. Both men and women living in other Nordic countries view a less stressful working life more often an argument for emigration. The distribution of preferences in Table 1.4 suggests, furthermore, that most parents viewed con- siderations related to their children to be neither in favor nor against emigration, but there is also a significant minority which saw considerations related to children as an important or very important reason to emigrate. This is most pronounced among parents migrating to the US. Baker et al. (2008), Havnes and Mogstad (2015) as well as Kottelenberg and Lehrer (2014) show that universal public child care might affect child outcomes negatively.11 Del Boca et al. (2014) find 11Kottelenberg and Lehrer (2014) show that the effects of universal childcare depend strongly on the child’s age. The negative effect is stronger when the children gain access at younger age, but turns positive for those that are 3 1.4 Big Picture 27 that parents’ time spent with their child is important for the child’s cognitive development. One possible explanation for our findings is that the respondents for whom considerations related to children are a reason to emigrate prefer that the children are taken care of at home, typically by the mother. We checked whether fertility decisions of the couples differ according to the country of residence but we did not find any systematic difference. Other Non- Nordic United Western Western countries States countries countries Male partners Less stressful working life in favor 25.5 14.3 21.8 15.8 against 4.3 28.6 23.2 31.6 Considerations for children in favor 6.4 26.2 16.2 23.7 (couples with children) against 2.1 2.4 3.5 0.0 Female partners Less stressful working life in favor 19.4 42.9 36.4 46.4 against 4.8 11.9 7.4 10.7 Considerations for children in favor 11.3 31.0 19.8 14.3 (couples with children) against 4.8 2.4 1.7 0.0 Source: Survey data Table 1.4: Selected arguments in favor or against emigration. With our data we are additionally able to address the question which partner had stronger pref- erences towards emigration. Respondents were asked whether both had equally strong migration preferences or whether one of the partners was more in favor of emigration. Table 1.5 shows, that in most cases the partners agreed on emigration and in a clear majority emigration was a joint preference of both partners.12 In every destination country group more than 50% of the respon- dents stated that they and their partner had equally strong preferences to emigrate. In 39.0% of couples, the male and in 9.1% the female partner was more strongly in favor of emigration. In years or older. Cornelissen et al. (2015) show for a universal child care program in Germany that attendance rates are higher for children from disadvantaged backgrounds. 12A potential concern in Table 1.5 might be a bias due to misreporting of the partners’ preferences depending on the characteristics of the partner who answered the questionnaire. Using register data, we were able to identify among our respondents 60 couples in which both the male and the female answered our survey and analyzed both partners’ mutual assessment of migration preferences in this subsample. Table A1 shows that in 49 of 60 cases the partners’ answers on their respective migration preferences were perfect matches. If there were deviations one partner mostly reported mutual agreement while the other stated stronger migration preferences of the male or female partner. Thus, we can expect joint migration preferences to be correctly assessed by one partner for most of the remaining observations in our sample. 30 1. International Family Migration and the Dual-Earner Model Male in the Female in the Both in the labor force, labor force, Both labor force female home male home at home No answer All 62.1 27.6 1.1 1.7 7.5 Nordic countries 80.7 6.3 1.8 2.8 8.2 Non-Nordic countries 57.1 33.2 1.0 1.5 7.3 United States 40.5 52.4 1.2 0.0 6.0 Other Western countries 62.7 28.1 0.8 2.3 6.1 Non-Western countries 56.1 28.8 1.5 0.0 13.6 Source: Survey data Table 1.7: Percentage shares for preferred labor force participation. labor force participation cannot be explained by long distance migration. Table 1.6 additionally shows that female labor force participation decreased particularly among couples with children in the household. Respondents were asked about their preferred division of labor in the household. In their answers they could indicate who of the partners they prefer to participate in the labor market.14 Table 1.7 reveals that low female labor force participation in 2008 seems to reflect the respondents’ prefer- ences towards their own and their partners’ participation in the labor market. Respondents who reside in the US are clearly more likely to state that they prefer the male partner to work while the female partner stays at home. In the Nordic countries, on the other hand, most respondents prefer a dual-earner household. In Table 1.8, we compare female labor force participation rates according to whether the female partner was a tied mover. We derive this information from the respondents’ main motivation to emigrate. In all country groups, except for the US, female labor force participation is lowest when the female is a tied mover, i.e. male respondents reported that they migrated for their own job opportunities and female respondents reported to have migrated mainly for family reasons. At the 14To alleviate the concern that respondents gave self-serving answers concerning their partner’s preferred labor force participation we analyze responses among the 60 couples in which both the male and the female answered our survey. The responses are in most cases mutually consistent among both partners as Table A3 shows. Given this subsample of couples we are confident that the responses do not depend on which of the partners answered the survey. In a clear majority of cases, the partners agreed that either both should work, or that the male should work and female should stay at home. 1.4 Big Picture 31 A: All couples. Nordic Other Western Non Western countries US countries countries Male tied mover 100.0% 40.0% 94.0% 75.0% (26) (5) (36) (4) No tied mover 97.0% 46.1% 75.8% 62.5% (33) (13) (62) (8) Female tied mover 90.0% 54.5% 65.4% 51.9% (50) (66) (165) (54) B: Couples without pre-school age children. Male tied mover 100.0% 100.0% 100.0% 100.0% (14) (2) (25) (1) No tied mover 100.0% 57.1% 86.9% 80.0% (21) (7) (46) (5) Female tied mover 90.9% 59.1% 66.4% 66.7% (33) (44) (116) (36) Source: Survey data Notes: Numbers of observation in parentheses. Table 1.8: Female labor force participation rates and tied mover status. same time, female labor force participation is highest among couples in which the male partner is a tied mover. Among couples without pre-school-age children, we find that in all considered country groups all women whose male partner is a tied mover participate in the labor force. These findings agree with Hypothesis 1 in our model. Tied mover’s partner should gain from migration, suggesting that wage differences abroad are larger than in Denmark. This makes it more likely that the tied mover (typically the female partner) works less, particularly if the couple migrates to a country where the dispersion of income is larger. This leads to lower female labor force par- ticipation, especially in the US and non-Western countries where incomes are more dispersed, and less so in other Nordic countries.15 Table A4 in the appendix shows that the pattern presented for the aggregated group of other Western countries hold separately for the UK, Other Western Europe and Australia, Canada and New Zealand. 15According to Klugman (2011) the income GINI coefficient or the income quintile ratio 2000-2011 is lowest in Nordic countries. For the US the income inequality measures are higher than for Western Europe and Australia, Canada and New Zealand. 32 1. International Family Migration and the Dual-Earner Model 1.4.3 Household Economies Table 1.9 provides further insights to the organization of household economies beyond the partners’ labor force participation. Respondents were asked whether they had separate economies or joint economies before migration as well as at the time of the survey in 2008.16 In separate economies the partners dispose of their own incomes and share only common expenses while couples with joint economies pool at least part of their individual incomes. Table 1.9 shows the couples’ types of household economies and provides information in case of joint household economies on which of the partners contributes more. The survey questions were referring to the situation before migra- tion (Panel A) and in 2008 (Panel B). We present the distribution of responses separately, again by country of residence in 2008. Before migration, most couples have joint household economies to which both partners contribute labor market income. Table 1.9 also shows that a considerable fraction of couples had separate household economies in Denmark. For 2008 we observe that the overall share of joint household economies is considerably higher and separate economies are less frequent among the couples in all country groups compared to the situation in Denmark. The share of household economies to which both partners contribute equally decreases. In general, we observe more couples in which the male partner is the main income earner. In 66.1% of cases the male partner contributed more in Denmark; this share increased to 76.6% abroad in 2008. There is substantial heterogeneity between household economies in different countries of residence. The share of females who stay at home is higher in 2008 for all except the Nordic countries. There are pronounced differences between the US and the Nordic countries. None of the couples in 16A potential concern related to our results might be that household economy dynamics are driven by a time effect which is unrelated to migration. Moreover, the composition of emigrant cohorts might have changed over time. Our findings could then be driven by a subgroup of households in our sample depending on the year of emigration. Table A5 addresses these issues and alleviates our concerns: We present separately the household economies for couples that emigrated before 1999 and in 2001/2002. We do not observe qualitative differences among the cohorts in terms of changes in household economies. For example, the decrease in the share of separate household economies among the 2000/2001 emigration cohorts is almost as high as for the couples that left Denmark in earlier years. Moreover, the composition of the cohorts in terms of countries of residence does not vary substantially over the years of emigration either as Table A6 shows. 1.5 Econometric Analysis 35 in favor of emigration for women suggests that other factors also play an important role in explain- ing low labor force participation rates of females in couples having migrated outside the Nordic countries. Moreover, the public provision of child care services provides incentives for labor market participation of both partners. Our model predicts that a lower price of child care would incen- tivize higher female labor force participation. Moreover, couples emigrating to a country where the provision of child care is low might be those preferring to spend more time with their children. In order to better understand the observed changes in labor force participation and household economies, we are going to jointly assess the role of various individual- and family characteristics in different countries of residence. 1.5 Econometric Analysis Female Labor Force Participation We use a linear probability regression framework to test the hypotheses from our theory with the data. For the subsequent analysis we exclude couples in which the female partner is retiree or student. In Table 2.5, we present factors associated with female labor force participation in the country of residence in 2008. As baseline explanatory variables, we include the couple’s power type (reference category: low power couples) and indicator variables for children in the household. We use the GINI coefficient to capture income differences in the country of residence. This coefficient refers to inequality in disposable incomes on the individual level taken from the WIDER WIID database (2016) for the year 2008. In an alternative specification, we include country group dum- mies (reference category: other Western countries) which additionally control for between country variation in language, culture and other factors which might be related to female labor force par- ticipation abroad. We also present separate regressions by country of residence groups. Estimates in Table 2.5 show that the presence of children is negatively associated with female labor force participation. This is especially the case for young children below the age of 7 which is in 36 1. International Family Migration and the Dual-Earner Model line with Hypothesis 2 from our model. The first column indicates, furthermore, that female labor force participation is lower in countries with wider income differences, which confirms Hypothesis 3. When controlling for country groups and comparing intercepts in the separate regressions we see that female labor force participation in the Nordic countries is more likely than in the US, other Western countries and non-Western countries.17 Our results do not reveal any significant effect of the presence of young children among couples having migrated to the Nordic countries while there are strong negative effects for the other countries. Female labor force participation is particularly low among couples with children aged 0 to 6 in the US and non-Western countries. In Table A9 we also present results including the "Starting Well Index", an index capturing the availability, affordability and quality of child care services in a subset of countries (The Economist Intelligence Unit, 2012). Results reveal that a higher value of this index is associated with higher female labor force participation which also is in line with our theory. The effect of the GINI coefficient is robust to including the "Starting Well Index". In Table A10 we additionally include an indicator for the work status of the female in Denmark before migration. If the female worked before migration, we estimate a significantly higher proba- bility for actual labor force participation abroad. Still the size and significance of most coefficient estimates remains unchanged. Moreover, we find that actual labor force participation reflects the preferred division of labor among the partners. In a similar framework to Table 2.5, we relate the explanatory variables to an indicator for whether the respondent prefers that either both partners work or only the female partner works. Results are reported in Table 1.11. In general, most coefficient estimates are similar in sign and size to the specifications in Table 2.5. The presence of children is negatively correlated with positive attitudes towards female labor force participation. Again, this relationship is weak and statistically insignificant for the Nordic countries. For the remaining countries, we observe 17We do not include the GINI coefficient in these regressions as most of the variation in the GINI is between the country groups rather than within them. Including the coefficient for the other Western destinations its estimate has a negative sign but is not statistically significant. Within the group of non-Western countries the estimated effect is negative and statistically significant at the 5% level. 1.5 Econometric Analysis 37 Nordic Other West. Non-Western All countries countries US countries countries Power couple -0.00310 0.0140 0.0155 -0.0984 0.0423 0.222 (0.0534) (0.0512) (0.0483) (0.190) (0.0788) (0.154) Female power couple 0.0531 0.0314 0.0390 0.0101 0.0111 0.277 (0.0699) (0.0657) (0.0442) (0.272) (0.114) (0.188) Male power couple -0.161** -0.146** -0.324* -0.0834 -0.135 -0.0565 (0.0727) (0.0685) (0.178) (0.195) (0.102) (0.175) Youngest child 0-6 -0.233*** -0.240*** -0.0696* -0.367** -0.239*** -0.454*** (0.0506) (0.0483) (0.0376) (0.173) (0.0755) (0.170) Youngest child 7+ -0.130*** -0.106** -0.0761 -0.200 -0.132** -0.00737 (0.0468) (0.0445) (0.0461) (0.160) (0.0639) (0.158) GINI -2.234*** (0.437) Nordic countries 0.214*** (0.0375) United States -0.178*** (0.0623) Non-Western countries -0.152** (0.0660) Constant 1.575*** 0.850*** 1.003*** 0.817*** 0.846*** 0.619*** (0.137) (0.0526) (0.0401) (0.183) (0.0718) (0.147) R-squared 0.136 0.160 0.179 0.071 0.056 0.205 Observations 465 505 105 81 255 64 Notes: OLS estimation. Robust standard errors in parentheses. *** Significant at the 1 percent level. ** Significant at the 5 percent level. * Significant at the 10 percent level. Table 1.10: Linear probability regressions: Female labor force participation abroad in 2008. 40 1. International Family Migration and the Dual-Earner Model Preference for Actual female labor force female labor force participation participation Power couple 0.0797 0.0948* -0.00367 0.0144 (0.0570) (0.0546) (0.0525) (0.0512) Female power couple 0.00378 -0.0157 0.0572 0.0884 (0.0834) (0.0787) (0.0698) (0.0660) Male power couple -0.147* -0.117 -0.160** -0.144** (0.0777) (0.0739) (0.0725) (0.0688) Youngest child 0-6 -0.241*** -0.245*** -0.241*** -0.243*** (0.0537) (0.0512) (0.0511) (0.0492) Youngest child 7+ -0.202*** -0.188*** -0.122*** -0.104** (0.0504) (0.0471) (0.0471) (0.0446) Mother out of LF at respondent’s age 12 -0.0880* -0.0796* -0.0635 -0.0224 (0.0491) (0.0474) (0.0455) (0.0445) GINI -2.187*** -2.271*** (0.474) (0.434) Nordic countries 0.184*** 0.213*** (0.0430) (0.0375) United States -0.228*** -0.179*** (0.0639) (0.0624) Non-Western countries 0.0109 -0.152** (0.0675) (0.0662) Constant 1.554*** 0.851*** 1.603*** 0.856*** (0.150) (0.0574) (0.135) (0.0527) R-squared 0.148 0.159 0.140 0.160 Observations 434 467 465 505 Notes: OLS estimation. Robust standard errors in parentheses. *** Significant at the 1 percent level. ** Significant at the 5 percent level. * Significant at the 10 percent level. Table 1.12: Linear probability regressions: Inter-generational transmission and female labor force participation. 1.5 Econometric Analysis 41 Overall, the results are in line with what our theory predicts. Differences in child care prices and in wage distribution can rationalize differences in female labor force participation between different countries of residence. In addition to the price mechanism, differences between couples living in different destinations could reflect Tiebout sorting. To evaluate whether couples differ in their behavior already in Denmark, so that couples with lower female labor force participation would self-select into destinations with wider wage differences and less generous child care pro- vision, we analyzed female labor force participation in Denmark, parallel to the analysis abroad in Table A11. The results suggest that there are no statistically significant differences between couples migrating to different destinations before emigration, once power type and the presence of children are controlled for. Furthermore, the effect of having young children in Denmark is very similar to the effect of having young children in other Nordic countries, which is again in line with what our theoretical model suggests. This does not exclude the possibility of Tiebout sorting in terms of preferences toward male breadwinner or dual-earner model, but it also shows that if there is such sorting, it is in terms of preferences whose realizations are not observable in Denmark. Fernández et al. (2004) find evidence for the US that wives of men whose mothers worked are themselves significantly more likely to work. Along the same lines, Moen et al. (1997) provide evi- dence that daughters’ and mothers’ gender role identities are closely related, including preferences towards the division of labor in the household. For Table 1.12, we take the first two specifications from Table 2.5 and Table 1.11 and add an additional regressor capturing labor force participation of the mother of the respondent at the age of 12. The idea is, first, to test whether preferences to- wards the division of labor are transmitted to the children. Columns 1 and 2 of Table 1.12 confirm that the respondent is more likely to prefer a male breadwinner model if the parents’ household economy was more traditional - after including the same covariates as in Table 1.11. Robustness analysis confirms that these results do not depend on whether the respondent was male or female. However, when explaining actual female labor force participation abroad, the coefficient estimate for labor force participation of the respondent’s mother becomes insignificant. As an additional test of Tiebout sorting, we analyze the effects of mother’s labor force participation on subsequent 42 1. International Family Migration and the Dual-Earner Model choice of the destination country. Table A12 relates labor force participation of the respondent’s mother to the likelihood of living in one of the considered country groups. We do not find evidence for sorting according to this indicator into different destinations. Preferences towards the division of labor which were transmitted from the parents to the respondents do not help to explain much of the variation in female labor force participation abroad, indicating that labor supply decisions and the division of labor among the partners abroad are rather shaped by economic incentives. Non-Nordic All Countries countries Power couple 0.0136 0.0774 -0.0485 0.100 (0.0541) (0.0586) (0.0668) (0.0738) Female power couple 0.0690 -0.00529 0.0607 -0.0251 (0.0691) (0.0845) (0.0933) (0.116) Male power couple -0.184*** -0.149* -0.113 -0.115 (0.0694) (0.0782) (0.0791) (0.0878) Youngest child 0-6 -0.271*** -0.230*** -0.321*** -0.295*** (0.0504) (0.0523) (0.0644) (0.0687) Youngest child 7+ -0.137*** -0.207*** -0.139** -0.234*** (0.0450) (0.0503) (0.0573) (0.0623) FLFP rate in country 0.318 0.422 -0.3730 0.0379 of residence (0.340) (0.405) (0.3886) (0.577) GINI -2.161*** -1.544** (0.488) (0.727) Constant 0.697*** 1.285*** 1.009*** 1.296*** (0.207) (0.277) (0.229) (0.285) R-squared 0.082 0.144 0.077 0.099 Observations 504 434 399 337 Notes: OLS estimation. Robust standard errors in parentheses. FLFP rate according to estimates by International Labour Organization for 2008. *** Significant at the 1 percent level. ** Significant at the 5 percent level. * Significant at the 10 percent level. Table 1.13: Linear probability regressions: Actual female labor force participation of emigrants and female labor force participation rates in the country of residence. A potential alternative explanation for our findings might be that labor force participation among females in our sample converges systematically to average female labor force participation rates in the destination countries. Among the native population female labor force participation rates are generally higher in the Nordic countries while they are lower in other Western countries and the 1.5 Econometric Analysis 45 Nordic Other West. Non-Western All countries countries US countries countries Power couple 0.0772 0.0761 -0.144 0.103 0.195* 0.115 (0.0722) (0.0687) (0.143) (0.161) (0.100) (0.155) Female power couple 0.0269 0.0417 -0.0147 0.0174 0.0743 -0.0650 (0.0987) (0.0743) (0.175) (0.205) (0.143) (0.269) Male power couple 0.240*** 0.227*** 0.270 0.317* 0.279** 0.195 (0.0776) (0.0723) (0.169) (0.166) (0.107) (0.145) Youngest child 0-6 0.304*** 0.295*** 0.215* 0.569*** 0.306*** 0.0599 (0.0680) (0.0651) (0.120) (0.189) (0.100) (0.152) Youngest child 7+ 0.378*** 0.350*** 0.319** 0.399** 0.411*** 0.127 (0.0620) (0.0600) (0.125) (0.182) (0.0871) (0.149) GINI 1.593*** (0.422) Nordic countries -0.153*** (0.0591) United States 0.0957* (0.0560) Non-Western countries 0.176*** (0.0563) Constant 0.138 0.387*** 0.404*** 0.326* 0.269*** 0.737*** (0.148) (0.0762) (0.149) (0.179) (0.0989) (0.158) Observations 351 379 99 47 193 40 R-squared 0.186 0.201 0.109 0.308 0.201 0.116 Notes: OLS estimation. Robust standard errors in parentheses. Sample: Both partners work abroad *** Significant at the 1 percent level. ** Significant at the 5 percent level. * Significant at the 10 percent level. Table 1.15: Linear probability regressions: Male partner contributes more to joint household economy in 2008 among dual-earner couples. 46 1. International Family Migration and the Dual-Earner Model dependent variable. Country dummies indicate a difference between the other Nordic countries, the US and the remaining Western countries. Dual-earner couples in the Nordic countries are less likely to have joint household economies where the male partner contributes more. Dual-earner couples in the US are slightly more likely to have joint household economies where the male partner contributes more, although the effect is only statistically significant at the 10% level. A separate estimation for the different country groups reveals that the presence of children is related to a higher male contribution to joint household economies among dual-earner couples in all West- ern countries. In non-Western countries, the presence of children is not related to higher male household economy contributions; however, the intercept estimate indicates that a higher male contribution in these countries is generally more frequent. Tables A14 and A15 show that our results still hold when including an indicator variable for whether the male partner contributed more to joint family income in Denmark before migration. If this was the case in Denmark, it is significantly more likely that the male contributes more in 2008, too. 1.6 Conclusion We analyzed motivations to emigrate, labor market participation and household economies of Danish couples who have emigrated between 1987 and 2002 and stayed abroad at least until 2008. Denmark is one of the most gender-equal countries worldwide, with a high female labor force participation rate. Despite this, we found big gender differences in main motivations to emigrate and in labor force participation abroad. Own work was the main reason to emigrate for 74.7% of men and 21.7% of women, partner’s job or other family reasons for 6.0% of men and 53.0% of women. Accordingly, Danish emigrant couples often adopt more traditional gender roles in non-Nordic countries abroad with a male breadwinner and the female staying at home. In the United States, the male works in almost all couples and the female in half. In the other Western countries, both partners work in two thirds of couples, and the male works while the female stays at home in one quarter of couples. The male stays at home and the female works in less than one 1.6 Conclusion 47 percent of couples that have emigrated. As a result of reduced female labor force participation, couples increasingly pool their earnings: the share of couples with separate economies declined from 12.6% before migration to 2.5% abroad, and the share of couples in which the male brings in all the money increased from 3.1% to 24.3%. Lower female labor force participation could reflect both, demand and supply side effects. Po- tential legal constraints or a lack of labor market opportunities could force the female partner out of the labor force, but it could also be that lower female labor supply is an adjustment to wider wage differences abroad and couples preferring a traditional male breadwinner model would be more likely to self-select into emigration. Our results suggest that both demand and supply side mechanisms are important. Demand side restrictions appear to play a role especially in non- Western countries, where the share of couples in which only the male works is considerably larger than the share of couples stating that they prefer such a situation. In the Western countries, preferred and actual female labor force participation are closely in line, suggesting the primacy of labor supply decisions. Our theoretical framework illustrates the role of incentives in this context: Wider income differences as well as higher prices for child care can economically rationalize that the female partner reduces labor supply or stays out of the labor force. The empirical findings for female labor force participation abroad confirm the hypotheses derived from our model. After migration we observe decreased female labor force participation in countries with wider income differences. Female labor force participation is lower in the non-Nordic countries where incomes are more dispersed, like in the US. Reduced female labor force participation in the US and non- Western countries is particularly evident among couples with pre-school age children. We do not find direct evidence for Tiebout sorting in terms of differences in pre-migration behavior. There are no big differences in female labor force participation rates in Denmark before emigration for couples living later in different countries. However, we observe that a significant fraction of couples with children migrating outside other Nordic countries viewed considerations related to children as a reason to emigrate, and very few as a reason against emigration. Among couples 50 2. Family Decision-Making on International Migration migration preferences. Using unique survey data we are the first to provide empirical evidence on the partners’ preferences towards joint emigration. Our empirical analysis uses survey data on Danish emigrants, restricting the attention to 522 respondents who had the same partner at the time of the survey as at the time of emigration. Denmark has one of the highest female labor force participation rates among OECD countries reaching 76% in 2010 (OECD 2011), although many women work only part-time. On top of this, the country is one of the world leaders in gender equality, having the third place in the United Nations Human Development Report (Klugman 2011). Therefore, we expect that family migration from Denmark would be more responsive to female preferences than family migration from less equal societies. Our model analyzes migration probabilities of singles and dual-earner couples in which both part- ners work. Job offers abroad depends on wage at home and an individual-specific component which may obtain positive or negative values. Our model predicts that in dual-earner couples, a higher own wage decreases the own probability of becoming a tied mover, but increases the partner’s probability of becoming a tied mover. Higher household surplus increases both partners’ proba- bilities of becoming a tied mover. The reason is that if household surplus increases, it becomes more likely that the partner who would gain privately from migration is willing to compensate the partner who would lose. Despite compensation, it is possible that the accompanying spouse is left worse off after migration (but never worse off than the outside option of staying and splitting up). We find that emigration is typically a shared family preference among the Danish emigrant cou- ples. Nonetheless, men are usually more strongly in favor of emigrating. If there is disagreement on emigration, it is usually the female partner who would have preferred to stay in Denmark. The emphasis on the male’s preference is strongest if the male partner has a college degree. However, even among couples in which the female partner is relatively higher educated, it is more common that the male rather than the female preferred to emigrate. Therefore, women seem to be more 2.1 Introduction 51 often "tied movers" even when being better educated. This finding is most pronounced among male power couples with children. When comparing different destinations, the share of couples with the male having the stronger preference to emigrate is somewhat lower and the share of couples with the female having the stronger preference to emigrate is somewhat higher to other Nordic countries than to the United States, the other Western countries or the non-Western countries, but even among couples migrating to other Nordic countries the male partner’s preference was about three times more often stronger than the female partner’s. We test our model’s hypotheses among dual-earner couples in which both partners work more than 60% of full working time in the year before emigration. Our results confirm that the likelihood for the male partner being in favor of emigration while the female partner had less strong preferences or was even against emigration decreases with earnings of the female partner. This is in line with predictions from the model. Theoretically, already Mincer (1978) and Mont (1989) showed that conflicting location preferences among partners lead to higher opportunity costs for relocation compared to single households. They abstracted from the possibility of separation and assumed that both partners agree to maxi- mize joint family income. Guler et al. (2012) analyze couples’ location constraints by developing a joint search theory. In their framework an external job offer to one partner changes the reservation wage of the other partner who might become a tied mover and search for a job at the new location if his or her initial earnings were sufficiently low. The model also builds on a unitary household in which both partners pool their incomes. Individual earnings losses of a "tied mover" would always be fully compensated in such a framework. Becker (1974) argues, however, that considering the household as a unitary decision-making agent imposes strong assumptions on individual prefer- ences. Contributions by Manser and Brown (1980) and McElroy and Horney (1981) explain family decision-making in a cooperative Nash bargaining model. Their framework allows to analyze the distribution of income within the family according to the partners’ outside options and their bar- gaining powers. Assuming full commitment, e.g. through the possibility of ex-ante transfers, the 52 2. Family Decision-Making on International Migration cooperative Nash bargaining model always implies efficient bargaining outcomes. The cooperative bargaining approach was conceptualized generalized further by Chiappori (1988, 1992) and Apps and Rees (1988) as the collective family model. In a dynamic context though, where commitment is not possible, Lundberg and Pollack (2003) argue that bargaining family bargaining might also lead to inefficient outcomes. A comprehensive overview on cooperative and non-cooperative bar- gaining in the household can be found in Browning, Chiappori and Weiss (2014). In the context of long term international migration we will formalize individual preferences and decision-making on migration among couples using a cooperative Nash bargaining model. The derived hypotheses are then tested with the survey data on Danish emigrant couples, confirming the theoretical pre- dictions. We restrict our analysis to one-time decision on whether to migrate. While there are a few papers that have analyzed repeated migration decisions (e.g. Gemici, 2011), a challenge in our context is that we cannot observe how potential job offers abroad would change over time. We re- frain from presenting a dynamic model of repeated decisions as our data would not allow testing it. The rest of the chapter is organized as follows. Section 2 presents a theoretical model on family bargaining and derives testable hypotheses from it. Section 3 describes our data and 4 presents stylized facts. In the light of our theoretical bargaining framework we present some empirical evidence on Danish emigrant couples in section 5. Section 6 concludes. 2.2 The Model 55 utility possibility frontier. This means that the migration decision is made to maximize the sum of individual incomes, net of eventual migration costs, and household surplus. Hence, with linear, additively separable utilities, joint emigration requires that the sum of both partners’ gains from migration is positive, i.e. xawa + xbwb − 2c > 0. (2.1) Staying in the home country, without migration incentives for neither partner the couple bargains under the following resource constraint: uc0a + uc0b = wa + wb + h. (2.2) If (2.1) holds and joint migration to 1 can be Pareto improving for both partners the resource constraint for bargaining is uc1a + uc1b = (1 + xa)wa + (1 + xb)wb − 2c+ h. (2.3) We consider a divorce-threat Nash bargaining framework. If (2.1) holds the couple maximizes a Nash bargaining function s.t. (2.3). The threat point of each partner is determined by his or her outside option, i.e. the individual migration incentive and optimal location choice as single. The partners’ exogenous bargaining powers are α for a and 1− α for b with α ∈ [0, 1]. In case neither partner has a migration incentive, or in the absence of migration possibilities, the Nash bargaining solution maximizes (uc0a − wa)α(uc0b − wb)1−α s.t. (2.2). This yields individual utilities of uc0a = wa + αh, uc0b = wb + (1− α)h. 56 2. Family Decision-Making on International Migration If both partners have an individual incentive to emigrate, there is no conflict of interest on the location and the Nash bargaining solution maximizes (uc1a − (1 + xa)wa + c)α(uc1b − (1 + xb)wb + c)1−α s.t. (2.3). The solution to the bargaining problem is then uc1a = (1 + xa)wa − c+ αh, uc1b = (1 + xb)wb − c+ (1− α)h. If (2.1) holds and partner a has a migration incentive but b not, the Nash solution maximizes (uc1a − (1 + xa)wa + c)α(uc1b − wb)1−α s.t. (2.3) if the couple emigrates. If (2.1) holds and partner b has a migration incentive but a not, the Nash solution maximizes (uc1a − wa)α(uc1b − (1 + xb)wb + c)1−α s.t. (2.3) if the couple emigrates. This yields uc1a = (1 + xa)wa − c+ α(h+ xbwb − c), uc1b = wb + (1− α)(h+ xbwb − c). 2.2 The Model 57 The corresponding result for a as a tied mover is uc1a = wa − c+ α(h+ xawa − c), uc1b = (1 + xb)wb + (1− α)(h+ xawa − c). These results yield that income losses of the tied mover are shared among the partners according to the bargaining powers. In this framework the tied mover is always worse off in terms of util- ity, compared to the situation before migration. He/she receives his/her outside option plus the remaining share of the household surplus net of own income losses. In case both partners would also migrate as singles, i.e. neither of them is a tied mover, there are no intra-family transfers. The partners divide the household surplus as in the home country according to the sharing rule derived above. From this bargaining solution we can also derive the sufficient conditions for joint emigration and household stability: The losses of a potential tied mover must be smaller than joint household surplus. In case b would face individual income losses, i.e. xbwb − c < 0, it has to hold h+ xbwb − c > 0. (2.4) The corresponding condition for a as a tied mover is h+ xawa − c > 0. (2.5) If (2.1) is satisfied but (2.4) or (2.5) is not, the partner who wants to emigrate could improve by migrating alone and the tied mover would be better off staying behind than migrating. Then, the couple would dissolve and give up joint household surplus h. This illustrates that unequal gains from migration can cause relationship instability, as pointed out by Mincer (1978) and Gemici (2011). Note that these conditions are independent of the partners’ bargaining powers, if coordi- nation on an efficient outcome and ex-ante transfers are possible. 60 2. Family Decision-Making on International Migration provides calculations for different corner solutions with x̄wb − c − h < 0 (and x̄wa − c − h < 0), including a distinction between the case of relatively small and large wage differences between a and b which becomes relevant then. Total Joint Emigration Probability The three cases together describe all possible events in which the couple emigrates: Without a tied mover, with a as a tied mover and with b as a tied mover. The probability of joint migration for the couple is, thus P4(joint emigration) = P1 + P2 + P3 = h(wax̄+ wbx̄− 2c− h) wawb + (x̄− c wa )(x̄− c wb ). Figures 2.1 and 2.2 illustrate the different cases in which joint migration occurs. They plot partner a’s job opportunities abroad on the horizontal and partner b’s job opportunities abroad on the ver- tical axis. As the distribution of xa and xb is uniform over the interval x to x̄ the surface indicated with P1, P2 and P3 corresponds to the migration probabilities derived above. The illustration in Figure 2.1 refers to the case in which wa = wb and h > c; Figure 2.2 provides an example for a situation in which wa > wb and h < c. 2.2 The Model 61 Figure 2.1: Conditions for joint emigration of a couple (illustration for wa = wb and h > c). 62 2. Family Decision-Making on International Migration Figure 2.2: Conditions for joint emigration of a couple (illustration for wa > wb and h < c). 2.3 Data 65 4. The impact of an increase in migration costs for both partners, for example, due to the presence of children, is unclear for the conditional probability of being a tied mover. Higher migration costs reduce the unconditional probability for being a tied mover. 2.3 Data We study household decision-making using unique survey data on Danish emigrants who had em- igrated in 1987, 1988, 1992, 1993, 1997, 1998, 2001 or 2002, and had not returned to Denmark by 2008. The survey was planned by Munk and Poutvaara within the project "Danes Abroad: Economic and Social Motivations for Emigration and Return Migration", financed by the Danish Social Science Research Council. The survey was carried out by Statistics Denmark. A detailed description on the data collection can be found in Poutvaara et al. (2009) for details on the sample used in the following analysis, see Munk, Nikolka and Poutvaara (2014). We focus on long-term emigration decisions of couples. The survey data provides information on several pre-migration characteristics of the respondents like household composition, education and work situation in Denmark. Most importantly, the survey respondents were asked about their motives and preferences for emigration. We restrict our sample to respondents in a long-term relationship being together with their partner since the time of emigration until the survey took place in 2008. The reason for this restriction is that if the couple would have separated between the time of emigration and survey conduction, the respondent might interpret the partner’s prefer- ences and an eventual conflict at the time of emigration in the hindsight of the relationship having ended. We also require that the respondent and the partner had lived together before emigration to focus on joint migration decision-making in the household as modeled above. Furthermore, we restrict the analysis to partners who are Danish citizens. The reason for this restriction is that in international couples, emigration from Denmark might imply returning to the home country of the partner, making a migration decision qualitatively different. Finally, we link respondents and their partners with the Danish full population register data on age, gender and earnings before migration. The remaining analysis is based on 522 respondents, as well as their partners, who 66 2. Family Decision-Making on International Migration satisfied all the above restrictions. In the subsequent analysis, we recoded the answers the survey respondents gave on their own and their partners’ situation and preferences. The recoded answers then refer to the male or female partner. For example, if a male answered that "I was in favor of migration, while my partner would have preferred to stay in Denmark", this is recoded as "Disagreement, female would have preferred to stay". 2.4 Descriptive Statistics Our data allow us to gain important insights into intra-family decision-making on migration of Danish emigrant couples and to test the predictions derived from the theoretical bargaining model. We are able to link the partners’ preferences to their individual characteristics, in particular their pre-migration earnings in Denmark. To our knowledge there is no empirical evidence on intra- household bargaining on international migration decisions so far. As a starting point for analyzing the complexity of household decision-making this section provides some descriptive statistics presenting the data and the variable used in subsequent econometric analysis. Our first question of interest is to what extent emigration was a shared preference among the partners. We group migration preferences in five categories. There are three categories con- taining joint migration preferences: equal migration preferences among the partners, agreement but stronger male preference, agreement but stronger female preference towards emigration. In two categories, there are couples having disagreed on emigration and either the female or the male partner would have preferred to stay. Table 2.1 provides an overview on the distribution of migra- tion preferences among the partners in our sample. A majority of 52.0% of respondents said they had equal preferences towards migration. In the case of unequal preferences it was mostly the male partner who was in favor of emigration (39.0%). 2.4 Descriptive Statistics 67 No child in DK Children in DK Total Equal preferences 52.8% 51.1% 52.0% Agreement, stronger male preference 29.6% 34.9% 32.6% Agreement, stronger female preference 9.0% 6.5% 8.5% Disagreement, female would have preferred to stay 7.9% 6.2% 6.4% Disagreement, male would have preferred to stay 0.7% 0.3% 0.6% Observations 244 278 522 Source: Survey data. Table 2.1: Migration preferences and child presence. 6.4% stated that only the male wanted to emigrate while the female disagreed on migration. The female wanted to emigrate while the male disagreed in only 0.6% of the cases. Stronger female migration preferences were more frequent among couples without children compared to couples with children. For couples with children stronger male migration preferences were relatively more frequent. Most couples, however, reported that migration was a shared preference. 93.1% of re- spondents report joint agreement on migration. Moreover, in Table 2.1 we do not observe a big difference in relative migration preferences between couples with and without children in Denmark. In the following, we analyze the partners’ migration preferences in the light of their pre-migration characteristics, as motivated by the theoretical model. We will emphazise especially migration decision of dual-earner couples in which partners potentially face more divergent preferences on international migration. Table 2.2 relates migration preferences of the partners to the power type of the couple. Following Costa and Kahn (2000) we will refer to different "power" types of couples in our analysis according to their level of education and earnings potential. Power couples are characterized by a college educated male and female partner. We refer to male or female power couples if only one partner holds a college degree. Low power couples are those where neither partner has completed a higher education. Among 38.9% of the low power couples the male partner preferred more to emigrate than the female, compared to a higher female migration preference for only 4.2%. For power couples and 70 2. Family Decision-Making on International Migration erences to be correctly assessed by one partner for most of the remaining observations in our sample. Median male income Median female income (monthly, DKK) (monthly, DKK) Equal preferences 27,734 15,152 Agreement, stronger male preference 28,538 16,434 Agreement, stronger female preference 25,501 12,254 Disagreement, 32,584 13,929 female would have preferred to stay Disagreement, 26,265 24,299 male would have preferred to stay Source: Survey data. Table 2.3: Income and migration preferences. Our theoretical framework presented above relates joint migration probabilities and the partners’ migration preferences to their pre-migration earnings. In Table 2.3 we present the median of monthly incomes in Danish Krone before emigration of the male and female partners according to migration preferences in our sample.4 We find that median male income is particularly high in couples where the male partner has stronger preferences towards emigration. This pattern does not seem to hold for female median income. A high female median income stands out only among the very few cases in which the male partner disagreed while the female partner had a strong preference towards emigration. Table 2.4 provides insights into the partners’ migration preferences for different destinations: the United States, Nordic countries, other Western countries and non-Western countries. Stronger male migration preferences are slightly more frequent among couples that emigrated to the US, and less frequent among couples that emigrated to other Nordic countries. Among couples having migrated to one of the other Nordic countries the share of stronger migration preferences of the female partner is relatively higher, 13.2%, compared to the 5.1% for the US. It is 7.0% for the other Western countries and 11.8% for the non-Western destination countries. On the other hand, the share of stronger male migration preferences is smallest among couples having migrated to the 4Earnings presented are annual earnings divided by 12 months from the Danish administrative register data. Annual earnings are labor income plus non-negative values for freelance income in the year before the couple has left Denmark. Earnings are deflated using the year 2000 as the reference year. 2.4 Descriptive Statistics 71 Other Non- Nordic Western Western countries US countries countries Total Equal preferences 52.1% 51.3% 52.5% 51.3% 52.0% Agreement, stronger male preference 28.1% 34.6% 33.8% 31.5% 32.6% Agreement, stronger female preference 12.4% 5.1% 6.2% 11.8% 8.5% Disagreement, female would have preferred to stay 6.6% 7.7% 6.6% 5.3% 6.4% Disagreement, male would have preferred to stay 0.8% 0.0% 0.8% 0.0% 0.6% Observations 114 73 259 76 522 Source: Survey data. Table 2.4: Migration preferences by destination country. Nordic countries. However, we would have expected gender differences with respect to migration preferences to be even smaller to the Nordic countries compared to the US and other destinations. In general, labor market policies in the Nordic countries, in particular for dual-earner couples, are more family friendly, for example due to the provision of public day care services. However, even among couples having migrated to the Nordic countries stronger male migration preferences are clearly more frequent than stronger female preferences. Still, for all destination countries a clear majority of couples emigrated in joint agreement, to all destination country groups 51.3% to 52.5% of both partners had even equally strong migration preferences. 72 2. Family Decision-Making on International Migration 2.5 Econometric Analysis Based on our theoretical framework and the descriptive analysis, we expect attitudes towards migration to be related to family characteristics and earnings potentials of the partners, in par- ticular in dual career couples. As we observe only very few couples where the female partner has stronger preferences towards joint emigration, we focus on stronger male migration preferences in the econometric analysis. In Table 2.5 we estimate a linear probability model for all couples for which we retrieve labor market information from the register data. Compared to the descriptive analysis the estimation results provide a more detailed insight on the relationship between pre-migration characteristics, in particular earnings potential of the partners, and the binary dependent variable for stronger male preferences towards joint emigration. Table 2.5 reports the regression results on a reduced sample of 449 observations. The reason for this is that we were not able to find administrative labor market data for partners in the sample. Our model predicted that the probability of being a tied mover conditional on joint emigration increases with the partner’s wage and decreases with the own wage. We include log earnings for both partners adding 1 to all values and thus also includ- ing log earnings of partners without any income in the year before emigration. We find that the probability for stronger male migration preferences in our sample increases with log male earnings. This finding is robust and statistically significant across all specifications in Table 2.5. However, we do not find any significant effect of female wages on stronger male migration preferences. We additionally include the power type of the family as a dummy variable in our regressions. Stronger male migration preferences are more likely among male power couples. We saw in our descriptive analysis that this seems to be driven to a large extent by the presence of children in the household. There are no clear results on the effect of female power couples and power couples on migration preferences in these specifications. Furthermore, estimation results do not reveal a clear effect for the presence of children in the household on stronger preferences of the male partner towards emigration. In the last specification, we include a dummy variable for la- 2.6 Conclusion 75 (1) (2) (3) Log female income in DK -0.126** -0.103** -0.110*** (0.0523) (0.0546) (0.0544) Log male income in DK 0.00883 0.0314 0.0514 (0.0475) (0.0500) (0.0506) Child 0-6 in DK 0.0339 0.0153 (0.0358) (0.0375) Child 7+ in DK 0.104* 0.0664 (0.0639) (0.0662) Cohabiting 5+ years in DK 0.0771** (0.0332) Observations 235 235 235 R-squared 0.031 0.027 0.040 Notes: OLS estimation. Constant included. Robust standard errors in parentheses. *** Significant at the 1 percent level. ** Significant at the 5 percent level. * Significant at the 10 percent level. Table 2.6: Linear probability regressions: Stronger male preferences towards joint emigration, dual-earner couples. 2.6 Conclusion To shed more light on the determinants of relative migration preferences among the partners in migrating dual career couples, we developed a model for family bargaining on international migra- tion. Most of the previous literature has only considered unitary decision-making of households in the context of migration. Our model endogenizes family stability and explains why a tied mover might face utility losses through migration, but still migrate with the family. From this framework we derive predictions on how the probability of becoming a tied mover, and thus rather preferring not to migrate, is related to earnings of partners in dual-earner couples. We are able to test the hypotheses with survey data on Danish couples having jointly emigrated. Empirically, we found that among the partners emigration is mostly a shared preference. How- ever, in many couples the partners did have different preferences towards joint emigration. Despite Denmark being one of the most gender-equal countries worldwide, with a high female labor force participation rate, we found that family migration is more often driven by the male partner. If 76 2. Family Decision-Making on International Migration there was disagreement on the migration decision, it was mostly the male who preferred to emi- grate and the female who would have rather stayed in Denmark. Gender differences in migration preferences were less strong among couples having migrated to one of the Nordic countries. Regression analysis shows that an increase in earnings of the male partner in Denmark is asso- ciated with higher probability that the male has the stronger preference towards emigration in the couple. This supports our theoretical predictions. Furthermore, we considered the partners emigration preferences in the light of the couples’ power types in our analysis, i.e. their levels of education. Our analysis revealed that stronger male preferences towards migration are more fre- quent among male power couples, in which only the male partner is college educated. Low power couples, female power couples and power couples do not differ much from each other in terms of migration preferences. For the sample of dual-earner couples we are able to directly test the hypotheses derived from the theoretical bargaining framework. Among these couples, lower female earnings are associated with stronger preferences of the male partner towards emigration. This is in line with theory. More- over, we find that stronger male preferences to emigration are more likely in case the couple was cohabiting for more than 5 years in Denmark, indicating that family migration with a tied mover is more likely if household surplus is large. Our study provides novel insights on couples’ migration decisions in case both partners have divergent individual labor market opportunities from migra- tion. Our empirical analysis is able to confirm theoretical implications from a bargaining model on migration decisions, a more general framework compared to a unitary household model. Chapter 3 Family Return Migration 3.1 Introduction This chapter investigates the role of family ties for return migration of immigrant couples. A major part of migration flows to OECD countries is of temporary nature (Dustmann 1995, 1997; Dustmann and Görlach, 2016). The effect of family ties on migration decision has been studied both theoretically as well as empirically (Mincer, 1978; Mont, 1989; Borjas and Bronars, 1992; Tenn, 2010; Gemici, 2011; Junge et al., 2014). As for the initial emigration decision in the first place, family ties can be expected to also play an important role for the decision to return home (Dustmann, 2003). However, the role of the family for temporary migration decisions has only received limited attention so far. There exists an extensive literature focusing on migrants’ ties to the home country when family members were left behind (For a survey see Docquier and Rapoport, 2006). In contrast to this literature, this chapter analyzes joint migration decisions of partners who immigrate together and decide whether to jointly return to the country of origin. In this con- text there is only little evidence on family related considerations for return migration so far. The present analysis is going to use administrative data from 1973 to 2010 to study return decisions of immigrant couples living in Denmark. Return plans are likely to affect outcomes on the family level like fertility and labor market attachment of partners challenging empirical identification of causal mechanisms. Many observable characteristics that potentially influence family return deci- sions are endogenously determined. 80 3. Family Return Migration children, too. Overall, the timing of family return decisions suggests an association with the chil- dren’s schooling. Nevertheless, robust evidence for the link between schooling quality and return migration of families cannot be established here. Factors related to home country ties, e.g. cultural identity might also have an effect on the de- cisions of families to return even among migrants that migrated together with their partner and children to Denmark. Fernandez (2007) and Fernandez and Fogli (2009) study the impact of cul- tural identity among immigrants as well as their descendants and find that economic decisions of first and even second generation immigrants in the host society are strongly associated with cultural background. This is likely to also affect the propensity to return. Sajons (2015) studies children’s eligibility to citizenship and return migration of families from Germany. Eligibility to host country citizenship is found to reduce return rates possibly through considerations related to the identity. The present analysis provides evidence that factors related to culture and identity are likely to be relevant for return decisions among subgroups of immigrant couples from Denmark as well: Relatively higher return probabilities for Turkish immigrants with a girl compared to those with a boy as the oldest child suggest an effect of considerations related to parents’ preferences towards gender roles and identity. Given that Denmark is a host country with high female labor force participation rates and high quality of schooling and a gender equal society, it seems unlikely that the differences in relative return propensities are due to labor market or schooling considera- tions concerning daughters in the family. Moreover, the effect of a policy change regarding family reunification rules also suggests that arguments related to cultural identity and ties to the home country which are not related to education or institutional quality are relevant for return decisions of families. The third contribution of this study links to the literature on temporary migration studying immi- grants’ self-selection into return migration on labor market characteristics. The present analysis studies self-selection into return on observable characteristics of the partners, in particular labor income, separately for all couples, for dual-earner couples, for couples with children and by different 3.1 Introduction 81 countries of origin. First of all, results reveal strong self-selection into return migration on primary earner’s income for couples with male as well as female primary earner. Self-selection patterns are strongest for non-Western countries. These results are in line with the literature on return mi- gration, in general, without addressing the role of family ties. Borjas and Bratsberg (1996) argue that the self-selection into return-migration accentuates selection patterns of the initial migration flow between two countries. Along these lines, Denmark, with a narrow income distribution would attract relatively more immigrants at the low end of the income distribution (Pedersen, 2005). In line with theory, self-selection patterns of returning migrant couples according to primary earner income is strongly positive to non-Western countries where incomes are often more unequally dis- tributed. Positive self-selection of immigrants into out-migration has also been shown for the case of Norway (Longva, 2001). For Sweden, Nekby (2006) finds U-shaped selection patterns with pos- itive self-selection of immigrants into return migration at the upper end of the income distribution. Extending the literature regarding individual self-selection into return migration with a house- hold level perspective can provide additional insights on how family ties shape decision making on return migration. In general, families migrating together often have to overcome co-location problems due to different individual migration incentives between the partners. Thus partners ex- perience unequal labor market gains from migration and one partner often becomes a tied mover (Mincer, 1978). While family ties have generally been found to reduce mobility (Mincer, 1978; Frank, 1978), the effect of family ties on self-selection patterns regarding individual characteristics of the partners are less clear. Junge et al. (2014) show that self-selection for emigrant couples from Denmark according to primary earner’s income is stronger than self-selection patterns for singles. Borjas and Bronars (1992), on the other hand, find weaker self-selection of immigrants with family ties into the US. They argue that family migrants are selected more randomly as they are more likely not to migrate primarily due to own income incentives. Co-location problems and divergent individual gains from migration might also affect return decisions from the host country. The effects to expect related to family ties of migrants and the partners’ self-selection into return migration on labor market characteristics are not clear ex-ante though. 82 3. Family Return Migration Analysis shows that individual and family characteristics of both partners contribute to explaining joint return propensities. Either partner being out of the labor force is associated with higher joint return propensities for immigrant couples from the Nordic, the other Western as well as the non-Western countries. Moreover, the positive correlation between higher propensity to return and higher income of the primary earner is lower among dual-earner couples in which both partners work more than 60% of full working time in the labor market. This finding is driven by immigrant couples from the non-Western countries. On top of this, the presence of children also seems to affect self-selection on primary earner’s income into return migration. Results, which are again driven by the non-Western sending countries, show that the children in the household weaken self- selection into return migration on primary earner income. A possible explanation for this is that other factors that play a role for return migration of families with children, which are uncorrelated with primary earner income, reduce selection patterns compared to singles and to couples without children. As outlined, the following analysis is going to show along different dimensions how family con- siderations are related to return migration decisions of immigrant couples. For policy makers it is of utmost importance to understand return decisions of immigrants in order to design policies aiming at attracting and retaining immigrants to overcome skill shortages, demographic challenges and to foster economic growth. To this, the present analysis shows that considerations related to the family play an important role and have to be taken into account in this context. The rest of the chapter is organized as follows. Section 2 provides background information on immigration to Denmark and introduces the data used in the empirical analysis. Section 3 presents stylized facts and descriptive statistics, section 4 econometric analysis and section 5 concludes. 3.2 Data The Danish administrative data contains information on all registered immigrants living in Den- mark in a given year. According to the definition of Statistics Denmark a person is considered as 3.2 Data 85 The sample will be restricted to individuals who are between 25 and 59 in order to capture the working age immigrant population. Furthermore, individuals have to be at least 18 when immi- grating; this ensures that they most probably migrated for own reasons to Denmark and did not come as children with the family. The major part of the empirical analysis restricts attention to a sub-sample of immigrants with a partner from the same country both fulfilling the age restriction above. Return behavior of couples with partners from different countries of origin is likely to be qualitatively different and should be analyzed separately, which is beyond the scope of this study. Unique individual and family identifiers make it possible to combine data for cohabiting partners as well as their children while they reside in Denmark.2 In order to allow for the possibility of sequential immigration of spouses (Borjas and Bronars, 1992), both partners do not necessarily need to have immigrated in the same year to Denmark. However, to be included in the analysis the partners have to cohabit immediately after the second mover immigrated. A return event in the subsequent analysis is defined as emigrating from Denmark to the country of origin. A couple returns if both partners migrate to their country of origin within the same year. Couples and singles are observed in the administrative registers over the observation period every year as long as their cohabitation status remains the same and as long as they reside in Denmark. Return migration patterns of couple households will be studied with respect to partner- and family characteristics and also be compared to single migrants. Returners and non-returners will be com- pared based on observable characteristics in the year before the potential return migration event. Results for returning to the country of origin in the following analysis are reported for returners not re-entering Denmark for the subsequent 5 years. 2Immigrants linked with a partner are either cohabiting at the same address, married or in a registered part- nership according to the administrative registers. Individuals in registered same-sex partnerships will be excluded because the number of observations is low in the immigrant population. 86 3. Family Return Migration 3.3 Descriptive Statistics Table 3.2 presents the data of the analyzed sample with the mentioned restrictions according to the origin countries of the migrants, separately for couples and singles. There are 318,377 individual- year pairs for male single migrants and 231,571 individual-year pairs for female single migrants in the data. There are 358,920 observations for individuals with a partner from the same coun- try of origin. According to the restrictions above in total 9,214 return events of couples can be observed. There are differences between the shares of the sending countries for singles and part- ners presented in Table 3.2. For Western countries the share of migrants in single households is higher than for migrants with a partner from the same country of origin. 5.2% of couples are from another Nordic country and 16.5% from one of the remaining Western countries. Among single males and females the corresponding shares are considerably higher. On the other hand, 78.3% of couple migrants are from the non-Western origin countries, compared to 47.2% among single males and 54.0% among single female. In particular for couple households in this group excluding the major refugee sending countries the most important sending country is Turkey (share of 29.6%). Singles Partners from males females same origin country Other Nordic countries 14.2 19.9 5.2 Other Western countries 38.5 26.0 16.5 Western Europe 33.8 22.1 14.8 US, NZ, CAN, AUS 4.7 3.9 1.7 Non Western countries 47.2 54.0 78.3 Turkey 7.6 5.7 29.6 Remaining countries 39.6 48.3 48.7 Observations 318,377 231,571 358,920 Source: Administrative data. Table 3.2: Origin countries. Table 3.3 reveals that return propensities of single and couple migrants differ considerably accord- ing to the country of origin. Among singles as well as couples from the non-Western countries returns to the home country are least frequent (2% for couples, 4% for single females and 7% for single males) while they are more likely for those from other Western countries (10% among all 3.3 Descriptive Statistics 87 groups) and most likely for those from other Nordic countries (18% for couples, 10% for female singles and 13% for male singles). In particular for couples returns are most frequent to the other Nordic countries while much less frequent among those from the non-Western countries. Of course, the average duration of stay varies between the different origin country groups, with migrant sin- gles as well as couples from Western European countries having on average shorter duration of stay than those from the non-Western countries. Potential reasons which might in general explain dif- ferences in temporary migration patterns between immigrants from the considered sending country groups are going to be addressed later. Due to the heterogeneity between the different countries of origin, much of the subsequent analysis is going to distinguish three groups of sending and return countries of migrants: The first group are other Nordic countries, the second group other Western countries consisting of the non-Nordic, Western European countries as well as Australia, Canada, New Zealand and the United States. Non-Western countries are the third group accounting ac- cording to observations in the data for a majority among couples as well as single migrants. Table 3.3 presents further average sample characteristics separately for singles and couples in the data. Females in couples are on average slightly younger while males are slightly older than in the corresponding sample of singles. Table 3.3 also reports the share of couples with children. With 78% a large fraction of couples has children below the age of 16 in the household.3 Children are also present in 33% of single female and 10% of single male migrant households. The income and tax register data provide information on labor market activity of the immigrant population in Den- mark. Table 3.3 shows that 23% of single men and women are out of the labor force in the sample. This share is higher among females in couples (33%), but lower among male partners (14%). The share of self-employed is relatively small in all groups. 44% of single men and 46% of single women work in the labor market, 30% of males and 31% of females in full-time employment.4 The share of employed is lower among females in couples (32%, full-time: 21%) and higher among males (49%, full-time: 38%). The share of couples in which both partners work full-time in the labor market is only 15%. The income register data reports annual gross labor and freelance income 3Older children are not directly recorded as household members and thus left out of the analysis. 4Full-time employment is defined as working more than 60% of the full-time equivalent working time in a given year.