Causes and consequences of the gender-specific migration from East to West Germany

Melzer, Silvia Maja Book Causes and consequences of the gender-specific migration from East to West Germany IAB-Bibliothek, No. 358 Provided in Cooperation with: Institute for Employment Research (IAB) Suggested Citation: Melzer, Silvia Maja (2016) : Causes and consequences of the gender-specific migration from East to West Germany, IAB-Bibliothek, No. 358, ISBN 978-3-7639-4104-9, W. Bertelsmann Verlag (wbv), Bielefeld, https://doi.org/10.3278/300902w This Version is available at: https://hdl.handle.net/10419/280211 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by-sa/4.0/ Silvia Maja Melzer 358 358 Causes and consequences of the gender-specific migration from East to West Germany Although German reunification took place 25 years ago, differences between East and West Germany still remain. After German reunification, people migrated from the east to the west because of differences in living standards and the new opportunities available. Silvia Maja Melzer analyses the causes and consequences of intra-German migration, both theoretically and empirically, from a gender-specific point of view and finds answers to the following questions: Which factors are decisive for the relocation of men and women? How does education influence the gender-specific decision to migrate? Who is relocating or commuting more often, men or women? In order to present a more comprehensive picture of gender-specific migration, contrasts are drawn respectively between men and women or singles and people in partnerships. How does the migration behaviour of East and West German couples and singles differ? What financial consequences result from migration? And: Are east-west migrants happier? ISBN 978-3-7639-4103-2 Causes and consequences of the gender-specific migration from East to West Germany 358 Causes and consequences of the gender-specific migration from East to West Germany Silvia Maja Melzer Dissertation zur Erlangung des Doktorgrades an der Fakultät für Soziologie der Universität Bielefeld 23. April 2013 vorgelegt von Silvia Maja Melzer geboren am 21. Juni 1979, in Danzig (Polen) Erstgutachter: Prof. Dr. Martin Diewald, Universität Bielefeld Zweitgutachter: Prof. Dr. Herbert Brücker, Otto-Friedrich-Universität Bamberg und Institut für Arbeitsmarkt- und Berufsforschung Herausgeber der Reihe IAB-Bibliothek: Institut für Arbeitsmarkt- und Berufsforschung der Bundes agentur für Arbeit (IAB), Regensburger Straße 104, 90478 Nürnberg, Telefon (09 11) 179-0 Redaktion: Martina Dorsch, Institut für Arbeitsmarkt- und Berufsforschung der Bundesagentur für Arbeit, 90327 Nürnberg, Telefon (09 11) 179-32 06, E-Mail: [email protected] Gesamtherstellung: W. Bertelsmann Verlag, Bielefeld (wbv.de) Rechte: Kein Teil dieses Werkes darf ohne vorherige Genehmigung des IAB in irgendeiner Form (unter Verwendung elek tronischer Systeme oder als Ausdruck, Fotokopie oder Nutzung eines anderen Vervielfältigungsverfahrens) über den persönlichen Gebrauch hinaus verarbeitet oder verbreitet werden. © 2016 Institut für Arbeitsmarkt- und Berufsforschung, Nürnberg/ W. Bertelsmann Verlag GmbH & Co. KG, Bielefeld In der “IAB-Bibliothek” werden umfangreiche Einzelarbeiten aus dem IAB oder im Auftrag des IAB oder der BA durchgeführte Untersuchungen veröffentlicht. Beiträge, die mit dem Namen des Verfassers gekenn zeichnet sind, geben nicht unbedingt die Meinung des IAB bzw. der Bundesagentur für Arbeit wieder. ISBN 978-3-7639-4103-2 (Print) ISBN 978-3-7639-4104-9 (E-Book) Best.-Nr. 300902 www.iabshop.de www.iab.de Bibliografische Information der Deutschen Nationalbibliothek Die Deutsche Nationalbibliothek verzeichnet diese Publikation in der Deutschen Nationalbibliografie; detaillierte bibliografische Daten sind im Internet über http://dnb.ddb.de abrufbar. Dieses E-Book ist auf dem Grünen Weg Open Access erschienen. Es ist lizenziert unter der CC-BY-SA-Lizenz. 3IAB-Bibliothek 358 Contents Acknowledgement .......................................................................................... 9 Chapter I: Introduction ............................................................................... 11 1 Motivation ................................................................................. 13 2 Contribution ............................................................................... 15 3 Historical context East-West migration in Germany ............. 18 3.1 Differences and similarities .................................................................. 18 3.2 East-West migration in Germany ....................................................... 19 4 The structure of this work ....................................................... 21 5 References .................................................................................. 24 Chapter II: Overview of migration theories .............................................. 29 1 Introduction ............................................................................... 31 2 Selection of migration theories ................................................ 31 2.1 Individual theories ................................................................................... 33 2.1.1 Neoclassical migration theory ............................................................. 33 2.1.2 New economics of migration ............................................................... 37 2.1.3 Signaling theory ....................................................................................... 41 2.1.4 Segmentation theory .............................................................................. 44 2.2 Migration of women ............................................................................... 48 2.2.1 Migration of unattached women ........................................................ 49 2.2.1.1 Why do women migrate less often?................................................... 49 2.2.1.2 Why do women migrate more often? ................................................ 50 2.2.2 Theories on family migration .............................................................. 52 2.2.2.1 Household economy ................................................................................ 53 2.2.2.2 Bargaining theory .................................................................................... 57 2.2.2.3 Gender role theory .................................................................................. 62 3 Critique ....................................................................................... 64 3.1 General critique........................................................................................ 64 3.2 Critique from a gender perspective ................................................... 65 4 Discussion................................................................................... 66 5 References .................................................................................. 68 Contents IAB-Bibliothek 358 4 Chapter III: Reconsidering the effect of education on East-West migration in Germany ................................................................ 79 Abstract .......................................................................................................... 81 1 Introduction ............................................................................... 81 2 Theory ......................................................................................... 84 2.1 Human capital theory ............................................................................. 84 2.2 Signaling theory ....................................................................................... 86 2.3 Segmentation theory .............................................................................. 87 2.4 Gender-specific differences .................................................................. 88 3 Data, measures, and methods ................................................... 89 3.1 Data ............................................................................................................. 89 3.2 Measures .................................................................................................... 90 3.3 Methods ..................................................................................................... 92 3.3.1 Random effects regressions .................................................................. 92 3.3.2 Selection bias ............................................................................................ 93 4 Results ........................................................................................ 95 5 Conclusion .................................................................................. 102 Acknowledgement .......................................................................................... 105 Funding .......................................................................................................... 105 6 References .................................................................................. 105 7 Appendix .................................................................................... 108 Chapter IV: Does migration make you happy? The influence of migration on subjective well-being .......................................... 111 Abstract .......................................................................................................... 113 1 Introduction ............................................................................... 113 2 Migration and subjective well-being ....................................... 116 3 Theoretical considerations ........................................................ 118 3.1 Subjective well-being of migrants ..................................................... 118 3.2 Group differences .................................................................................... 119 4 Data and methods .................................................................... 120 4.1 Data ............................................................................................................. 120 4.2 Methods ..................................................................................................... 123 Contents 5IAB-Bibliothek 358 5 Results ....................................................................................... 124 5.1 Descriptive results ................................................................................... 124 5.2 Analytical results ..................................................................................... 125 6 Conclusion .................................................................................. 130 Acknowledgement .......................................................................................... 132 7 References ................................................................................. 132 8 Appendix .................................................................................... 137 Chapter V: Reconsidering gender-specific commuting and migration between East and West Germany .............................................. 139 Abstract .......................................................................................................... 141 1 Introduction .............................................................................. 141 2 Previous research ...................................................................... 144 3 Theoretical framework .............................................................. 145 3.1 Mobility costs ........................................................................................... 145 3.2 Gender-specific differences .................................................................. 147 4 Data and methods ..................................................................... 148 4.1 Data ............................................................................................................. 148 4.2 Methods ..................................................................................................... 150 5 Results ........................................................................................ 151 5.1 Descriptive results ................................................................................... 151 5.2 Determinants of commuting and migration .................................... 153 6 Conclusion .................................................................................. 160 7 References .................................................................................. 162 Chapter VI: Why do couples relocate? Considering migration from East to West Germany ...................................................... 167 Abstract .......................................................................................................... 169 1 Introduction ............................................................................... 169 2 East to West migration in Germany ........................................ 170 3 Theoretical background ............................................................. 172 3.1 Household economic theory ................................................................. 172 Contents IAB-Bibliothek 358 6 3.2 Bargaining theory .................................................................................... 172 3.3 Gender role theory ................................................................................. 173 4 Family migration in the East German context ........................ 174 5 Data and variables ..................................................................... 176 5.1 Data ............................................................................................................. 176 5.2 Variables ..................................................................................................... 176 6 Methods ..................................................................................... 178 7 Results ....................................................................................... 179 7.1 Descriptive results .................................................................................. 179 7.2 Analytical results ..................................................................................... 180 8 Conclusions ................................................................................ 184 Acknowledgement .......................................................................................... 186 9 References .................................................................................. 186 Chapter VII: Explaining the puzzling effects of household migration: Why do East German women lose and West German women gain?.............................................................................. 191 Abstract .......................................................................................................... 193 1 Introduction ............................................................................... 193 2 Determinants and consequences of migration – literature review ........................................................................ 196 2.1 Mobility of individuals ........................................................................... 196 2.2 Mobility of households .......................................................................... 196 2.3 Explanation for different mobility patterns of men and women in households ............................................................................. 197 2.4 Gender arrangements in East and West Germany ......................... 198 3 An alternative approach: the bargaining model of migration ............................................................................... 199 3.1 Migration decisions from a bargaining perspective ...................... 199 3.2 Regional determinants and couples’ migration decisions in East and West Germany .................................................................... 201 4 Methods and data ..................................................................... 202 4.1 Data ............................................................................................................. 202 4.2 Methods ..................................................................................................... 204 Contents 7IAB-Bibliothek 358 5 Results ........................................................................................ 205 5.1 Determinants of couples’ migration in East and West Germany .......................................................................................... 205 5.2 Effects of couples’ migration on income .......................................... 209 5.3 Occupational structure before and after the move ....................... 211 6 Summary .................................................................................... 213 7 References .................................................................................. 214 Chapter VIII: Conclusion ................................................................................. 219 1 Summary of research ................................................................ 221 2 Contribution to East-West migration in Germany ................ 221 2.1 Why do East German women relocate more often to West Germany than East German men? ........................................... 221 2.2 Why do the migration patterns of East German single women and those in partnerships differ so substantially? ......................... 224 3 Contribution to the field ......................................................... 228 4 Limitations and future research ............................................... 229 5 References .................................................................................. 230 Abstract .......................................................................................................... 233 Kurzfassung .................................................................................................... 235 IAB-Bibliothek 358 14 Introduction Second, focusing on the migration of women as part of a family implicitly results in a lack of research on the migration of highly skilled women (Kofman 1998; Kofman and Raghuram 2009; Pedraza 1991), which is key because the availability of highly qualified workers is important for countries’ long-term economic growth (cf Lucas 1988) and because increasingly more countries are attempting to attract highly skilled migrants. In general, migration research pays scant attention to the fact that migrant women also hold professional and managerial positions (Kofman 1999).3 This is true even though the migration of highly skilled women combines the two most current trends in migration research: the feminization of migration and the migration of highly skilled workers (Dumont et al. 2007). Bringing women’s mobility into focus should therefore widen our understanding of migration, including the “brain gain, brain drain” dimension. Third, the main body of research seeks to explain the migration of women using qualitative studies. As Curran et al. (2006) note, by the mid-1990s the focus of migration studies shifted from research on the migration of women to research on gender and migration. Moreover, the methodological framework shifted from quantitative to qualitative research. The qualitative studies concentrate on the gender issues of migration, whereas the quantitative research generally fails to address gender differences. Quantitative research on women’s migration can help to deepen our understanding of migration by accounting for factors that influence men’s and women’s migration decisions at the micro, macro and meso levels differently. Finally, as Calavita (2006, p. 125) indicated, “If there is one bias that penetrates much of [migration] literature, it is an almost singular focus on immigrants who are poor”. The migration flows that receive the most attention originate in (poor) developing countries, where the migrants aim to relocate to (wealthy) industrialized and post-industrialized countries (e.g., Durand et al. 2001; Massey and Espinosa 1997; Taylor 1987). Migration within industrialized countries receives less attention; however, the reasons for migration may differ for people who relocate from or within developing and industrialized countries. Factors such as income levels and access to jobs, which generate migration in industrialized countries, may also lead to migration in developing countries. Nevertheless, it can be expected that the manner in which these factors influence the migration decisions varies according to the degree of industrialization, the culture and the political context of the country. 3 For research on the mobility of highly qualified women, see Hugo (1993); Kofman (1998); Kofman and Raghuram (2009); Pedraza (1991). 15Chapter 1 Contribution 2 Contribution This work aims to fill at least some of the gaps in the literature focusing on the East-West migration in Germany, which represents one of the recent migration flows in which women outnumber men. This study examines the reasons behind women’s higher migration rates from East to West Germany. The fact that women are more likely than men to leave East for West Germany (cf Dienel and Gerloff 2003; Gerloff 2004; for the higher migration rates among women, see also Hunt 2006; Windzio 2007) is particularly curious because their higher migration rates are not consistent with the predictions of most theoretical frameworks. Moreover, a comparison of the wage differences between East and West Germany shows that the wage gap between East and West Germany is smaller for women than for men, which suggests that men should be more likely to migrate. This topic deserves attention because it reinforces problematic demographic developments as aging of the population. Moreover, together with other trends as lower return migration to East Germany by women or higher migration of women abroad it leads to severe demographic consequences. For example, in 2007, in nearly all regions of East Germany, there were fewer than 90 women in the age group between 21 and 25 for every 100 men. This ratio was even worse for slightly older women (see Figure 1 and cf Grünheid 2009; Kröhnert and Klingholz 2007). In the age group between 26 and 30, fewer than 86 women lived in East Germany for every 100 men. Gender and migration I investigate which factors at the micro, macro and meso levels may influence migration decisions and outcomes in a gender-specific manner. The main challenge of analyzing the causes and consequences of women’s migration is not only that migration can be influenced in a gender-specific manner at the micro and macro levels but also that women’s social networks may determine different migration decisions and outcomes for single women compared with women in partnerships. I expect that gender alone does not account for the differences between migrant men and women but that gender and partnerships interact and play a mediating role in migration. The interrelation of these two dimensions results in not only different migration patterns between men and women but also entirely different migration outcomes for single women and women in partnerships. I focus on individual migration in the first two articles and on family migration in the three last articles. The manner in which gender influences migration creates a highly complex research scenario, placing high demands on the data and empirical methods used. IAB-Bibliothek 358 16 Introduction Figure 1: Ratio of women living in Germany per 100 men; the first age group is 16 to 20 years; the second 21 to 25 and the third age group is 26 to 30 Source: Statistical Federal Office Germany; data 2007; own calculations. 17Chapter 1 Contribution To account for the topic’s complexity, I investigate migration from different angles (causes and consequences of migration) and examine different groups (singles and couples) while maintaining a focus on the influence of central factors such as education and qualifications. Finally, I also compare the mobility patterns of persons who relocate under different economic conditions, more precisely, people relocating from East to West Germany and within West Germany. Gender and migration theories This study adds to the limited theoretical discussion of women’s migration by providing an overview of theoretical frameworks. The fact that women’s migration can only be understood when investigated from different angles is reflected in the choices of the migration theories reviewed. First, I present an overview of the classical migration theories that are used to investigate individual migration and can also be used as a theoretical framework for the special case of East-West migration in Germany. I also provide a summary of the migration theories used to investigate family relocation. Although the role that gender plays in the decision to migrate is central to family relocation, that role is mostly neglected in migration theories that focus on individuals. Finally, I discuss briefly two theoretical ideas that are specially designed to describe the differences in migration patterns of men and women who relocate individually. Data and methods For the empirical investigation of the causes and consequences of migration, I use the German Socio-Economic Panel (SOEP) data that were collected by the German Institute of Economic Research (DIW) beginning in 1990. The existing research on the specific context of East-West migration in Germany was primarily conducted shortly after reunification, with the longest period investigated at the individual level spanning into the year 2001. Meanwhile, the available data since reunification cover more than 20 years and allow for more complex research questions, such as analyses of gender-specific influence of education on the decision to migrate and the application of more sophisticated research methods such as multilevel models with random (see Rabe-Hesketh and Skrodal 2012a; Rabe-Hesketh and Skrodal 2012b) or fixed effects (see Engel 1998; Snijders and Bosker 1999) or Heckman selections (see Heckman 1979; Heckman and Smith 1996). IAB-Bibliothek 358 18 Introduction 3 Historical context East-West migration in Germany 3.1 Differences and similarities Over 20 years ago, East and West Germany were combined to form a single country with the same institutional framework. Reunification equalized the political system in East and West Germany and other aspects of people’s lives such as pensions, education, and systems of taxation. However, East and West Germany remain divided by their historical past, their economic situation and certain cultural aspects. The Federal Republic of Germany (FRG) and the German Democratic Republic (GDR) have experienced significantly different historical pasts: the FRG was a western liberal democracy, and the GDR was a socialist Warsaw Pact country. Moreover, the economic systems in the FRG and the GDR also differed significantly, and East Germany’s economy continues to lag far behind that of West Germany in terms of development. In 1991, the gross domestic product (GDP) of the new federal states (excluding Berlin) accounted for only seven percent of the GDP of the united Germany (with East and West Berlin, eleven percent), even though the new federal states held approximately one-third of the territory and approximately one-fourth of the population. Until 2009, this GDP percentage rose by only five percentage points to a level of twelve percent (with East and West Berlin included, fifteen percent) (calculations on data of the Statistical Agency of the Federal Union and the Länder). In addition, income levels vary between East and West Germany. In 2009, the hourly wage for West German men was 20.1 Euros, whereas the hourly wages for East German men was 30.3 percent lower at 14 Euros per hour. The difference between East and West German women’s incomes was only 12.9 percent; the East German women’s incomes were 13.2 Euros per hour, and the incomes of West German women were 15.2 Euros per hour (Statistisches Bundesamt 2010, p. 49). Notably, the gender wage gap is higher in West Germany than in East Germany (in the East, the incomes of women are only 80 cents, or 5.7 percent, lower). In West Germany, the difference is 24.7 percent or 5 Euros (Statistisches Bundesamt 2010, p. 49). Moreover, the unemployment rates remain nearly twice as high in the East as in the West (Bundesagentur für Arbeit Statistik 2013). The political ideology in the GDR emphasized gender equality and the labor market participation of women. In the FRG, the main breadwinner family model was common (Pfau-Effinger 1996). In the GDR, the employment of both partners was most common (Lauterbach 1994) and men and women led more similar lives in the GDR (Blossfeld et al. 1995). Currently, women in East Germany participate in the labor market more often than their West German counterparts and contribute 19Chapter 1 Historical context East-West migration in Germany a greater amount to the household income. The percentage of household income contributed by women, according to various sources, is between 40 (Lemke 2002) and 43 percent (Dölling 2002). The incomes of West German women account for only 18 percent of the entire household income (Dölling 2002; Lemke 2002). Until today, East German men and women express more egalitarian gender views than West Germans (Kreyenfeld and Geisler 2006; Lück and Hofäcker 2003; Matysiak and Steinmetz 2008). Finally, the structure of families in East and West Germany remains significantly different. In 2009, in East Germany, 40.2 percent of adults were married compared with 46.3 percent of the West German population. Moreover, in East Germany, 8 percent of couples cohabited compared with 6.2 percent in West Germany during the same year. Thus a higher share of the East Germany population lived alone. Although the current breadwinner model, encompassing traditional gender roles, is supported by the German tax system (Dingeldey 2000, p. 125), alternative living arrangements are more common in East than in West Germany. It is also notable that the differences in East and West Germany were greater in 2009 than they were in 1996, which indicates that at least in this domain, there is little harmonization. 3.2 East-West migration in Germany The East-West migration in Germany is indivisible from German reunification and the historical changes that occurred in 1989. The most prominent event was the fall of the Berlin Wall, which was visible evidence of the breakdown of the GDR and the entire Eastern Bloc as well as the end of the Cold War (Zelikow and Rice 1996). The fall of the Berlin Wall on November 9, 1989 enabled the East German population to freely and legally relocate to the FRG for the first time in 28 years. For the GDR, the fall of the Berlin Wall was only one step in the direction of reunification; that event was followed by the first free election in March 1990 and then the Unification Treaty, which accomplished the consolidation of the monetary, economic and social union of Germany in July 1990 (Bahrmann and Links 1999; Pond 1993). The final step occurred in October 1990, when the GDR and the FRG became politically unified. For the people of East Germany, this process opened the possibility of migrating to the West and removed the restrictions on where they could live. Unsurprisingly, East Germany experienced high outmigration, with 400 thousand people migrating to West Germany in 1990 alone (Figure 2). However, the connection between migration and German reunification goes deeper. Apart from the poor economic situation in the GDR, which went hand-in-hand with the economic crisis in the IAB-Bibliothek 358 20 Introduction Soviet Union, the mass protests and outmigration that occurred shortly before the fall also contributed to the breakdown of the GDR. Migration to West Germany became feasible. The Hungarian government opened its borders to Austria in May 1989, disregarding the conventions of the Warschauer Pakt (Bahrmann and Links 1999; Pond 1993). By 1989, 390 thousand people had migrated to the West (Figure 2). The high migration rates in 1989 and 1990 matched the migration outflow of 1957, the highest figures of outmigration before the Berlin Wall was built. In the first two years after the fall, more men migrated to West Germany than women Figure 2 and 3: East-West migration in Germany 250 200 150 100 50 0 250 200 150 100 50 0 Source: Statistisches Bundesamt (2011) Bevölkerung und Erwerbstätigkeit. Wiesbaden: Statistisches Bundesamt, p. 53 ff. 1957 1957 1961 1961 1965 1965 1969 1969 1973 1973 1977 1977 1981 1981 1985 1985 1989 1989 1993 1993 1997 1997 2001 2001 2005 2005 2009 2009 net migration men net migration women East-West men East-West women West-East men West-East women Net Migration between East and West Germany (in k) Migration between East and West Germany (in k) 21Chapter 1 The structure of this work (Figure 3). This picture changed after 1991, when more women migrated than men (cf also Grünheid 2009; Mai 2006; Schlömer 2004). However, beginning in 1991, the migration flows for both men and women declined steadily. In 1993, the flow of those leaving East Germany stabilized at approximately 80 thousand men and women per year. Beginning in 2003, the level dropped again by ten thousand to approximately 70 thousand. This indicates that even if the migration flows declined shortly after reunification, the eastern portion of Germany has continued to lose approximately 25 thousand men and slightly more women every year since reunification. Overall, by 1995, the East German population had declined by 7.9 percent compared with its pre-reunification levels, and by 2000, it had declined by 10.7 percent.4 By 2008, the East German population had declined by 11.7 percent, or 1.7 million people. However, other factors also contributed to this decrease in the population, including declines in fertility and migration abroad (Statistisches Bundesamt 2010, p. 10). The German reunification allows for the investigation of unique research questions possible and should also enable this study to improve our understanding of the East-West migration in Germany and international migration. For example, the identical political and institutional framework in combination with extensive economic differences renders comparisons of the migration patterns between East and West Germany particularly useful, primarily because reasons for migration that are connected to political or educational systems common in international migration can automatically be eliminated. 4 The structure of this work This study begins with an overview of the theoretical frameworks that exist in the migration literature. In addition to the discussion of the most common individuallevel migration theories, this overview reviews theories that describe family migration and outlines the few theoretical ideas that focus on the differences in migration behavior between single men and women. The theories are presented as theoretical concepts that can be utilized simultaneously rather than as competing frameworks (cf Massey et al. 1993). The second section comprises the empirical contribution of this work, which is structured into five chapters, each containing one article. All of the articles focus on the causes or the consequences of East-West migration for men and women, although the articles have various emphases. 4 For the population of the GDR in 1989, please refer to the Staatliche Zentralverwaltung für Statistik (1989, p. 335). The figures presented do not include Berlin because the statistics office does not differentiate between East and West Berlin after 2000. IAB-Bibliothek 358 22 Introduction The first article focuses on the causes of individual migration from East to West Germany for both men and women. The purpose is to investigate the genderspecific differences in the migration process. This article focuses on the influence of education on migration and the self-selection processes involved in migration. This article is motivated by findings in recent international research that indicate that level of education influences the migration decisions of men and women differently (Dienel and Gerloff 2003; Dumount et al. 2007; Feliciano 2008; Gerloff 2004; Stecklov et al. 2010). This article accounts for the gender-specific interplay of factors at the micro and macro levels. For example, it can be shown that education determines women’s migration decisions to a greater degree than it determines men’s migration decisions. Women are self-selected with regard to their education. Only women with at least an upper secondary education are able to profit from migration and are thus willing to relocate. Because the economic conditions are more favorable for men, men are able to profit from migration even if they have less education. Thus, self-selection based on education is lower for men who migrate from East to West Germany. The second article shifts the research perspective and focuses on the consequences of migration for men and women from East to West Germany. This section focuses on the effects of migration on non-monetary factors such as subjective well-being (SWB). This article adds to the understanding the development of SWB through the process of migration. This topic has not yet been investigated using longitudinal data and information from before and after relocation. The results indicate that migrants are indeed able to improve their SWB relocating. For men, this improvement can be associated with better economic conditions in West Germany. The third article goes beyond the investigation of the determinants of East- West migration in Germany and investigates a broader spectrum of mobility decisions of men and women. This article accounts for the complexity of mobility decisions and that commuting between East and West Germany and migration may be interrelated. Commuting may serve as an alternative or stepping stone to migration. The aim of this article is to show how the mobility choices of men and women differ, which factors have gender-specific influence on the decision to commute and to migrate and, finally, how these processes are interrelated. A question motivating this article is whether the different choices in mobility forms men and women prefer may help explain the higher migration rates found in East- West migration for women than for men. To seriously consider the effect of marital status on the migration of men and women, researchers must not only control for the family context but also investigate the forces that drive family migration decisions as both partners’ characteristics. The fourth article presented here investigates the migration patterns of men and 23Chapter 1 The structure of this work women who relocate with their partners, with the migration of persons who relocate alone and, thus, do not need to consider the location preferences of partners. This article adds to the literature on East-West migration in Germany by investigating the meso level and analyzing how a partner’s characteristics influence men’s and women’s migration decisions. This article’s contribution to international research is an analysis of the migration of people who have been socialized in a post-socialist country. Information on the interplay of marital status, family groupings and gender norms in determining the migration behavior of men and women and comparing individual and family migration is provided. The importance of being in a partnership for migration decisions is emphasized. Women are much more restricted in their mobility decisions due to partnership than are men. Moreover, whereas men abstain from migration because their partner’s characteristics, such as high income, restrict migration, women in partnerships retreat from migration simply because of the presence of a partner, regardless of that partner’s characteristics. The last article presented here contributes to the topic of family migration by investigating both the determinants and consequences of family migration. Although family migration is a joint decision, migration may have different effects on the income and employment of men and women in partnerships. This article focuses on these differences. In addition, this article takes advantage of the fact that the situation in the labor market differs significantly in East and West Germany even though the institutional settings are identical. Merging the analyses on the causes and consequences of migration, this article combines the inconsistent results obtained at individual and family levels. Moreover, this article examines why East German women in partnerships, who appear to be more egalitarian and make greater contributions to household income, are unable to benefit from their higher education when relocating. The work concludes by combining the results of the different articles and providing a more general overview and description of the East-West migration of women in Germany. The results of this study may prove interesting for future research and policy makers. Politicians may be interested in the mechanisms that drive migration from East to West Germany. Moreover, information gained from research on East-West migration may enhance our understanding of other migration flows such as the relocation of people to Germany following the enlargement of the EU because data on such recent developments remain limited. Most importantly, the investigation of women’s migration should add to our understanding of the general phenomenon of migration, not only because gender is often neglected but also because gender may be a dimension that causes inequality (cf Portes 1997, p. 816). 31Chapter 2 Selection of migration theories 1 Introduction No migration theory exists that can be used generally or that provides a theoretical subordinate framework for all the different aspects of migration (Arango 2000; Bakewell 2010; Castles 2010; Molho 1986). As Massey et al. (1993, p. 432) indicate, “[a]t present, there is no single coherent theory of international migration, only a fragmented set of theories that have developed largely in isolation from one another, sometimes but not always segmented by disciplinary boundaries”. 1 Migration research is characterized by a variety of theories focusing on different aggregation levels (i.e., micro vs. macro levels) and describing the migration of different units (i.e., families vs. single individuals) providing various explanations for the relocations. 2 Selection of migration theories The selection of theories presented here was driven by consideration of suitability to investigate migration from East to West Germany focusing on gender-specific differences. Because of some very fundamental characteristics of Germany and because East-West migration takes place within one industrialized country, some migration theories, such as the institutional theory of migration, are a priori not applicable. Other theoretical frameworks, such as network theory (see, e.g., Haug 2000a; Haug 2008; Massey 1990; Massey et al. 1987; Massey et al. 1993; Massey and Gracía España 1987) are not discussed, as the available data does not allow to investigate the theoretical ideas. One of the selection rules applied to the theoretical overview is that the theoretical frameworks presented here should place individual action in their center, should provide specific selection rules for the migration, and should inform us about the reasons why individuals choose migration. Therefore, all chosen theoretical frameworks should provide us with an explanation of why individuals or families with certain characteristics migrate and why others do not. In this respect, these theories differ from macro-level theories that address aggregated movements and the influence of structural conditions such as unemployment, which simultaneously abstracts from the description of the mechanisms at the individual level (Haug 2000a, p. 32). Therefore, none of the common macro-level theories are presented here. At the same time, those selection rules may even be problematic for some of the theories presented 1 For an overview of migration theories, see, for example, Faist (2000), Gross and Lindquist (1995), Haug (2000b), Kalter (1997), Massey et al. (1993), Massey et al. (1998) and Molho (1986). IAB-Bibliothek 358 32 Overview of migration theories here. Both signaling and segmentation theory place the employer rather than the migrant at the center of their consideration, while the individual is only able to react, making more or less rational decision based on her understanding of the selection processes employers impose. Moreover, the gender role theory, as presented among the theories of family migration, is problematic, when the selection criteria are taken seriously. The gender role theory is based less on utility maximization than on the assumption that people act according to their socialization. Finally, the presented theories are not limited to one discipline. Economic and sociological theories are described. As Cooke (2008, p. 258) noted when referring to the work of Halfacree (1995) concerning theories on family migration, disciplines develop theories and, thus, assumptions that are not challenged but rather taken for granted. This reduces not only the comparativeness of migration research across disciplinary boundaries but also the scientific gain from the empirical work based on those theories. Using a broader selection of theories, regardless of their originating discipline, I try to avoid any unnecessary limitation and to view the phenomenon of migration from various angles. Flexible use of theories As Massey et al. (1993) and others (e.g., Faist 2000; Kurekova 2009; Portes 1997) suggest, it is important to understand these theories not as competing ideas but as complementary frameworks that can be mixed together. This understanding has two implications for the development of a theoretical framework aimed at explaining a specific empirical situation. First, ideas can be summed and borrowed (Boswell 2008; Boswell and Mueser 2008). The second possible implication is that varying theoretical approaches can be applied simultaneously. To account for the specific situation an individual faces, it might be reasonable to adapt the theoretical framework to a person’s living situation using different theoretical implications for singles and for married individuals. Overall, the more flexible use of multiple theories should allow not only for a better understanding of the development of migration and the explanation of persistent and new migration trends, but it also lowers the danger of pressing the empirical findings into an unfit theoretical construct. In the following discussion theories referring to individual and family migration, which might provide a theoretical framework for the investigation of the gender-specific East-West migration are selected. 33Chapter 2 Selection of migration theories 2.1 Individual theories The first set of theories presented below provides explanations for individual migration. In the following, neoclassical migration theory, new economics of migration, signaling, and segmentation theory are discussed. The theories discussed in this section make no special assumptions about how the migration behavior of women differs from the migration patterns of men, and they assume that the processes are gender-neutral and that men and women behave alike. Regarding the research question this assumption is clearly a disadvantage. However, the theories bear the potential to investigate the differences of migration patterns using assumptions on differences in the characteristics of men and women and to derive in this way gender-specific hypotheses. Moreover, in Section 2.2 two theoretical ideas that try to explain individual migration of women are sketched briefly. 2.1.1 Neoclassical migration theory Building on the connection between wage differentials and migration, which was already central in the neoclassical macroeconomics literature (Hicks 1932; Lewis 1954), neoclassical migration theory provides the first theoretical framework focusing on individual migration within a microeconomic model. Sjaastad treats migration as an “investment increasing the productivity of human resources” (Sjaastad 1962, p. 83), which is analogous to investments in schooling or on-the-job training. Similar to investments in education, an investment in migration increases a person’s future rates of return in the form of higher earnings. A precondition for benefiting from migration, however, is that regional or international income differences must exist. The greater these differences are, the more individuals can profit from relocation. The decision to migrate is based on a comparison of the benefits and costs of migration. Individuals compare both the monetary and non-monetary costs and benefits (Sjaastad 1962; Speare 1971). As Sjaastad explains, “The private costs can be broken down into money and non-money costs. The former include the out-of-pocket expenses of movement, while the latter include forgone earnings and the “psychic” costs of changing one’s environment” (Sjaastad 1962, p. 83). The main contribution of Sjaastad (1962) model is the idea that individual characteristics influence the benefits and costs of migration. Characteristics such as age, education and firmspecific experience might incentivize certain groups to migrate, while creating barriers for other groups. To define this idea, Sjaastad (1962) examined the effect of age on migration behavior. Because younger people face longer employment periods at their destination, they experience greater long-term profits from IAB-Bibliothek 358 34 Overview of migration theories the investment in migration. Thus, younger people should be more willing to relocate. Additionally, the fact that younger individuals usually have more general human capital in relation to firm-specific human capital allows them to earn higher returns from migration. Firm-specific knowledge is obtained from the company or the job in which a person is employed and is strongly associated with seniority; therefore, it is not easily transferred, and it loses its value if the person changes employers. In contrast, people usually acquire general knowledge in educational institutions. The universal nature of this knowledge enables them to switch employers more easily, as their productivity should be comparable among different companies. Because younger individuals usually have invested less in firm-specific capital than older individuals, a young person’s ratio of general-to- specific knowledge favors migration and binds them less to the current location than an older person’s knowledge ratio.2 Neoclassical migration theory provides an intuitive and appealing explanation of the migration process (Arango 2000; Bielby and Bielby 1992). Its credibility is bolstered by its universality. In terms of Colman’s (1990) boat or bathtub concept, this theoretical approach incorporates not only the individual-level conditions for migration but also accounts for the fact that individuals react to macro-level conditions such as changes in income levels. The advantages of this theory may explain why scholars have extended it in several ways. Two extensions of the model were proposed by Chiswick (1999), who concentrated on the costs of migration to explain the selectivity of migrants more accurately. The first extension is related to the concept of efficiency. Highly educated migrants should be more capable of finding the best location to migrate to and to coping more quickly with the new situation, e.g., by finding a new job or a new accommodation. They may use better search strategies or be more familiar with modern search tools such as the internet. Their abilities, which usually help them be more productive, also make them more efficient in migration (Chiswick 1999). Highly educated individuals might also have better language skills. Because ‘time is money‘, the higher efficiency of highly educated individuals should result in lower absolute migratory costs, which implies that more highly educated individuals will benefit from migration in situations in which less-educated individuals can no longer profit. The second extension is related to the idea that for highly educated migrants the relative migration costs are lower. More highly educated individuals face similar (or lower) absolute costs but lower relative costs than less-educated individuals because of the higher incomes the highly educated 2 There are several other reasons individuals with varying human capital or occupations are more or less likely to migrate. For example, individuals with higher general knowledge usually have lower non-monetary costs of migration because of, for example, their broader friendship networks (Brücker and Trübswetter 2007, p. 374). 35Chapter 2 Selection of migration theories already earn more in the country of origin, which allows them to spend a smaller fraction of their income on the relocation. The larger the monetary costs of migration, the greater the differences between the migration costs of highly and less-educated individuals. As a result, the selectivity of migrants is higher when the costs of migration are higher. Because the monetary costs of migration are relatively small in the case of regional mobility, this extension of the human capital model should be less important to regional migration. However, this concept should confer higher explanatory power for international migration, where the monetary costs of migration differ considerably among countries. Borjas (1987) extends the classic human capital approach based on the Roy (1951)3 model. Borjas (1987) provided a model that accounts for the selectivity of migrants by considering specific income structures in the labor markets of the sending and receiving countries. Because of the variance in the income distributions, individuals with different educations might gain varying returns from migration even if relocating from the same sending country to the same receiving country. If the income gap between two countries or regions is large enough, then every group in the population should profit from migration regardless of the income distributions in the sending and receiving regions, and the selection of migrants remains low. However, the smaller the income gap becomes, the more dependent the individual gains are on the differences in the income distributions of both regions and the more important migrants’ skills become. Borjas’ (1987) model assumes that education is positively and similarly connected to the earnings in all countries. However, in a country with a broader income distribution, the value of an additional year spent in education is higher, and thus, the difference between the income received, for example, with upper and lower secondary degrees, is greater. When differentiating between two groups, it can be shown that highly educated individuals prefer regions with broader income distributions as they face a comparative advantage in productivity in such regions. Thus, it can be expected that if earnings in the destination are less equal people with higher educational levels will migrate. 3 Roy (1951) illustrates this point by using two occupational groups: hunters and fishers. His model shows that if the productivity levels and the incomes of two occupations differ with regard to ability, then the more-able individuals will make occupational choices different from those who are less able. The model incorporates two occupations that differ in the scope of their outcome distributions and incomes to show that more-capable individuals choose those distributions that are more unequal and therefore broader. Less-able individuals working in an occupation with a broader income distribution may suffer extremely low incomes; concurrently, individuals who are more able and work in the same occupation may earn exceptionally high incomes. Because occupations with broader income distributions are more dependent on ability, capable individuals can obtain incomes that are unattainable in occupations with narrow income distributions and are therefore attracted to those occupations. In contrast, less-able individuals are more likely to choose occupations with narrow productivity and income distributions. Individuals cannot obtain high incomes within these occupations, but they are also protected from very low outcomes because these distributions bear less risk for less-educated individuals. IAB-Bibliothek 358 36 Overview of migration theories Less educated migrants, in turn, favor destinations with narrow income structures. In regions or countries with more equal incomes, the penalty for the lack of education is lower because the divergence in living standards of highly and poorly educated is smaller. Therefore, if earnings in the destination country are more equal than in the country of origin, less-educated people will migrate. However, a broader income distribution in the sending country than in the receiving one does not necessarily indicate that migrants are negatively selected with regard to their skills. Just because the income distribution is unfavorable it does not mean that highly educated migrants lose all the other advantages of migrating, such as the higher generality of their human capital, their greater search efficiency or the relative advantages relating to migration costs, as neoclassical economic theory implies. More likely, the selection is less positive than it would be if the incomes were distributed differently. Using information on the income distribution of the sending and receiving countries, Borjas’ (1987) extended human capital approach can explain why the characteristics of individuals migrating between two countries with similar income gaps but different income distributions might vary considerably. This concept also suggests that the income allocation in the receiving country might not always be favorable for highly educated individuals and that, in such a case, individuals migrating might be in the group of middle- or less-qualified persons. This is the major difference between the ideas expressed in the simpler neoclassical model described by Sjaastad (1962), which always expects highly educated individuals to be more migratory in nature, and in Borjas’ model. This theoretical framework provides a more flexible mechanism for analyzing migration than Sjaastad’s classical model. However, despite all the positive characteristics provided by Borjas (1987), his model extension can also be criticized. Because the Roy (1951) model, which serves as the basic framework, does not include any costs, Borjas’ (1987) migration model treats the costs of migration in a rudimentary fashion. Borjas only defines migration costs as a stable fraction of the incomes of the migrants’ destination country, which is the same for all migrating individuals regardless of their abilities or educational levels. However, this condition is not very realistic (Chiswick 1999). Brücker and Defoort’s (2009) contribution to this problem proposes an additional extension of the human capital model based on Roy (1951). The authors suggest, as in earlier work, that costs exhibit educationspecific variations (Brücker and Defoort 2009; Brücker and Trübswetter 2007). Their approach is similar to the ideas expressed by Chiswick (1999) and goes back to the idea that education and migration costs are connected. However, the main difference is that Brücker and Defoort integrate this idea into a more demanding model based on Roy’s (1951) and Borjas’ (1987) consideration of extended human 37Chapter 2 Selection of migration theories capital. Their model accounts for the fact that the costs of migration should decline with higher education. This idea is exactly the opposite of the way Borjas (1987) treated migration costs. Brücker and Defoort (2009) propose that one reason for the declining migration costs is the broader regionally spread networks that highly educated individuals usually have. Chiswick’s (1999) argument regarding the efficiency of highly educated individuals can be used here as well. Brücker and Defoort’s (2009) extension implies that highly educated migrants might not only receive higher gains from migration to regions with advantageous income distributions but may also have an additional advantage as they spend a lower proportion of their incomes on the relocation. Despite the various model extensions, which can be partly criticized by itself, some general critiques remain. The most important, perhaps, is that within the neoclassical economic framework, migration is motivated by economic considerations only (Castles 2010, p. 1573; Faist 2000, p. 36; Massey et al. 1998, p. 8). However, migration also occurs because of non-economic reasons, such as political or cultural. The focus on monetary costs and gains and the subsequent failure to include nonmonetary factors constrains the adaptation of additional ideas and the development of new hypotheses (Kalter 1997, p. 51). Moreover, as was demonstrated by Massey et al. (1998) based on an overview of the empirical literature, the neoclassical theory assumes a linear connection between migration and regional wage differences (see also Kurekova 2009). It ignores market imperfections and the fact that people are usually unable to make their migration decision based on perfect information (Castles 2010, p. 1573; Kalter 1997, p. 51; Kurekova 2009; Molho 1986, p. 397). Concentrating on economic behavior, the neoclassical migration theory reduces migrants to workers (Arango 2000; Schwenken and Eberhardt 2008). Moreover, it assumes that migrants make their decisions as autonomous individuals, and it does not account for the influence of family members, friends and social networks (de Haas 2010; Massey 1990; Massey et al. 1987; Mincer 1978; Radu 2008; Stark 1991; Stark and Bloom 1985; Stark and Levhari 1982). Despite the efforts made by Sjaastad (1962), this theoretical framework is unable to explain why so many people do not migrate even in the face of significant wage discrepancies between two countries (Arango 2000; Kalter 1997, p. 51; Kurekova 2009). Finally, the theoretical framework is unable to account for the social and historical changes influencing migration (for a discussion of this critique point see: Massey 1990). 2.1.2 New economics of migration Together with several coauthors, Stark proposed a theoretical framework, which challenges the ways of thinking in traditional neoclassical economics IAB-Bibliothek 358 38 Overview of migration theories and its migration models (Arango 2000; de Haas 2010; Kurekova 2009; see also Stark and Bloom 1985). The most important feature of Stark’s framework (Stark 1991; Stark 2006; Stark and Bloom 1985; Stark and Levhari 1982; Stark et al. 2009; Stark and Taylor 1989; Stark and Taylor 1991; Stark et al. 1986; Taylor 1999) is that migration is no longer the decision of an autonomous individual, a point that was criticized with regard to neoclassical economic theory (Arango 2000, p. 288; Kurekova 2009, p. 4; Massey et al. 1993, p. 436), but is rather the joint decision of a household. Even if only a single person actually migrates, families decide together on the relocation of one or more household members. The family chooses the member best suited for the migration. Stark and his coauthors do not provide a selection rule or describe the characteristics a person might have to be chosen from the family to migrate. However, it can be assumed that the ability to find work and the expected incomes are important to the decision. From a sociological point of view, cultural norms might also influence which of the household members must migrate. A possible connection with regard to gender is that different familiar expectations for men and women might determinate migration behavior (Kanaiaupuni 2000; Kandel and Massey 2002). Contrary to neoclassical economics, in the new economics of migration, the income gains at the individual level do not influence migration decisions but the gains of the entire family do. This feature of the migration decision is included in a concrete model by incorporating the utility of the household members who stayed behind into the utility function of the migrant (Stark 1995). Thus, the migrant’s well-being depends on the household members’ utility; simultaneously, the household members’ well-being depends on the utility of the migrant. This connection is possible to determine, as this approach focuses not on income maximization but on risk minimization (Stark and Levhari 1982). Families migrate and diversify the sources of income, thus minimizing their risks. With this aspect, Stark follows the ideas expressed by Hicks (1967) regarding risk minimization through spreading (see Stark 1991, p. 41), particularly when some of the family members are employed in geographically removed regions. This ensures, that if the family’s total income is derived from more than one source then the loss of one of the income sources is less traumatic. What this means in technical terms has mainly been demonstrated by Stark’s (1995) later work on altruism. In Stark’s model on altruism, a family should be able to achieve gains from scale economics if its income sources are negatively correlated and if the likelihood that the family loses all its income sources at the same time is nearly zero. Therefore, the theoretical framework of risk minimization can also explain why migration occurs between countries with no significant income differences. 39Chapter 2 Selection of migration theories This feature of the model contradicts the assumptions from the neoclassical models, i.e., that income is a homogenous good, indicating that the source of the income matters (Massey et al. 1993, p. 438). Moreover, this feature of the model integrates the theme of remittances into an economic framework of migration (Taylor 1999). Because the family members profit from each other’s well-being, remittances increase the well-being of the sending family member and of the receiving members (Stark 1991; Stark et al. 1986; Taylor 1999). Stark’s approach is especially fruitful for the investigation of return migration, as the time perspective used is rather myopic. Persons migrate for limited periods. The family’s principal source of income and main state of residence remains in the region of origin. After the end of a successful migration project (e.g., the migrant earns enough money to purchase a desired good), the migrant returns home. Thus, this concept implicitly accounts for return, repeated and circular migration. Despite its substantial differences, the new economics of migration and the neoclassical economic theory can be criticized for a similar reason. Stark’s model also treats migration as being purely motivated by economic considerations (Faist 2000, p. 41). Even if persons migrate to minimize their risks, the ultimate aim is of an economic nature and the decision is based on a rational decision (Arango 2000). Similar to neoclassical theory, Stark’s model does not consider family migration, i.e., migrations in which one family member follows her partner, or other non-economic migration flows (Arango 2000). Faist (2000, p. 41) indicates two additional shortcomings of the new economics of migration approach. Even this theoretical approach assumes that migration is a household strategy. The approach ignores conflicts and power relations within the household and is unable to note who made the decision and how it was legitimized. Gross and Lindquist (1995, p. 327), citing Folbre (1984, p. 5), note that “the household [is used] as an individual by another name” (see also Bakewell 2010, p. 1693). Another, problem of the concept is that even though migration decisions are made to account for future developments, the past is ignored. However, it can be expected that experiences in the past make people believe they need a type of insurance, which was cited as the reason families differentiate income sources (Faist 2000, p. 41). In a second approach, Stark and coauthors (Stark 1991; Stark and Taylor 1989; Stark and Yitzhaki 1988) contributed to the sociological idea of social comparison, which was described in 1966 by Runciman (Stark and Taylor 1989). Runciman (1966) suggested that individuals compare themselves with others or with a picture of themselves years into the future and that they might feel relatively deprived if they want something others have or if they want to have something in the future they do not (yet) have. Stark and Taylor (1989) replace Runciman’s (1966) somewhat vague concept by beginning their analysis with a IAB-Bibliothek 358 46 Overview of migration theories market to find a job in the internal labor market. Migrants, people with other ethnic backgrounds and women especially face greater disadvantages and might have problems accessing the internal labor market, despite adequate education (Gordon 1995; Hudson 2007).7 For Germany, Sengenberger (1987) distinguishes a third fraction of the market, the craft-specific labor market, which accounts for the high share of individuals with specific occupational qualifications who work in small- and middle-sized companies.8 This partial market shares some of the characteristics of the internal market and, simultaneously, some of the characteristics of the external market, and they can be set between the external and internal labor markets (Blossfeld and Mayer 1988). Gordon (1995), referring to his own work (see Gordon 1994) and that of McNabb and Ryan (1990), noted that the presented ideas of internal and external markets can also be understood as two extreme positions of a continuum. As with the internal market, the income structure of the craft-specific market is well defined, where individuals face a predefined career path. Here also, the interdependency between employer and employee is high (Blossfeld and Mayer 1988). However, as previously mentioned, most of the firms in this part of the market are smaller, which has two implications for the market (Blossfeld and Mayer 1988). First, the companies are unable to absorb all economic shocks; therefore, as with external markets, employees are let go in times of economic hardship. Second, these companies cannot offer such well-defined career paths within the company. To climb the career ladder, individuals must change their employer. Correspondingly, the fluctuation between firms is higher in this sector than in the internal labor market, but it is still lower than in the external market. Similar to the internal market, qualifications are the key to the craft-specific market. These qualifications are acquired during vocational training, which is highly differentiated and occupationally orientated in Germany. Craft guilds and legal restraints guarantee highly standardized vocational training and highly generalizable apprenticeships. While the differences between the craft-specific and external labor markets are obvious, and the qualifications play a key role in the division, it is more difficult to clearly differentiate between the craft-specific and internal markets. In a sense, the migration decision faced by people relocating between countries is similar. Within a country, however, the legal framework ensures the general 7 Gordon (1995) indicates that the argument is sometimes made that women are less committed to their work. According to the author, even if this argument is not supported empirically, it has negative effects on the excess of women in the internal market as it works as a signal for potential employers. Moreover, women are crowded into specific jobs, with a continuous and more flexible supply of workforce than for men, thus reducing the value of their work and their earnings (Gordon 1995). 8 For a description of the development of the segmentation theory, see Köhler et al. (2007). 47Chapter 2 Selection of migration theories comparability of educational and vocational certificates. For individuals migrating abroad, the approval of the qualifications is not guaranteed, which is especially important for those who have higher skills. Thus, the segmentation theory would focus once more on the elements borrowed from the signaling theory and on discrimination. Katz and Stark (1987) indicate that it could be more difficult for an employer to evaluate the true potential of an employee with a foreign certificate. For the employee, this means she cannot be certain that her qualifications are indeed approved and that her position and her earnings are appropriate for her qualifications. Moreover, employers may be reluctant to offer immigrants jobs that require investments in human capital or on-the-job training, as they may fear the immigrants will only stay at the destination temporarily (Offe and Hinrichs 1977). Persons with good qualifications, such as those employed in internal and craft-specific markets in their home country, should be reluctant to relocate, as their qualifications are not necessarily approved abroad. Moreover, highly qualified individuals already have relatively good pay and stable jobs and, therefore, face higher incentives to relocate abroad than do individuals who are employed in the external labor market. Even if the wage gap is extensive, it should be less attractive for highly qualified individuals to give up their positions as qualified workers and to work in the external labor market. If the employees are unable to certify the individual’s qualifications, these individuals are only able to find employment in the external sector. Thus, the highest willingness to migrate abroad should be found for the individuals employed in the external labor market because other individuals would be more reluctant.9 Relying on segmentation theory, Gordon (1995) proposed a similar idea where he differentiates between speculative and contracted moves. Contracted moves allow workers to change jobs within a region or over regions with relatively low risk, as they move after receiving a job offer. Thus, such job changes are associated with the primary market. Speculative job changes, on the other hand, bear higher risks and are associated with the secondary market. Speculative job changes are less attractive, especially when entry jobs into the primary market are missing and the unemployment at the destination is high. Gordon (1995) differentiation 9 The idea that migrants usually find employment in the external labor market was also expressed by Piore (1979), who suggested a division of the labor market into two segments (see also Footnote 5). However, Piore’s (1979) argument follows a different reasoning. Because of the demand for work in the external market and because it is not possible to raise the incomes of the individuals at the bottom of the income hierarchy without increasing the incomes higher up, the demand for unskilled work in the industrialized countries can only be satisfied by hiring workers from abroad. Because migrants are more interested in the financial aspects of their work than in such factors as occupational hierarchies or prestige, they are willing to fill those positions. However, Piore’s (1979) concept is on the macro level and thus does not offer an explanation on the micro level as to why migrants decide to relocate. IAB-Bibliothek 358 48 Overview of migration theories between contracted and speculative moves also allows also for gender-specific hypotheses on the mobility behavior. First, while the relocation of the primary wage earner is usually a contracted move, the following partner has to find a job after the arrival and must change her job speculatively. The roles in the households, however, are not distributed randomly, and women are usually play the role of secondary earners.10 Thus, it can be expected that women who follow their partners more often experience speculative job changes. Moreover, Gordon (1995) notes that women usually work in less-specialized occupations, so there is less need to search for female workers over regionally. Segmentation theory provides a more sociological explanation of migration that emphasizes the differences between market segments. The idea that certain markets do not obey the economic forces of supply and demand but are ruled by social norms (Doeringer and Piore 1971; Piore 1970) provides a concept that is entirely different from the ideas discussed earlier. For the dual-labor market and segmentation theories, it is relatively difficult to determine who is employed in which market sector, a problem that induces a general inaccuracy in this concept (cf Arango 2000; Kurekova 2009). A clear theoretical and methodological distinction of the segments is missing (Gordon 1995; Leontaridi 1998). The problem of segmentation is intensified by the changes that occur in the labor market and the trend toward diversification within the segments (Cappelli 1995). 2.2 Migration of women The theoretical frameworks discussed thus far are gender-neutral and do not account for differences in the migration behaviors of men and women. Among the theoretical approaches that account for the differences in men’s and women’s migration patterns, there are two lines of argumentation that remain unconnected. First, several frameworks explain the gender-specific differences based on the migration behavior of married persons, arguing that constraints associated with the relocation of the entire household and the regional preferences of at least two household members must be considered. These constraints influence the migration behavior and the migration outcomes gender-specifically. These theoretical ideas concerning household migration will be discussed in section 2.2.2. Less common are approaches that explain the migration patterns of unattached women and how their migration behavior differs from those of men. These ideas will be sketched only briefly in the following section. The discussion starts with Markham and Pleck’s 10 The existing research provides several reasons for women’s positions as secondary earners. The explanations range from gender-norms over lower education and labor market participation of women to explanations that note the importance of the gender pay gap. 49Chapter 2 Selection of migration theories (1986) suggestions, which follow neoclassical economic theory and explain why women should relocate less often. Next, the analytical model of Thadani and Todaro (1979) is examined. This model is the oldest approach to explaining the migration of unattached women and was originally designed to explain the migration behavior of women in developing countries. The model suggests that women should be more likely to migrate than men. The last framework, developed by Edlund (2005), is mentioned only briefly, as the main conclusions are in line with Thadani’s and Todaros’ (1979) implications. 2.2.1 Migration of unattached women 2.2.1.1 Why do women migrate less often? In a series of papers, Markham and coauthors proposed various explanations for the differences in men’s and women’s migration patterns, drawing from neoclassical economic theory (Markham et al. 1983; Markham and Pleck 1986). Markham’s ideas cannot serve as an independent theory. Rather, they can be understood as an extension of the neoclassical economic framework. The explanations for the differences in the migration patterns of men and women rely on various auxiliary constructions and the general assumption that the migration behaviors of men and women are identical, as was present in the neoclassical framework. Both men and women act rationally, comparing the costs of migration and balancing them against the gains, but differences in their living situations result in varying strategies. Thus, the central idea is that periods of family formation reduce women’s labor market participation, their labor market outcomes and, therefore, their mobility patterns. This pattern is not only true for women who already have a family but also for young women who face future periods of family formation. Based on the very problematic concept of perfect information, the theory assumes that actors are able to see ahead and that young women take into account their desire to have children as they plan their careers and make migration decisions. Anticipating career interruptions and that the overall benefits from their career are lower, young women adjust their behavior and invest less in their employment, which reduces their labor market experience (cf also Anker 2001; Hakim 1991; Hakim 1995) and automatically reduces their mobility. The concept of Markham and his coauthors (1983; 1986) is based on a rigidly conservative gender role understanding (cf Schwenken and Eberhardt 2008). The assumption that women’s primary focus is on marriage and children, not on their labor market participation, is built on conservative gender stereotyping, as was Becker (1991) specialization idea. Markham’s argumentations, which may have been accurate in the 1980s when the theoretical concept was developed, must not IAB-Bibliothek 358 50 Overview of migration theories necessarily mirror modern societies’ and women’s thinking 30 years later. Using those concepts today seems dependent on overly conservative patterns. 2.2.1.2 Why do women migrate more often? As early as 1979, Thadani and Todaro (1979) noted a lack of theoretical or analytical frameworks able to explain the migration of unattached women. Although some theoretical considerations, such as the ideas expressed by Mincer (1978, discussed below), Sandell (1977) and Blood and Wolfe (1960), provide explanations for the differences in the migration patterns of married women and men, few theories and models explain why single women migrate and how their migration behavior differs from that of men. Based on observations from developing countries and emphasizing the fact that women in developing countries usually move to their husbands’ locations when marrying, Thadani and Todaro’s (1979) model provides explanations for the higher migration rates of women. In this model, as in neoclassical economic theory, migration takes place in response to economic considerations. Men and women try to improve their financial situation by migrating. The economic conditions of men improve because of occupational factors if they are, for example, able to find a better paying job. In line with the argument used in neoclassical economic theory, the main force driving male migration is higher remuneration in the country or region of destination resulting from the wage gap. This driving force also applies to women, who can improve their economic status through better employment opportunities offered at the destination. However, women may also improve their situation by marrying, which is the main difference between women and men. Although the authors make no clear statement on this point, the idea is based on the expectations that a woman’s employment is secondary and that the defining factors of the family’s status and economic situation are the husband’s occupational and social status. In an attempt to categorize the different types of female migrations, the authors differentiate between migrations entirely motivated by economic situations and migrations motivated by marital obligations. They also differentiate between married and single women. In this way, four categories of motives for migration emerge. The migration of married women might be motivated by economic considerations, such as the desire to earn higher wages by themselves, or married women may have non-economic reasons for migration and follow their partners who have found work at a new location. Single women, in turn, migrate because of better employment opportunities and, secondly, because they want to marry at the destination. Because a woman’s social status is dependent on her husband’s status, men with higher earnings and better social statuses are more 51Chapter 2 Selection of migration theories attractive to women than low-status men. Thus, the authors argue that the wage differential between two regions might increase the attractiveness of husbands in the region with higher earnings, which could motivate women to migrate. At first glance, it seems difficult to apply these ideas to a more economically developed society in which women are independent and marriage may play a subordinate role. However, women are still responsible for the upbringing of children in the modern societies of economically developed countries (Coltrane 2000; Hochschild 1989; Poortman and Van der Lippe 2009), and they interrupt their employment more often than men when their children are young. During such times, women are more dependent on their partners’ earnings and employment status (Edlund 2005). Moreover, the literature covering the marital behavior of men and women indicates the persistence of marital patterns in which women marry men of higher, or at least equal, status (Blossfeld 2009; Blossfeld and Timm 2003; Matthijs 1998; Schulz 2010; Skopek et al. 2011). If the availability of high-income men varies between regions, for example because more highly educated men live in cities (also true for economically better-developed countries or regions), cities might be more attractive for women (Cooke 2011; Edlund 2005; Gautier et al. 2010).11 As Edlund (2005) noted, the connection between economic and non-economic motives, and therefore of the migration categories, which are separate in Thadani and Todaro (1979) model, produce a double advantage for women with higher education levels in the case of migration and can help to adapt this framework to economically developed countries. First, it can be assumed that women, especially those who are highly educated, profit economically from migration to a region with better financial compensation. This assumption can also be derived from neoclassical economic theory. Second, women might profit from the higher availability of potential partners with higher earnings (Edlund 2005). Such an assumption goes beyond neoclassical economic theory. In their analytical model, Thadani and Todaro (1979) account for the homogeneity of marriages and develop three scenarios to explain why the availability of higher-status men in some regions might attract not only highly educated women but also those with lower education levels. Their model shows that the migration of less-educated women is especially likely when highly educated women are less likely to marry or if highly educated men do not differentiate between women with higher and lower education levels when looking for a partner. When women are more likely to find and marry a highstatus man in another region, and by doing so improve their own economic status, 11 Other approaches that highlight the marriage markets to explain the migration were provided by Gautier et al. (2010) and Behrman and Wolfe (1984). IAB-Bibliothek 358 52 Overview of migration theories migration might also become attractive for less-educated women, who would not otherwise profit from the wage differences between the regions. Bringing the non-economic and economic factors together, as in Edlund’s concept (2005), and relying on the assumption proposed by the general neoclassical economic theory that more-educated individuals are more likely to migrate, the following scenario emerges: more-educated women might profit from the economic conditions at the destination as well as from the higher earnings of their potential partners, which should encourage them to migrate. For lesseducated women, the incentives to migrate should be lower, as they profit only from the higher earnings of potential partners, whereas improvements in income alone should not be high enough to cover the migration costs. However, if the density of high-earning male partners differs significantly between two regions or, as proposed by Edlund (2005), between cities and rural areas, migration might also be attractive for less-educated women. The main and at the same time the most serious cirique of Thadani’s and Todaro’s (1979) is that it focuses strongly on marriages, but women’s migration is driven by many more factors than just interest in finding partners (Lim 1993, p. 228 refering to; Ware 1981). 2.2.2 Theories on family migration Scholars realized early on that analyzing the migration patterns of atomized actors could not explain all features of the different migration patterns of men and women, and they began to include the family perspective. The developed theoretical approaches depart from the earlier common assumption that women are passive actors in the migration process who migrate only if their partner has decided to do so (Lutz 2010, p. 1648; see also Lee 1966 who expressed this opinion). They consider the migration decisions of two actors simultaneously, which also represents the main challenge these theories have to overcome. In the first step, two economic approaches are outlined. These approaches describe how rational actors behave to maximize their utility with regard to migration based on two different maximization strategies. Similar to the migration decisions described in neoclassical economic theory, the first discussed framework, the household economy, conceptualizes migration as an investment in the productivity of individual human capital. Within this framework, however, the migration decision can actually lead to financial disadvantages for one of the partners (Kalter 1998; Mincer 1978; Ott 1992). This approach and the second approach based on bargaining theory assume that rational individuals and families migrate if they can obtain financial benefits. Household economy relies on a framework of utility maximization based on household income. Bargaining 53Chapter 2 Selection of migration theories theory uses a utility maximization approach based on individual calculations, which takes the view that individuals try to maximize their individual incomes through household migration. The third theoretical approach deviates entirely from this reasoning, as it is based on sociological concepts that suggest that the behavior of individuals is also influenced by learned cultural and normative patterns (Fenstermaker and West 2002; West and Zimmerman 1987; West and Zimmerman 2009). These patterns are different for men and women, who play different roles in the decision-making process regarding migration (Bielby and Bielby 1992). 2.2.2.1 Household economy In the 1970s, several economists attempted to incorporate a family perspective into the analysis of migration decisions (see: DaVanzo 1972; Mincer 1978; Polachek and Hovath 1977; Sandell 1977). The proposed models differ in their designs and are based on different ideas. Sandell (1977), for example, incorporates migration into a framework of work and leisure choices. Nevertheless, these models all highlight the influence of both partners’ characteristics, such as employment and earnings, on the family’s migration decision. Starting from the ideas of Sjaastad (1962) and Becker (1974), Mincer (1978) proposed a model in which the family maximizes its gains from the productivity of the human capital due to migration. The following section discusses Mincer’s (1978) model in greater detail. Among all the models developed during the early 1970s, this model gained the most attention in the migration literature (e.g. Compton and Pollak 2007; Cooke et al. 2009; Geist and McManus 2012; Nivalainen 2004; Shauman and Noonan 2007; Smits et al. 2003; Tenn 2010) because of its simple and persuasive structure (Bielby and Bielby 1992, p. 1242). Like atomized individuals, families migrate if the net household returns from migration are high enough and if the potential gains at the new location exceed the migration costs and earning prospects at the current place of residence. Families migrate if they expect net gains Gf at the household level.12 G denotes the net gains, and the subscript f indicates the family level. The net household income combines the net gains of both partners, denoted by Gm for men and Gw for women. Within this framework, the family migrates if the net returns are higher than zero. Theoretically, this outcome can occur in three different situations. First, if both partners experience positive net gains (Gm > 0 and Gw > 0). This outcome 12 Actually, the situation in which the family stays intact and both partners migrate together or stay where they are is described as a special case of the Mincer’s model. Mincer (1978, p. 756) notes also that a partner’s incentives to take a job in another region may have a destabilizing effect on the relationship. The relationship between marriage instability and increased migration rates is characterized as a “two-way street”. IAB-Bibliothek 358 54 Overview of migration theories is unusual (Mincer 1978; Kalter 1998). More common is the case in which one of the partners (either male or female) experiences a net gain and the other partner a loss. Thus, individuals within a household might relocate despite individual disadvantages. This might be the case because the decision to migrate is based on the greater common household welfare and occurs ( Gf > 0, Gm > 0, Gw < 0, and |Gw| < Gm ), if the gains of one of the partners from migration exceed the losses of the other partner. The person who moves despite experiencing individual losses is called a “tied mover.” Mincer’s (1978) model is gender-neutral in principle and does not imply specific disadvantages for women. That is, the model does not suggest that men can more easily convince their families to relocate. However, the empirical research based on this model clearly indicates that families will migrate more often to support the husband’s career (e.g. Compton and Pollak 2007; Geist and McManus 2012; Nivalainen 2004; Smits et al. 2003; Tenn 2010). How this is possible is discussed below. Mincer’s model is based on a comparison of incomes. Within this framework, a migration takes place if the gains of one partner can compensate for the losses of the other. If the incomes are asymmetrical, i.e., if one partner has a high income and the other earns little, then the partner who is financially better off can easily compensate for the losses of the other partner (e.g., the loss of this partner’s job). Even if the assumption itself is completely gender-neutral, this feature of the model predicts that women will initiate migration less often. Based on the persistence of income inequality between men and women and women’s lower labor market participation, this model predicts that women are less often financially able to compensate for the losses of their partners due to migration and will initiate migration less often. Another argument frequently used to explain male dominance in the migration decision is that women are likely to invest less in their human capital (cf Markham et al. 1983; Markham and Pleck 1986).13 Empirical research shows that despite the disadvantages associated with the role of the tied mover, women follow their partners more often than do men (Bielby and Bielby 1992; Bird and Bird 1985; Boyle et al. 2001; Compton and Pollak 2007; Duncan and Perrucci 1976; Jacobsen and Levin 1997; Jürges 1998; Jürges 2006; Long 1974; Nivalainen 2004; Shihadeh 1991; Smits et al. 2003; Tenn 2010; for a review see: Cooke 2008). This pattern indicates that when men receive job offers at new locations, they are more likely to realize high-enough gains 13 However, this argument is becoming increasingly irrelevant, as young women now invest more in their education than young men (cf for the educational participation of women OECD 2011, p. 44 ff). 55Chapter 2 Selection of migration theories to compensate for the losses of their partners. In turn, men are also more likely to be the tied stayer because men often find that they can realize high-enough income gains to offset their migration costs and those of their partners, especially if their partners are employed (Mincer 1978; Shaklee 1989; Spitze 1984). This theoretical approach appears to be capable of explaining the situations of men and women with regard to the migration process based on economic factors alone and without relation to cultural, normative or biological differences. However, it is an important point of critique that gender-specific differences in incomes might themselves result from gender-specific behavior or culturally and normatively motivated forms of gender discrimination, which undermines the earlier argument (Elson 1999; Halfacree 1995). Migration exacerbates the inferior labor market positions of women, and the inferior labor market positions of women reinforce their positions as tied movers. It is possible that only the worse labor market integration of women allows families to migrate and makes the financial gains of their male partners high enough to outbalance their losses. Thus, the reason the actual migration can take place is not because the income of the female partner is lower, but it can be traced back to the socialization of women as secondary earners and housewives. The term “tied” indicates that the individual’s migration outcome is dependent on other family members. For both the tied mover and the tied stayer, another situation will maximize their individual gains. Based on the individual gains and losses, remaining in the location of origin would be more profitable for the tied mover. In contrast, the tied stayer would profit individually from moving to the destination. However, for the tied mover, the calculation at the household level reveals that the benefits of moving outweigh the losses. Therefore, the migration occurs. The reverse situation is true for the tied stayer. The tied stayer’s gains are not high enough to offset the losses of the partner in the case of a migration. As long as the characteristics of the partners are very different and the partners specialize in, for example, labor or household work, as suggested by Becker (1991), the “tied” phenomenon is rare. However, a crucial feature of this model is its prediction of a greater commutation of the “tied” phenomenon the more similar the earnings of men and women are. Because gains and losses can be distributed differently, this model also predicts that migration goes hand in hand with specialization (Geist and McManus 2012, p. 199). In a second step Mincer (1978) provided an extended version of this model that considers the role of location. This model can explain why couples occasionally live in regions that are not ideal for either of the partners. To understand this extended model, we must imagine that certain regions would be ideally suited to the respective needs of each partner. However, these regions are not necessarily IAB-Bibliothek 358 62 Overview of migration theories an agreement. As the costs increase, the outside options become more attractive. This implies that persons or couples with extremely good outside options will not invest much time or effort into finding a solution or in their relationship, which might result in shorter relationships. Joint investments include children or the purchase of joint goods such as real estate, the amount of time spent together or certain legal arrangements (e.g., marriage) (Abraham 2003). The higher the investments already are, the greater are a couples’ incentive to hold to the bargained agreements and the higher the likelihood the couple will relocate. This correlation holds even if one of the partners has to endure personal setbacks to relocate. This framework provides not only assumptions about how the partners’ characteristics and their relationship influence the migration decision but, as Nisic (2010) demonstrates, allows for assumptions about how regional factors may influence migration decisions. In line with the arguments of Chiswick (1999) and Borjas (1987) for individual migration, Nisic (2010) notes that depending on the income distribution in the sending and receiving countries, the migration pattern of couples might differ. Couples who can expect greater regional income gains will be less positively selected (e.g., lower education levels) than couples who can expect very small regional income differences upon migration. The higher the income gaps, the more likely couples are to profit from the relocation. This relationship implies that couples are more likely to realize high gains at the household level from the migration even if the “tied mover” loses relatively. When the income gaps are smaller, only couples who are able to secure especially high profits, such as when both partners profit independently from migration, will migrate. The main criticism of the bargaining theory, as already mentioned, is that its extended and complex framework ultimately produces hypotheses that are very similar to those generated by household economic theory, which is a much easier framework (cf Nisic 2010, p. 521). Moreover, as previously stated, the sociological criticisms of the household economy model apply here as well; the approach does not take into account the possible existence of cultural and normative beliefs that ascribe different values to the earnings or employment of men and women (Bielby and Bielby 1992). 2.2.2.3 Gender role theory These approaches “ignore the household roles husbands and wives occupy, the gender-role beliefs they subscribe to regarding those roles, and the effect of these beliefs on both the process and outcome of couples decision making” (Bielby and Bielby 1992, p. 1245). According to the household economy model and the 63Chapter 2 Selection of migration theories bargaining theory discussed above, a woman should not experience additional disadvantages, and her partner should pay the same respect to her migration wishes as the woman affords to those of the man, if both migration proposals guarantee the same gains. However, recent research shows that this statement is not necessarily true, as many characteristics, such as the educational and employment characteristics of men, are of greater importance for the migration decision (Compton and Pollak 2007; Nivalainen 2004; Tenn 2010). Women are usually unable to initiate a move and lose in terms of earnings and employment (Boyle et al. 2001; Boyle et al. 2009; Clark and Withers 2002; Mincer 1978; Rabe 2011; Shauman and Noonan 2007; Shihadeh 1991), even if they previously earned more or had a higher status than their partners (Boyle et al. 2009; Shauman and Noonan 2007). Therefore, the sociological research proposes explanations of male and female behavior based on the normative and cultural values individuals learn in the process of socialization (Fenstermaker and West 2002; West and Zimmerman 1987; West and Zimmerman 2009). This framework indicates that partners might react differently depending on their gender attitude and on whether the male or the female proposes the migration plan (Bielby and Bielby 1992). Bielby and Bielby’s (1992) framework implies that gender attitudes are the moderating factors, and it refers to the distribution of financial means between partners as secondary factors. Gender role attitudes can explain why one family agrees to migrate and the other does not even though the two households have identical migration gains, costs and income distributions between the partners. This theoretical framework suggests that both partners behave during the migration decision-making process in accordance with their normative and cultural values. The idea that men and women must behave a certain way to succeed in society can be traced back to Goffman (1974; 1977). In his main work, (Goffman 1974) developed the idea that people adapt their behavior to certain frameworks, which dictate how they should behave in different circumstances. In subsequent work, Goffman (1977) adapted the idea to the behavior of men and women, indicating that not only do specific situations influence the way people react but so does the person herself, whose gender is “framed” by specific activities. Following this argument, to succeed in their roles as men and women, people must behave accordingly. This concept can also be applied to migration. By meeting their expectations and behaving like women, women support their gender roles. The gender roles used by men and women conform to the dominant pattern within the society (Schulz 2010, p. 93). Couples with traditional views assign varying degrees of importance to the earnings of men and women; their roles are not interchangeable (Potchek 1997). Even if women contribute significantly to the household’s income, men’s IAB-Bibliothek 358 64 Overview of migration theories jobs remain more important for traditional couples. The position of women as secondary providers and unequal co-providers determines the varying behaviors after men and women receive a job offer. Within a family with traditional views, women are likely to be tied movers or tied stayers, depending on whether they would have been able to obtain better job opportunities at the place of origin (when the family moves) or in another place (when the family stays). Men with traditional gender norms are more likely to migrate when they receive a better offer in another region. This situation is different for household with non-traditional gender role views. In these households, a job offer with the same features given to the male or the female partner should result in the same outcomes. The husband should be much more reluctant to relocate for the sake of his own career than the husband of a traditional couple because his wife’s career is accorded the same value as an equal co-provider’s career. Relocations that occur because of the male partner’s career should be less common for couples with non-traditional views because they are not automatically agreed upon by both partners. At the same time, relocations for the sake of the woman’s career will increase because these relocations will not be automatically dismissed, as would be the case for traditional couples. For nontraditional couples, the characteristics of either partner will similarly influence the decision to migrate. Couples with egalitarian views, therefore, behave in a way that was already described by the economic and bargaining theory. One major critique of the model can be derived from the general critique of the doing gender framework. Even if traditional vs. non-traditional or egalitarian gender models are considered, this framework cannot explain how “new” gender arrangements develop (Schulz 2010, p. 93). Thus, within this type of framework, it is difficult to explain why some couples have more traditional views and others less traditional in the same society. Within this framework, it is impossible to explain development within the society. This framework describes only two possible extreme positions, but nothing between those poles. 3 Critique 3.1 General critique As Massey et al. (1998) note, empirical investigations usually support all theoretical concepts to a certain degree. All theories have aspects that might be preferable for the explanation of some specific migration flows; however, these theories also have weak points and aspects that are not supported empirically and are open to criticism. Because theories are selected to fit specific situations, the entire concept 65Chapter 2 Critique of theoretical testing can be undermined (Portes 1997, p. 804), as a rejection of the hypotheses does not necessarily lead to a rejection of the theoretical idea but only indicates that the theory does not fit the specific situation (Bakewell 2010, p. 1692). In the discussion above, specific critiques have already been outlined; however, other general critiques remain. All the theoretical concepts exclude the influence of policies, as Arango (2000) notes (see also: Kurekova 2009). International agreements, quotations and, of course, immigration policies play a powerful role in migration, as they define the migration costs and, thus, who is able to profit from the relocation. Ignoring political aspects might lead to wrong interpretations. Ignoring political features might be especially unsatisfactory for research comparing migration flows because political barriers may explain why two countries with similar economic characteristics have differences in migration patterns. Second, the theoretical frameworks usually concentrate either on the sending or the receiving country or region (Castles 2010, p. 1571) but are mostly unable to account for the characteristics of the countries of destination and origin at the same time. In sociology, the focus has long been on theories that concern the conditions at the destination and that take into account the integration of migrants into the labor market while ignoring the reasons migration occurs in the first place (Brettell and Hollifield 2000; Schmitter Heisler 2000). Third, the theoretical frameworks usually ignore historical contexts (Massey 1990) and focus, as already implied, on economic factors (Bakewell 2010, p. 1690; Castles 2010, p. 1573; Faist 2000, p. 36; Massey et al. 1998, p. 8). Another point of critique is that most of the concepts are unable to explain why migration does not take place even when it would improve the individual’s or family’s situation (Kalter 1997). Here, the concept “bounded rationality” (Jones 1999; Simon 1997) may be helpful to deepen our understanding of migration and explain why many people do not even consider migration (Kalter 1997, p. 51). In addition, research that understands migration as more of a step process, in which the first step is to develop the intention to migrate (see De Jong 2000; Kalter 1997), can help address this criticism. Finally, most of the theoretical frameworks have a static perspective and are unable to explain changes in migration when the external circumstances remain stable (cf Kurekova 2009). 3.2 Critique from a gender perspective The overview of the migration theory clearly outlined that the existing theoretical concepts do not focus on gender differences in the migration patterns of men and women. As Kofman et al. (2000, p. 17) noted, studies providing theoretical IAB-Bibliothek 358 66 Overview of migration theories frameworks for women’s migration are missing. Although the theoretical concepts of migration are usually described as being gender-neutral, they are better fitted to explain the migration patterns of men than of women, and they rely on the assumption that women are tied movers (Kofman 2000; Kofmann et al. 2000; Schwenken and Eberhardt 2008). This is especially true because the understanding of migration and the underlying mechanisms on which the concepts rely are based on a ‘prototype’ of migrants that is often referred to as “homo economicus,” a man whose characteristics have been described as “narrow rationality, selfishness and social isolation” (Schwenken and Eberhardt 2008, p. 13). A view from the gender perspective reveals that when analyzing the migration of men and women and relying on the described concepts, gender-specific differences might be ignored even if the theoretical framework is described as gender-neutral, and gender in fact remains a reason for the differences revealed. For future work in this field, it is important to realize that such problems exist and that the picture presented by the models might be oversimplified. A great challenge for future research is to find a balance between taking into account that men’s and women’s migration patterns might be influenced by different types of socialization or discrimination on the one hand and, on the other hand, ensuring a certain generality that theoretical concepts and models can rely on. The answer to the problem is not to abstain from quantitative research, but to be aware of the limitations of the models used and that gender differences exist. 4 Discussion Migration is a complex phenomenon in which rational actors make decisions. These decisions might be influenced not only by individual characteristics but also by conditions at the macro-level in the origin and the destination countries and those at the meso-level (e.g. Coleman 1990; de Haas 2010; Esser 1991; Haug 2008). As already mentioned above, the theoretical frameworks described herein already provide some solutions to integrate the differences into the theoretical construct and, thus, to account for gender-specific differences, even if such a process has received criticism. At the same time, considering the living circumstances of men and women and all three levels of explanation (the micro-, meso- and macro-levels) is a challenge to every theoretical framework. Every level might bear gender-specific differences, which might be more or less important for the specific migration flows and target group under investigation (cf Coleman 1990; and also Esser 1991, p. 113; Haug 2008, p. 590; Lutz 2010, p. 1658), as we can see in Figure 1. For married women, the meso-level and their position in the household as well as the gender role beliefs of their partner should be especially 67Chapter 2 Discussion important. How female participation in the labor market is understood is also important for them, for example, as secondary earners or equal co-providers. For single women, in turn, their labor market position, such as a hierarchical position in the company, might enable them to access attractive jobs in other regions. Finally, married women might also migrate not because it is beneficial from their own cost-benefit calculations but because their partners wish to migrate. It is practically impossible to account for all the different aspects, situations and conditions that a theoretical framework would have to consider to explain all aspects of men’s and women’s migration behaviors within a single framework (cf Bakewell 2010; Castles 2007; Castles 2010; Portes 1997). Especially at the macro-, meso- and micro-levels, it is difficult to account for the different migration patterns of married and unattached women, which is why it might be better to rely on the various theoretical frameworks and to adapt or combine them to fit the specific situation under investigation. It might be advantageous to aim not for the grand theory of migration, which cannot explain all aspects of migration at the destination and origin or the differences in men’s and women’ behavior, but for “theories of the middle-range” (Arango 2000; Bakewell 2010; Castles 2007; Castles 2010; Portes 1997).17 These middle-range theories have the benefit of accounting for some patterns and similarities of the highly fractured phenomenon of migration without being able to account for all its aspects (Castles 2010, p. 1574). This point brings us back to the argument put forth by Massey et al. (1993) and clarifies why it is necessary to understand migration theories as theoretical constructs, which can be combined and used simultaneously, rather than as competing frameworks. 17 Castles (2010, p. 1574) refers to Merton’s (1957) “theories of the middle-ranges”. Figure 1: Macro-meso-micro model of migration Macro level: gender-specific situation • Labor market conditions • Cultural context Macro level: gender-specific migration rates Meso level: gender-specific context • Social position at the labor market • Social position within family Micro level: gender-specific individual characteristics • Economic capital • Human capital • Cultural capital Actor Migration decision IAB-Bibliothek 358 68 Overview of migration theories 5 References Abraham, M. 2003. “Die Stabilisierung von Partnerschaften durch bilaterale Investitionen. Das Beispiel der Unternehmensbesitzer.” Zeitschrift für Soziologie 32: 50–69. Abraham, M. 2006. Berufliche Selbstständigkeit: Die Folgen für Partnerschaft und Haushalt. Wiesbaden: VS-Verlag. Anker, P. 2001. “Theories of occupational segregation by sex.” Pp. 120–155 in Women, gender and work, edited by M.F. Loutfi. Geneva: ILO. Arango, J. 2000. “Explaining migration: A critical view.” International Social Science Journal 52: 283–296. Arrow, K.J. 1973. “Higher education as a filter.” Journal of Public Economics 2: 193–216. Auspurg, K. and M. Abraham. 2007. “Die Umzugsentscheidung von Paaren als Verhandlungsproblem.” Kölner Zeitschrift für Soziologie und Sozialpsychologie 59: 271–293. Bailey, A.J. and T.J. Cooke. 1998. “Family migration and employment: The importance of migration history and gender.” International Regional Science Review 21: 99–118. Bakewell, O. 2010. “Some reflection on structure and agency in migration theory.” Journal of Ethnic and Migration Studies 36: 1689–1708. Baron, J.N. and W.T. Bielby. 1986. “Sex segregation within occupations.” The American Economic Review 76: 43–47. Becker, G.S. 1974. “A theory of marriage.” in The economics of the family: Marriage, children, and human capital, edited by T.W. Schulz. Chicago: University Chicaco Press. Becker, G.S. 1991. A treatise on the family. Cambrige: Harward University Press, 1981 Enlarged edition. Behrman, J.R. and L.B. Wolfe. 1984. “Micro determinants of female migration in a developing country: Labour market, demographic marriage market and economic marriage markets incentives.” Research in population economics. A research annual 5: 137–66. Bergmann, B. 1995. “Becker’s theory of the family: Preposterous conclusion.” Feminist Economics 1: 141–150. Berk, R.A. and S. Fenstermaker Berk. 1983. “Supply-side sociology of the family: The challange of the new home economics.” Annual Review of Sociology 9: 375–95. Bernasco, W. and D. Giesen. 2000. “A bargaining approach to specialization in couples.” Pp. 42–43 in The management of durable relations. Theoretical models and empirical studies of households and organizations, edited by J. Weesie and W. Raub. Amsterdam: Thela Thesis. 69Chapter 2 References Bielby, W.T. and D.D. Bielby. 1992. “I will follow him – family ties, gender-role beliefs, and reluctance to relocate for a better job.” American Journal of Sociology 97: 1241–1267. Bird, G.A. and G.W. Bird. 1985. “Determinants of mobility in 2-earner families – does the wifes’ income count.” Journal of Marriage and the Family 47: 753–758. Blau, F., M. Feber, and A.E. Winkler. 1992. The economics of women, men, and work. New Yersey: Prentice-Hall. Blau, P. 1964. Exchange and power in social life. New York u.a.: Wiley. Blood, R.O. and D.M. Wolfe. 1960. Husbands & wives. The dynamics of married living. Glencoe: Free Press. Blossfeld, H.-P. 2009. “Educational assortative marriage in comparative perspective.” Annual Review of Sociology 35: 513–530. Blossfeld, H.-P. and K.U. Mayer. 1988. “Labor market segmentation in the Federal Republic of Germany: An empirical study of segmentation theories from a life course perspective.” European Sociological Review 4: 123–140. Blossfeld, H.-P. and A. Timm. 2003. Who marries whom? Educational systems as marriage markets in modern societies. Dordrecht: Kluwer Acad. Boje, T.P. 1986. “Segmentation and mobility: An analysis of labour market flows on the danish labour market.” Acta Sociologica 29: 171–178. Borjas, G.J. 1987. “Self-selection and the earnings of immigrants.” The American Economic Review 77: 531–553. Borjas, G.J. and B. Bratsberg. 1994. “Who leaves? The outmigration of the foreignborn.” The Review of Economics and Statistics 78: 165–76. Boswell, C. 2008. “Combining economics and sociology in migration theory.” Journal of Ethnic and Migration Studies 34: 549–566. Boswell, C. and P.R. Mueser. 2008. “Introduction: Economics and interdisciplinary approaches in migration research.” Journal of Economic and Migration Studies 34: 519–529. Boyle, P., T.J. Cooke, K. Halfacree, and D. Smith. 2001. “A cross-national comparison of the impact of family migration on women’s employment status.” Demography 38: 201–13. Boyle, P., Z. Feng, and G. Vernon. 2009. “A new look at family migration and women’s employment status.” Journal of Marriage and the Family 71: 417–431. Brettell, C.B. and J.F. Hollifield. 2000. “Introduction: Migration theory. Talking across disciplines.” Pp. 1–26 in Migration theory. Talking across disciplines, edited by C.B. Brettell and J.F. Hollifield. New York and London: Rotledge. Brücker, H. and C. Defoort. 2009. “Inequality and the self-selection of international migrants: Theory and new evidence.” International Journal of Manpower 30: 742–764. IAB-Bibliothek 358 70 Overview of migration theories Brücker, H. and P. Trübswetter. 2007. “Do the best go west? An analysis of the selfselection of employed East-West migrants in Germany.” Empirica 34: 371–395. Cain, G.G. 1976. “The challenge of segmented labour market theories to ortodox theory: A survey.” Economic Literature 14: 1215–1257. Canziani, P. and B. Petrongolo. 2001. “Firing costs and stigma: A theoretical analysis and evidence from microdata.” European Economic Review 45: 1877–1906. Cappelli, P. 1995. “Rethinking employment.” British Journal of Industrial Relations 33: 563–602. Castles, S. 2007. “Twenty-first century migration as a challange to sociology.” Journal of Ethnic and Migration Studies 33: 251–71. Castles, S. 2010. “Understanding global migration: A social transformation perspective.” Journal of Ethnic and Migration Studies 36: 1565–1586. Castles, S. and M.J. Miller. 1993. The age of migration. International population movements in the modern world. Basingstroke: Macmillan. Chiswick, B.R. 1978. “The effect of americanization on the earnings of foreign-born men.” Journal of Political Economy 86: 897–921. Chiswick, B.R. 1991. “Speaking, reading and earnings among low-skilled immigrants.” Journal of Labor Economics 9: 149–170. Chiswick, B.R. 1999. “Are immigrants favorably self-selected?” The American Economic Review 89: 181–185. Clark, W.A.V. and S.D. Withers. 2002. “Disentangeling the interaction of migration, mobility and labor-force participation.” Enviroment and Planning 34: 923–945. Coleman, J.S. 1956. Comunity conflict. Glencoe, IL: Free Press. Coleman, J.S. 1990. Foundation of social theory. Cambridge, Mass., and London: Harvard University Press. Coltrane, S. 2000. “Research on household labor: Modeling and measuring the social embeddedness of routine family work.” Journal of Marriage and Family 62: 1208–1233. Compton, J. and R.A. Pollak. 2007. “Why are power couples increasingly concentrated in large metropolitan areas?” Journal of Labor Economics 25: 475–512. Constant, A.F. and D.S. Massey. 2005. “Labor market segmentation and the earnings of german guestworkers.” Population Research and Policy Review 24: 489–512. Cooke, T.J. 2005. “Migration and employment among the civilian spouses of military personnel.” Social Science Quarterly 86: 343–355. Cooke, T.J. 2008. “Migration in a family way.” Population Space and Place 14: 255–265. Cooke, T.J. 2011. “Marriage markets and the intermetropolitan distribution of skilled couples.” Growth and Change 42: 98–110. Cooke, T.J., P. Boyle, K. Couch, and P. Feijten. 2009. “A longitudinal analysis of family migration and the gender gap in earnings in the United States and Great Britain.” Demography 46: 147–167. 71Chapter 2 References Cooke, T.J. and M. Rapino. 2007. “The migration of partnered gays and lesbians between 1995 and 2000.” Professional Geographer 59: 285–207. Curtis, R.F. 1986. “Household and family in the theory on inequality.” American Sociological Review 51: 168–183. DaVanzo, J. 1972. “Analytical framework for studing u.s. interregional migration.” RAND Corp. Res. Report. De Haas, H. 2010. “Migration and development: A theoretical perspective.” International Migration Review 44: 227–264. De Jong, G.F. 2000. “Expectations, gender, and norms in migration decision-making.” Population Studies 54: 307–319. Doeringer, P.B. and M.J. Piore. 1971. Internal labor markets and manpower analysis. Lexington, Massachusets: Heath Lexington Books. Duncan, R.P. and C.C. Perrucci. 1976. “Dual occupation families and migration.” American Sociological Review 41: 252–261. Edlund, L. 2005. “Sex and the city.” Scandinavian Journal of Economics 107: 25–44. Elson, D. 1999. “Labor markets as gendered institutions: Equality effciency and empowerment issues.” World Development 27: 611–627. Esser, H. 1991. Soziologie. Allgemeine Grundlagen. New York: Campus. Faist, T. 2000. The volume and dynamics of international migration and transnational social spaces. Oxford: Oxford University Press. Fenstermaker, S. and C. West. 2002. “”Doing difference” revisted: Problems, prospects and the dialogue in feminist theory.” Pp. 105–114 in Doing gender, doing difference: Inequality, power, and institutional change, edited by S. Fenstermaker and C. West. New York: Routledge. Ferber, M.A. and B.G. Birnbaum. 1977. “The ´new home economics´: Retrospects and prospects.” Journal of Consumer Research 4: 19–28. Folbre, N. 1984. “Cleaning house: New perspectives on households and economic development.” Journal of Development and Economics 22: 5–40. Gautier, P.A., M. Svarer, and V.N. Teulings. 2010. “Marriage and the city. Search frictions and sorting of singles.” Journal of Urban Economics 67: 206–218. Geist, C. and P.A. McManus. 2012. “Different reasons, different results: Implications of migration by gender and family status.” Demography 49: 197–217. Gibbons, R. and L.F. Katz. 1991. “Layoffs and lemons ” Journal of Labor Economics 9: 351–380. Goffman, E. 1974. Frame analysis: Essays on the organization of experience. Northeastern: University Press. Goffman, E. 1977. “Interaktionen und Geschlechter.” In Das Arrangement der Geschlechter, edited by E. Goffman. Frankfurt a.M.: Suhrkamp Verlag. Gordon, I. 1994. Pay, conditions and segmentation in the london labour market. University of Reading, Reading: mimeo. IAB-Bibliothek 358 78 Overview of migration theories Stark, O. 2006. “Inequality and migration: A behavioral link.” Economics Letters 91: 146–152. Stark, O. and D.E. Bloom. 1985. “The new economics of labor migration.” The American Economic Review 75: 173–178. Stark, O. and D. Levhari. 1982. “On migration and risk in LDCS.” Economic Development and Cultural Change 31: 191–196. Stark, O., M. Micevska, and J. Mycielski. 2009. “Relative poverty as a determinant of migration: Evidence from Poland.” Economics Letters 103: 119–122. Stark, O. and J.E. Taylor. 1989. “Relative deprivation and international migration.” Demography 26: 1–14. Stark, O. and J.E. Taylor. 1991. “Migration incentives, migration types – the role of relative deprivation.” The Economic Journal 101: 1163–1178. Stark, O., J.E. Taylor, and S. Yitzhaki. 1986. “Remittances and inequality.” The Economic Journal 96: 722–740. Stark, O. and S. Yitzhaki. 1988. “Labor migration as a response to relative deprivation.” Journal of Population Economics 1: 57–70. Steijn, B., A. Need, and M. Gesthuizen. 2006. “Well begun, half done? Longterm effects of labour market entry in the Netherlands, 1950–2000.” Work, employment and society 20: 453–472. Taylor, J.E. 1999. “The new economics of labor migration and the role of remittances in the migration process.” International Migration 37: 63–88. Tenn, S. 2010. “The relative importance of the husband’s and wife’s characteristics in family migration, 1960–2000.” Journal of Population Economics 23: 1319–1337. Thadani, V.N. and M.P. Todaro. 1979. “Female migration in developing countries; a framework for analysis.” Working Paper (Population Concil: Center for Policy Studies) no. 47. Waller, W. and R. Hill. 1951. The family: A dynamic interpretation. New York: Holt, Rinehard and Winstron. Ware, H.R.E. 1981. Women, demography and development. Canberra: Australian National University. Weiss, A. 1995. “Human capital vs. signaling explanations of wages.” Journal of Economic Perspectives 9: 133–154. West, C. and D.H. Zimmerman. 1987. “Doing gender.” Gender & Society 1: 125–151. West, C. and D.H. Zimmerman. 2009. ” Accounting for Doing Gender.” Gender & Society 23: 112–122. Chapter III Reconsidering the effect of education on East-West migration in Germany1 1 This chapter is published in European Sociological Review 29 (Issue 2 April 2013), p. 210–228. 81 Introduction Chapter 3 Abstract This article analyzes migration from East to West Germany, focusing on the influence of education on migration and on the self-selection processes involved in decisions regarding education and migration. Using human capital, signaling, and segmentation theory, hypotheses are derived on the influence of education on migration. The migration patterns for men and women are investigated on the basis of SOEP data from 1992 to 2007. The results of the hierarchical logit regression models show that the level of education influences the migration decisions of both men and women. However, Heckman selection models reveal that only the migration patterns of women are defined by a selection of upper secondary education. For women, the results suggest that the same mechanisms drive their participation in upper secondary education and in migration. 1 Introduction Twenty years after reunification, incomes and corresponding living standards continue to vary considerably between East and West Germany. In 2006, for instance, the average gross income of men in East Germany was only 69 percent of the income in West Germany, having already reached 57 percent in 1992 (for women, the gap declined from 32 to 19 percent).2 Hardly any other European country has such large regional income differences and such high incentives for relocation; such income differences are more pronounced for international migration. In contrast, to international migration, in the case of East-West migration, the costs of relocation are extraordinarily low due to the lack of legal constraints, low transportation costs, and absence of language barriers. This remarkable combination should represent powerful incentives to migrate for all population groups. However, even in East Germany, migrants tend to be self-selected; only those who will profit from migration are willing to relocate, and these individuals usually have different characteristics than the remaining population. In East Germany, migrants show more favorable characteristics than the average population, e.g. they are younger and better qualified than the remaining population. This relationship between migrants and their favorable characteristics and its consequences have been the focus of public discussions and political debates for some time. Some new federal states have even launched programs to reduce migration or to attract former migrants back and facilitate return migration.3 2 My calculations are based on the IAB Beschäftigten-Historik (BeH) V7.01, Nuremberg 2007. 3 Mecklenburg-Western Pomerania launched the Internet site mv4you, which provides information on child care, working conditions, transport connections, and job offers for those interested in relocation. IAB-Bibliothek 358 82 Reconsidering the effect of education on East-West migration in Germany Although numerous studies and various disciplines have addressed East-West migration using individual and aggregate data, our understanding of the migration process, and especially the influence of education on migration, remains limited. Demographic studies at the macro level primarily focus on migration trends that influence the development and structure of the population, marking differences between East and West (e.g., Heiland 2004). In contrast, economic studies based on aggregate data focus on the economic push and pull factors driving migration (cf Arntz 2010; Hunt 2006). Macro-level studies focusing on the impact of education on migration and investigating the “brain drain” are lacking in the case of East-West migration. For East Germany, research on the influence of education on migration tends to be based on individual data, though this line of research provides mixed evidence. Although it is apparent that migrants tend to be younger (Hunt 2006; Schwarze and Wagner 1992; Wagner 1992; Windzio 2007) and to have earned already more in East Germany than comparable “stayers”, i.e., those who decided to stay instead of migrate (Brücker and Trübswetter 2007; Hunt 2006; Windzio 2007), no clear trend is evident for the influence of education on migration. Using ordinal logit models and controlling for commuting or backward commuting, Hunt (2006) states that migrants and commuters are generally more likely to have higher education than people who stay in East Germany. According to Hunt (2006), individuals with a tertiary education are 83 percent more likely to leave East Germany than those without an educational or vocational degree. Young college graduates are especially mobile: they are five times more likely to migrate to the West than persons without an educational or vocational training. Hunt (2006) also identifies a group of young migrants with low education, who can be described loosely as pupils or students, who migrate to continue their education or start apprenticeships. Other studies do not identify any effect of education on migration (e.g., Brücker and Trübswetter 2007; Windzio 2007).4 Instead, Brücker and Trübswetter (2007) show that migrants are positively selected from the remaining population with regard to their income and thus are likely to have unobserved positive characteristics. These authors also point out that the wage premium for highly educated individuals is lower in the East than in the West, which would create additional incentives for this group to migrate. Recent international research (cf Carrington and Detragiache 1999; Docquier and Rapoport 2007) has also suggested that migrants are self-selected 4 The data samples used by authors who find an influence of education differ from those who do not find this influence. Hunt (2006) uses SOEP data, whereas Windzio (2007) and Brücker and Trübswetter (2007) base their analyses on IABS data. 83 Introduction Chapter 3 on education.5 These analyses have provided initial indications that individuals who move to the West might be self-selected according to their education. However, migration and education are probably interwoven with each other to a higher degree. The human capital approach implies that the reasons for participating in education and migration are defined by the same mechanisms, as both processes can be understood as investments. The first process can be understood as an investment in human capital, and the second can be understood as an investment in the productivity of human capital. A strong career orientation, for example, can influence the decision to invest in both education and migration. Migration can also serve as a tool to realize full gains from previous educational investments. To account for the relationship between education and migration, it is therefore necessary to control for observed and unobserved characteristics of individuals. The second issue that has received significant public attention is the lack of young women in the East. In the public opinion, highly educated young women leave East Germany, whereas less educated men with right-wing political views remain there.6 In fact, there are 115 men between the ages of 18 and 30 who live in East Germany for every 100 women who live there (Grünheid 2009, p. 36 ff). This large difference between the numbers of young men and women living in East Germany can be used as an “argumentum e contrario” for the higher migration rate of young women. Moreover, these numbers may indicate different migration patterns for men and women.7 Although these numbers seem demographically and socially alarming, this topic has seldom been addressed in empirical research. The few studies that do address gender differences in the German East-West migration explicitly link gender-specific participation to education and migration and explain the higher migration rates of young women by their higher educational achievements (Dienel and Gerloff 2003; Gerloff 2004). Analyses of extended data considering both the gender and heterogeneity of migrants are needed to provide further insight into the influence of education on migration in East Germany. If the migration patterns of men and women indeed differ, investigating education without taking genderspecific behavior into account would provide an incomplete if not a biased picture. 5 Doquier and Rapoport (2007), for example, used data spanning five continents to show that the share of highly skilled workers is usually higher among migrants than among the remaining population. 6 See, e.g., Berlin-Institut für Bevölkerung und Entwicklung (2007) Not am Mann: Vom Helden der Arbeit zur neuen Unterschicht?; DRadio Wissen-Kultur (05.08.2010) Not am Mann; The Economist (28.06.2007) We ain’t got dames; Die Welt (30.05.2007) Im Osten fehlen die Frauen. 7 Recent international research on developing as well as industrialized countries provides additional evidence of differences in the mobility patterns of women and men (Dumount et al. 2007). In the UK, for example, young women are more likely than young men to leave their city to study. After finishing their tertiary education, young women are again more likely to leave their city for a new destination, whereas their male fellow students are more likely to stay in their current residence or return to their city of origin (Faggian et al. 2007). IAB-Bibliothek 358 84 Reconsidering the effect of education on East-West migration in Germany This article aims to fill these gaps, focusing on the influence of education on migration and on the self-selection processes involved in the education and migration decisions of males and females. The aim is to answer the following questions. First, to what extent does education influence migration from East to West Germany? Second, are migrants self-selected based on education? Third, do men and women differ in terms of education and self-selection? In the following, I use theories of human capital, signaling, and segmentation to derive hypotheses about the influence of education on migration. I then investigate the migration patterns of men and women based on the individual behavior causing the macro-level movements. This analysis is based on SOEP data from 1992 to 2007 and includes measures for education, on-the-job training, and labor market participation. The dataset is composed of men and women between 18 and 60 living in East Germany. Using hierarchical regression models, I investigate the influence of education on migration from East to West Germany. I estimate separate models for men and women to examine migration patterns and carry out combined estimations to investigate factors with a gender-specific influence. Analyzing the migration patterns of men and women not only improves our understanding of sources of gender-specific inequalities but also provides general information on the causes of migration and how these might be influenced or even reduced. Finally, the analysis provides information that can be useful in estimating the migration potential of future events, such as the eastern expansion of the EU. The individual decision to migrate is also embedded in the process of regional change. Therefore, I control for regional factors, including timedependent, gender-specific variables of regional unemployment and income levels (NUTS 3 level) as well as time-stable characteristics of the region, such as population density. In a further step, the Heckman selection model is used to control for unobserved characteristics of migrants that influence both the decision to pursue further education and the decision to migrate. The analysis of selectivity should deepen our understanding of migration and provide information for further research on the processes driving migration. 2 Theory 2.1 Human capital theory In human capital theory, migration is treated as an allocation strategy to maximize individual income and consumption based on education and work experience 85 Theory Chapter 3 (Borjas 1987; Chiswick 1999; Chiswick 1978; Sjaastad 1962). Migration is a risky investment in the future, similar to schooling or on-the-job training. Therefore, people must be willing to carry present costs to gain future benefits (Sjaastad 1962). The costs of migration are of a monetary and non-monetary nature. Whereas the monetary costs are mostly transportation costs, non-monetary costs are more substantial. They cover the location-specific costs such as, for example, the loss of family and friendship networks (DaVanzo 1983) and opportunity costs. Spatial income gaps offer incentives for migration: the higher they are, the more individuals can profit. Two kinds of human capital, specific and general knowledge, dictate the returns of migration. Specific knowledge includes information obtained through the institution in which a person is employed; it is usually acquired on the job and is strongly associated with seniority. It cannot be easily transformed and loses its value after a change of employer. As a result, productivity as well as financial compensation should drop in a new job. Individuals with higher levels of specific knowledge should be less likely to migrate (H1: human capital theory). General knowledge, however, is usually acquired in educational institutions. Its universality enables people with high levels of general education to switch employers more easily and permanently, as productivity levels should be comparable among institutions. Individuals with higher levels of general education should be more likely to migrate (H2: human capital theory). Moreover, individuals with higher general knowledge usually have reduced non-monetary migration costs, for example, due to broader friendship networks (cf Brücker and Trübswetter 2007, p. 374). The ratio of specific to general knowledge defines the gains from migration. Individuals with low proportions of specific to general knowledge will profit most from relocation. The ratio differs strongly but not randomly among population groups. Younger people who have recently left general education and are at the beginning of their careers have a lower ratio of specific knowledge to general education than older employees. Younger individuals should be more likely to migrate (H3: human capital theory). This also follows from the fact that younger individuals have a longer remaining duration to spend in the labor force, during which time they profit from higher wages. The theoretical approach by Sjaastad (1962) provides general hypotheses, which do not take into account the situation in the East and West German labor markets. Borjas (1987) and Chiswick (1978; 1999) extend this classic human capital approach, taking into account discrepancies in income distributions that cause different returns on migration for individuals with varying education levels. This provides a way to account more precisely for the specific situation in East and West Germany. If the income gap is large enough, all groups of the population should IAB-Bibliothek 358 86 Reconsidering the effect of education on East-West migration in Germany profit from migration regardless of the income distribution in the sending and receiving regions. If the gap is small, the individual’s gain is dependent on income distribution, and only certain groups profit from migration. If the earnings at the destination are less equal than at the origin, people at the higher end of the income distribution – usually those with higher educational levels – will migrate. They benefit not only from the income gap but also from the higher marginal value of their education at the destination. The lower educated, in turn, favor destinations with narrow income structures. The penalty for the lack of education is lower in those regions because the divergence in living standards between rich and poor is smaller. As a former socialist country with high income equality, East Germany is still influenced by the past and shows considerably lower income inequality than West Germany (Statistisches Bundesamt 2004, p. 627 ff). Due to the differences in income distribution between the two parts of Germany, highly educated individuals from the East should profit most from migration.8 The extended human capital theory thus provides the same prediction as summarized in H2. 2.2 Signaling theory Signaling theory extends the human capital approach by addressing differences in productivity among workers that can only be observed indirectly via education (Spence 1974; Weiss 1995). Employers are usually not able to measure all characteristics that influence workers’ productivity of directly, and they are prohibited from asking some questions. At the same time, companies that need highly qualified employees – requiring expensive on-the-job training – are interested in productive and reliable workers. The main idea behind this approach is that differences in worker productivity are caused by unobserved characteristics that drove the workers’ prior educational choices and achievements and that affect their current performance (Spence 1974; Weiss 1995). At this point, education provides a reliable proxy for productivity. Individuals with higher educational levels are not a random sample of the population; they are generally healthier, less likely to smoke or take drugs, and less likely to quit (Weiss 1995). For migrants, education plays an even more critical role (Katz and Stark 1987; Stark and Bloom 1985). For example, employers have a more difficult time checking previous employment references for individuals who did not previously work in the region because they cannot easily make use of unofficial channels for background checks. The employer therefore takes a higher risk by hiring a person 8 There is a potential counter-mechanism that must be mentioned. In societies with compressed income structures, high-skilled jobs are less expensive relative to low-skilled jobs. This might encourage companies with high demand for skilled work to relocate to East Germany and reduce the migration of highly skilled workers. 87 Theory Chapter 3 from another country or region. Educational and vocational qualifications reduce this risk. In Germany, every effort has been made to ensure the comparability of educational and occupational qualifications between East and West Germany.9 The Unification Treaty (article 37) enables East Germans to verify their educational and vocational qualifications. The hypothesis derived from signaling theory agrees with H2, which is derived from human capital theory. Individuals with higher levels of general education should find jobs more easily in West Germany and thus should be more likely to migrate. 2.3 Segmentation theory The idea behind this approach is that the different “segments” of the labor market offer different conditions and thus require different adoption strategies. Not all of these market segments obey the law of supply and demand; some are ruled by agreements or social norms. In their original articulation of these segments, Doeringer and Piore (1971) and Piore (1979) distinguish between the internal and external market. For Germany, Lutz and Sengenberger (1980) and Sengenberger (1987) broaden this approach and distinguish between two external markets, defining a third segment: the professional labor market. The internal labor market offers high job stability, high prestige, good pay, and well-defined career opportunities (Blossfeld and Mayer 1988). The possibility that an internal market exists increases with the size of the company. It is characterized by high interdependency between employers and employees and a need for special skills. Such skills cannot be acquired outside the company, and training costs are high. In the internal market, employers are interested not only in short-term productivity but also in the continuity of work and long-term gains. Additionally, because replacing workers is expensive, employers offer workers incentives to prevent them from quitting. Consequently, employees are not hired just for a single job but rather are given a contract stipulating progressive job responsibilities with a clear career definition, and they are paid based on the position they occupy. The opposite of the internal market is the unstructured external market, which is governed by the neoclassical laws of supply and demand.10 Incomes are low and 9 The Unification Treaty (article 37) ensures the comparability of vocational and educational qualifications. Initial steps in this direction were taken as early as August 1990 with the passage of West German legislation on vocational training. Vocational training was already relatively similar between East and West Germany due to the standardization and formalization of occupations (Konietzka 2003). 10 Kleber (1988) points out that internal and external markets are not equivalent to primary and secondary. He shows that although the differentiation between internal and external labor markets can be derived from the special conditions in those segments, the differentiation between primary and secondary is based to a greater degree on a description of the phenomena. IAB-Bibliothek 358 94 Reconsidering the effect of education on East-West migration in Germany taking the unobserved heterogeneity into account, a selection control described by Hackman and Smith (1996, p. 69 ff) is used.16 In their empirical application, Heckman and Smith (1996) investigate the impact of employment and training schemes on re-entry into the labor market. The aim was not only to analyze the direct impact of the scheme on re-entry but also to consider that the unobservable characteristics of people participating in such programs may differ from those who do not participate. A similar situation can be found in the case of migration. People participating in education are more likely to migrate. However, this might not be due solely to the fact that they have a higher educational degree; rather, it may be because of unobserved characteristics such as career orientation that drive participation in both education and migration. Finally, other than fixed and random effects regressions, the Heckman selection not only controls for unobserved heterogeneity but can actually show whether the estimations are influenced by unobserved heterogeneity. In this approach, y defines the probability of migration, and z* defines the probability of attending upper secondary school and serves as a correlation bias. The Mills ratio is estimated from z * (Equation 2) and is introduced into the first estimation to investigate the process of interest: migration (Equation 1). δ defines the impact of the probability of upper secondary school attendance on migration for all individuals, even those who have not yet completed upper secondary schooling.17 y = xi β + z δ + μi (1) z* = wiγ + εi (2) z* > 0 if zi = 1; zi = 0 otherwise y dependent variable of interest x variables that predict the dependent variable β coefficient of x εi, μi error terms z* selection rule w variables that predict the z* γ coefficient of w z inverse Mills ratio δ coefficient of z 16 Björklund and Moffitt (1987) show that this mechanism could be used for other applications, such as to investigate the impact of migration, union membership, or education on earnings. 17 For the formulas, see Heckman and Smith (1996) and Björklund and Moffitt (1987). 95 Results Chapter 3 4 Results In this section, the results of the random effects hierarchical logit regressions for migration from East to West Germany are discussed. Table 1 presents the results for women, and Table 2 presents the results for men. The estimations indicate that women with upper secondary education are more likely to migrate to the West than those without upper secondary education, with a lower secondary degree or with tertiary education. The high mobility of the group with secondary education may be partly due to the fact that they migrate to continue tertiary or vocational education. Individuals with tertiary education are less likely to relocate than those with upper secondary education. The migration of such highly educated individuals may be lower because the highly educated may have better chances of finding jobs in the East; for example, unemployment rates are generally lower for more qualified individuals. In addition, having received their tertiary education in the East, these individuals may also be able to establish better networks, which could improve their access to the labor market. With every year a women spends in a company, her likelihood of migration declines by five percent.18 Firm-specific knowledge cannot easily be transferred between companies. Therefore, for migration to become attractive to individuals who have spent a long time with a company and have accumulated significant specific knowledge, other factors must take precedence over the loss of seniority, and the rewards of education or general labor market experience must be exceptionally high. At the same time, seniority increases wages over the life course in East and West Germany to the same degree, such that for each additional year of seniority, the wages rise by the same degree in the East and in the West (Orlowski and Riphan 2009). However, this is not true for the labor market experience gained in all jobs over the life course in East German companies. Experience accumulated in the East is valued less in West Germany than experience accumulated in the West (Orlowski and Riphan 2009), which is especially important for individuals who have already accumulated significant experience. Workers with higher seniority who migrate to the West are at a disadvantage not only because it is difficult to transfer firm-specific knowledge but also because their overall labor market experience gained in East German companies is valued less in the West. 18 Note that the effects found in random effects models cannot be interpreted as causal effects of x on y. For an interpretation of the effects, odds ratios can be calculated with the following formula: exp(-0.0550)-1=0.0535. IAB-Bibliothek 358 96 Reconsidering the effect of education on East-West migration in Germany Table 1: Determinants of the migration from East to West Germany; females, 1992–2006; application of random effect logit regressions women I II III IV individual characteristics reference group: lower secondary education intermediate secondary education -0.0310 -0.1161 -0.0855 -0.0925 (0.1714) (0.1800) (0.1825) (0.1830) upper secondary education 0.6514*** 0.4805* 0.4699* 0.4337* (0.1884) (0.1913) (0.1976) (0.1985) tertiary education 1.3176*** 0.2611 0.2672 0.2484 (0.2332) (0.2327) (0.2425) (0.2446) vocational training -0.5671*** -0.1700 -0.1865 -0.1367 (0.1302) (0.1355) (0.1388) (0.1394) age 0.0897 0.1300* 0.1042 (0.0493) (0.0543) (0.0541) age squared -0.0020** -0.0024** -0.0020** (0.0007) (0.0007) (0.0007) partner -1.1046*** -1.0515*** -0.9990*** (0.1494) (0.1501) (0.1450) reference group: no children in the household children younger than six -0.0986 -0.1537 -0.1861 (0.1828) (0.1910) (0.1908) children between six and 19 -0.4500* -0.4706* -0.3935 (0.2011) (0.2036) (0.2044) duration spent in the last company -0.0547** -0.0550** (0.0182) (0.0184) cumulative duration spent in unemployment 0.0210 -0.0056 (0.0428) (0.0442) gross income (in €100) 0.0545 0.0411 (0.0452) (0.0450) reference group: unemployed employed part-time -0.6338 -0.5715 (0.3728) (0.3710) in apprenticeship 0.0514 0.2249 (0.3128) (0.3123) employed full-time -0.3082 -0.2023 (0.3329) (0.3326) homeowner -0.6848*** (0.1395) 97 Results Chapter 3 women I II III IV regional characteristics regional income level -0.0122*** (0.0034) unemployment rates -0.0064 (0.0130) distance to the old federal states (10 km) 0.0014 (0.0008) urban area 0.2155 (0.1415) n person-years 26284 26284 26284 26284 n person 3975 3975 3975 3975 standard deviation of the random effect term 0.8984 0.5370 0.5340 0.5303 variance reduction in percent -0.4 -1.9 -1.9 -1.9 log likelihood -1714.6 -1585.8 -1574.8 -1551.9 SOEP data 1992–2006; *** sign. P ≤ 0.001; ** sign. P ≤ 0.01; * sign. P ≤ 0.05. 1 Grand mean: standard deviation 0.9150; log likelihood -1,763.0. 2 The variance reduction (also r² aggregate) is a measurement of the quality of the model. It indicates how much of the variance can be explained through the introduction of the variable corresponding to the grand mean. Usually, no positive values should be found; however, sometimes it can happen that the variance is greater when additional variables are included. For the estimation of the variance reduction, the variance of the second level of the grand mean model is divided by the variance of the second level of the corresponding model. Then the value is subtracted from one: 1 – (var_2ed level g. m.)/(var_2ed level) (see Snijders and Bosker 1999). The variance reduction for the first model is estimated as follows: 1 – (0.9150²)/(0.8984²) = -0.4. We can see that with every new model, the variance was reduced by some percentage. 3 A test referring to the quality of the model is the log likelihood test, which measures the improvement between models. The log likelihood test statistic is estimated in the following way: =(2ln(L2) – 2ln(L1); degrees of freedom (m2 – m1).The log likelihood test indicates for all models, except Model 3, an improvement at the 0.001 significance level. Table 2 contains similar estimations for men. A first impression that arises from a comparison of Tables 1 and 2 is that for men, more variables show significant effects on migration. For men, the effect of education shows a pattern similar to that found for women. The results indicate that males with upper secondary education are more likely to move to the West than those without secondary education or with lower secondary education, and they are more likely to do so than males with tertiary education. Seniority also reduces the probability of migration for men. The effect of vocational training seems to differ, albeit not significantly, between men and women. For women, vocational training reduces the likelihood of migration, whereas for men, the effect displays a positive but not significant impact. The varying effect may also indicate that the segregation of men and women into different occupations may lead to different chances of finding a job in the West. IAB-Bibliothek 358 98 Reconsidering the effect of education on East-West migration in Germany Table 2: Determinants of the migration from East to West Germany; males, 1992–2006; application of random effect logit regressions men I II III IV individual characteristics reference group: lower secondary education intermediate secondary education 0.4343* 0.2892 0.3049 0.3206 (0.1826) (0.1896) (0.1928) (0.1935) upper secondary education 0.7166*** 0.8180*** 0.7581*** 0.7374** (0.2110) (0.2188) (0.2239) (0.2254) tertiary education 1.5075*** 0.7371* 0.4545 0.4383 (0.2830) (0.2863) (0.3092) (0.3109) vocational training -0.0248 0.1631 0.1369 0.2082 (0.1331) (0.1426) (0.1453) (0.1481) age 0.1023* 0.0857 0.0643 (0.0485) (0.0545) (0.0541) age squared -0.0021** -0.0018* -0.0015* (0.0007) (0.0007) (0.0007) partner -0.4389* -0.4510* -0.5324** (0.1738) (0.1760) (0.1701) reference group: no children in the household children younger than six -0.3551 -0.3540 -0.3480 (0.2283) (0.2280) (0.2282) children between six and 19 -0.7896** -0.7567** -0.6220* (0.2455) (0.2464) (0.2472) duration spent in the last company -0.0346* -0.0366* (0.0153) (0.0154) cumulative duration spent in unemployment -0.0700 -0.1330 (0.0682) (0.0711) gross income (in €100) 0.1505** 0.1324* (0.0534) (0.0545) reference group: unemployed employed part-time -0.8178 -0.7807 (0.6066) (0.6111) in apprenticeship -1.2594** -1.1290** (0.4109) (0.4164) employed full-time -1.0103* -0.8458* (0.3973) (0.4050) homeowner -0.9486*** (0.1625) regional characteristics regional income level 0.0220 (0.0145) unemployment rates -0.0034 (0.0031) distance to the old federal states (10 km) 0.0022* (0.0010) urban area 0.0478 (0.1554) 99 Results Chapter 3 men I II III IV n person-years 25260 25260 25260 25260 n person 3932 3932 3932 3932 Standard deviation of the random effect term 0.5497 0.5330 0.5344 0.5380 variance reduction10.0 -0.6 -0.6 -0.5 log likelihood2-1,422.5 -1,353.0 -1,345.6 -1,318.0 SOEP data 1992–2006; *** sign. P ≤ 0.001; ** sign. P ≤ 0.01; * sign. P ≤ 0.05. 1 Grand mean: standard deviation 0.5506; log likelihood -1,437.2. 2 The log likelihood test indicates for all models, except Model 3, an improvement at the 0.001 significance level. Additionally, some of the other variables show gender-specific differences. For men as for women, having a partner reduces the likelihood of migration, which is in line with the household economics of Mincer (1978). This suggests that for individuals in relationships, the monetary and non-monetary costs of the partner have to be considered when deciding on migration. In contrast to women, for men, the likelihood of migration rises by 14 percent with every €100 increase in income. This may seem puzzling at first, but it is in line with the results of other studies (Brücker and Trübswetter 2007; Hunt 2006; Windzio 2007). Moreover, a positive effect of income may indicate a positive selection of migrants (cf Brücker and Trübswetter 2007). For men, employment status also shows a strong and significant impact on migration. Men who are employed full-time or in an apprenticeship are less likely to leave East Germany than the unemployed or those employed on an irregular or marginal basis. For women, employment status was less significant. Regarding regional characteristics, the distance to the old federal states or West Berlin shows a positive effect for men but not for women. It appears that men who live close to the former border can easily commute to work. When the distance increases, the monetary as well as the non-monetary costs of commuting increase (cf Stutzer and Frey 2008), and migration may become the preferable option. In contrast to women, for men the regional income levels show no significant effect on migration. Even if they are mostly insignificant, the signs of the results are in line with those of Hunt (2006), who investigated the impact of regional characteristics on the basis of aggregated data. She found a negative effect of income level in the region of origin (here significant only for females) and a positive effect of unemployment rates, indicating that an increase in regional income levels (unemployment) reduces (increases) migration to the West. She also found a higher impact of source wages than that of source unemployment, which is the case here for females. To account for gender-specific differences, the data samples for men and women were pooled, and interaction terms for gender and all other covariates were generated and included in the estimations. In all models, the effect on relationship and the effect of being in an apprenticeship (but only in the first model) differed by IAB-Bibliothek 358 100 Reconsidering the effect of education on East-West migration in Germany gender at the 5 percent significance level. All other variables showed similar impacts for both sexes. Selection bias The results of the Heckman selection are displayed in Table 3. The selection models are estimated with probit regressions.19 For the selection equation, the probability of attending upper secondary education is used. The introduction of the inverse of the Mills ratio may introduce severe multicollinearity into the model; indeed, the multicollinearity is so high that the variables measuring general education have to be excluded.20 This reduces our interpretation possibilities, as the pure effect of education on migration and the influence of the unobserved characteristics can no longer be distinguished. To control for the longitudinal structure of the data, the error terms are clustered for individuals.21 Table 3: Determinants of the migration from East to West Germany; Heckman selection on upper secondary education; 1992–2006; application of probit models with clustered error terms women I women II men I men II individual characteristics reference group: lower secondary education intermediate secondary education -0.0366 0.1125 (0.0733) (0.0751) upper secondary education 0.1843* 0.2818 (0.0797) (0.0859) tertiary education 0.0993 0.1675 (0.1025) (0.1239) vocational training -0.0440 0.0831 (0.0537) (0.0567) age 0.0253 0.1949*** 0.0172 -0.0493 (0.0199) (0.0428) (0.0205) (0.0501) age squared -0.0006* -0.0028*** -0.0005 0.0003 (0.0003) (0.0006) (0.0003) (0.0006) partner -0.3878*** -0.3478*** -0.2022*** -0.0934 (0.0555) (0.1093) (0.0651) (0.1623) 19 I use one-step estimators and STATA 10 to obtain the results. Parental education is used as the single exclusion restriction. 20 This solution was also used by Li et al. (2000). The problems that might arise from such exclusion are discussed by Briggs (2004, p. 415). 21 The estimations of the logit (Table 1 and 2) and probit (Table 3) models can be compared when the coefficients of the probit model are multiplied by 1.7. However, the models used in Tables 1 and 2 are estimated with random effects logit models, whereas the results in Table 3 are based on probit models with clustered error terms; therefore, the results may deviate. Moreover, because the probit estimations are based on normal errors and the logit regression on logistic ones, the significance levels of the variables included in the estimation may also differ (see Long 1997). 101 Results Chapter 3 women I women II men I men II reference group: no children in the household children younger than six -0.0763 -0.3208* -0.1387 -0.2063 (0.0749) (0.1423) (0.0848) (0.1786) children between six and 19 -0.1350 -0.5436*** -0.2217* -0.2025 (0.0762) (0.1575) (0.0872) (0.1368) duration spent in the last company -0.0171* -0.0351** -0.0118 -0.0057 (0.0075) (0.0122) (0.0061) (0.0108) cumulative duration spent in unemployment -0.0022 -0.0098 -0.0452 -0.1472 (0.0163) (0.0412) (0.0270) (0.1134) gross income (in €100) 0.0162 -0.0369 0.0568** 0.1547** (0.0192) (0.0406) (0.0217) (0.0542) reference group: unemployed employed part-time -0.2320 -0.0376 -0.3356 -0.8567 (0.1466) (0.3138) (0.2403) (0.5416) in apprenticeship 0.0892 0.5066 -0.4732** -1.0162* (0.1327) (0.2644) (0.1646) (0.4151) employed full-time -0.1015 0.0640 -0.3743* -0.8034* (0.1387) (0.2909) (0.1620) (0.3719) homeowner -0.2768*** -0.2523*** -0.3647 *** -0.3538*** (0.0544) (0.1024) (0.0597) (0.1402) regional characteristics regional income level -0.0047*** -0.0023 -0.0013 0.0031 (0.0013) (0.0027) (0.0012) (0.0023) unemployment rates -0.0038 -0.0002 0.0075 0.0153 (0.0052) (0.0102) (0.0054) (0.0109) distance to the old federal states (10 km) 0.0006 -0.0007 0.0009* 0.0002 (0.0003) (0.0007) (0.0004) (0.0008) urban area 0.0798 0.1239 0.0277 -0.1675 (0.0551) (0.1055) (0.0598) (0.1214) inverse Mills ratio -0.3329* -0.1913 (0.1679) (0.1751) reference group: cohort 1970 and 1980 cohort 1960 -0.1438* 0.0883 (0.0696) (0.0754) cohort 1950 0.1004 0.4692*** (0.0687) (0.0746) cohort 1940 -0.2230** 0.4164*** (0.0858) (0.0816) cohort 1930 -0.5956*** 0.2121 (0.1359) (0.1218) parents have upper secondary education 0.7033*** 0.8535*** (0.0694) (0.0701) n persons 3975 3975 3932 3932 n person-years 26284 26284 25260 25260 n person-years censored 21566 20609 n person-years uncensored 4718 4651 log likelihood -1553.1 -12223.8 -1318.3 -11675.4 rho -0.3211* -0.1890 SOEP data 1992–2006; *** sign. P ≤ 0.001; ** sign. P ≤ 0.01; * sign. P ≤ 0.05. IAB-Bibliothek 358 102 Reconsidering the effect of education on East-West migration in Germany The lower part of Table 3 displays the variables used to generate the selection mechanisms. Women born after 1970 and women within the 1950–1959 cohort are the most likely to have completed upper secondary education. The effects are different for men; here, the cohorts born in the 1950s and 1960s are those with the highest likelihood of having completed upper secondary education.22 The impact of parents’ education shows a positive influence for men as well as for women. Children of parents who completed upper secondary education are more likely to complete upper secondary education themselves. We can see that rho23 and the inverse Mills ratio are both significant in Model 2 of Table 3; therefore, the selection model should indeed be used for females. Whereas the Heckman selection clearly reveals a selection process involved in the migration of women, the investigation of the effects for men indicate no such process. Rho and the Mills ratio in Model 4 of Table 3 remain insignificant. The estimation of the Heckman selection controls reveal that for women, unobserved characteristics influence the decision to migrate, as they have already influenced the decision to participate in education. The results also show that for women, both education and migration are driven by similar underlying mechanisms. The results thus support the common human capital theory view that both decisions are investments in human capital or its productivity. Using selection controls changes the significance and the magnitude of some of the variables. For example, the effects of children in the household become significant, which suggests that after controlling for the unobserved characteristics of females, the presence of children shows a negative impact on migration. 5 Conclusion The purpose of this article was to investigate migration from East to West Germany. The focus was on the influence of education on migration and on the selfselection processes involved in both education and migration decisions. Moreover, gender-specific differences between men and women were analyzed. Human capital, signaling, and segmentation theory were used to derive hypotheses on the influence of education on migration. Random effects hierarchical regressions and Heckman selection models were estimated. 22 The effect corresponds partly to the statistics released by the federal and state statistical offices (Statistische Ämter des Bundes und der Länder 2006, p. 31), demonstrating that in the new federal states, the percentage of men with upper secondary education rather than lower or intermediate secondary education is higher among individuals aged 55–64 than among those 25–34. The relation differs for women according to the statistic, as young women (25–34) in East as well as in West Germany are more likely to hold a secondary degree than those 55–64 years old. 23 Using the terminology from formula 1 and 2; rho refers to the correlation of εi and μi . 103 Conclusion Chapter 3 The following questions may be asked: Which hypotheses received empirical support? For women, hypotheses had to be tested based on the Heckman selection presented in Table 3; for men, the hierarchical regression models displayed in Table 2 are accurate. Both estimations show similar effects. For both men and women, the first hypothesis (H1) can be confirmed: longer durations in a company accompanied by higher seniority reduce the likelihood of migration. H3 also receives empirical support, as younger individuals are in fact more likely to migrate to the West.24 H2, derived from both the human capital and signaling theory and indicating that individuals with higher levels of specific knowledge should be more likely to migrate, cannot be confirmed. Individuals with upper secondary education are indeed more likely to leave East Germany than those with no, lower secondary, or intermediate secondary education. However, it is not individuals with tertiary education but those with upper secondary education who are the most likely to leave their region. Furthermore, the assumptions derived from the segmentation H6, suggesting that individuals employed in the professional labor market should be the most likely to migrate, and H4 (individuals employed in the internal labor market are less likely to relocate than individuals employed in the professional market) did not withstand the empirical test. For men, the impact of vocational training on migration is indeed positive, although it is weaker than the impact of general education and is also not significant. For women, completing vocational training reduces the likelihood of migration. Males with the lowest education and those without vocational training as well as those employed in the unskilled labor market are indeed the least mobile group. Thus far, the results are consistent with the outline of hypothesis five. Nevertheless, H5 cannot be confirmed for males because the effects on educational degrees are not always significant. For women, the effect differs. Females with intermediate secondary education are less mobile than the reference group with the lowest education, and vocational training shows a negative effect. Gender-specific differences are discussed below. Was it necessary to control for the selection on education? Yes, because the migration patterns of women are in fact defined by a selection on upper secondary education. For women, the results suggest that the same mechanisms drive participation in upper secondary education and migration and that differences in unobserved characteristics are the “true” reason behind the migration 24 We should, however, be very careful in the interpretation of the age effect in the selection model beyond its sign. Usually, variables included in the selection model as well as the second estimation have to be interpreted by taking the coefficients in the selection model as well as in the estimation of the effect of interest into account (Siegelman and Zang 1999). Here, age is measured in different ways in both models, which makes interpretation difficult. IAB-Bibliothek 358 110 Reconsidering the effect of education on East-West migration in Germany Table 6: Transitions between educational degrees, men and women no answer no degree lower secondary degree intermediate secondary degree upper secondary degree tertiary degree no answer 52.38 1.46 7.00 17.23 14.69 7.23 no degree 0.23 77.35 8.92 9.38 3.43 0.69 lower secondary degree 0.00 0.27 97.51 1.83 0.27 0.11 intermediate secondary degree 0.02 0.03 0.52 98.35 0.63 0.45 upper secondary degree 0.00 0.07 0.25 1.36 95.79 2.52 tertiary degree 0.15 0.15 0.58 5.44 17.40 76.29 Table 7: Distribution of within and between variance; intermediate secondary degree, men and women intermediate secondary degree mean S.E. min max overall variance 0.5347 0.4988 0 1 between variance 0.4854 within variance 0.1406 Table 8: Distribution of within and between variance; upper secondary degree, men and women upper secondary degree mean S.E. min max overall variance 0.1962 0.3971 0 1 between variance 0.3813 within variance 0.1248 Table 9: Distribution of within and between variance; tertiary degree, men and women tertiary secondary degree mean S.E. min max overall variance 0.2338 0.4233 0 1 between variance 0.4003 within variance 0.0829 Chapter IV Does migration make you happy? The influence of migration on subjective well-being1 1 I would like to thank my supervisors, Herbert Brücker and Hans-Peter Blossfeld, for their advice and support. I am grateful to Michael Tåhlin, Ruud Muffles, Jennifer Hunt, Thomas Rhein, Dominik Morbitzer, and Journt Mandemakers for their comments. This research was conducted in part during a research visit at the Swedish Institute for Social Research (SOFI) and in part during a research visit at the Department of Sociology at the University of Tilburg, whose hospitality is gratefully acknowledged. I would also like to thank the TransEurope Research Network for their generous financial support during the research visit to the SOFI. This chapter is published in Journal of Social Research & Policy 2 (Issue 2 December 2013), p. 73–92. 113 Introduction Chapter 4 Abstract In the field of neoclassical economics, migrants are expected to move to improve their economic situations, but what are the effects of moving on the subjective wellbeing (SWB) of migrants? Using longitudinal data from the German Socio-Economic Panel Study (SOEP) (1990–2007), I investigate the influence of migration from Eastern to Western Germany on SWB. The hypotheses in this study are derived from neoclassical economics and from the psychology literature. Following the rational choice framework, I expect that migration improves SWB in the long term. Fixedeffects models distinguish between the effects of unobserved heterogeneity, such as varying personality traits, and migration on SWB. The results reveal that migration has a positive, long-term effect on SWB. In addition, the favorable labor market conditions in Western Germany account for the increasing SWB that is reported by male migrants but does not account for that reported by female migrants. 1 Introduction According to neoclassical economics, migrations are typically financially motivated. Migrants generally improve their economic situations after moving to a new location, but they also experience the non-monetary losses of family and friendship networks (see, e.g., Borjas, 1987; Chiswick, 1999; Sjaastad, 1962).2 Can such financial gains compensate for the possible loss of friendship and family networks? How does migration affect the SWB of migrants? Are migrants happier after they move? Few studies have addressed these questions (De Jong et al., 2002) because of the unavailability of data pertaining to the influence of migration on SWB. Ideally, longitudinal data containing information on SWB before and after migration are necessary; however, such data rarely exist. In the country of destination migrants may participate in surveys only after they move, while no information from before are available. In the country of origin migrants drop out from the data sets after the relocation and the data contains no information after the in their countries of origin. Therefore, most studies rely on cross-sectional data that are collected after such moves, and these data contain no previous information on SWB. Hence, migrants are compared with natives (Amit, 2010; for studies on older immigrants, see Amit and Litwin, 2009; Ba�lt¸a�escu, 2007; Bertram, 2010; for studies on second-generation immigrants, see Neto, 1995; Safi, 2010). Other studies ask the respondents directly about their SWB before and after a move (De Jong et al., 2002; Lundholm and 2 There are additional reasons to migrate; for example, some individuals migrate to improve their quality of life (see, e.g., Benson and O’Reilly, 2009). IAB-Bibliothek 358 114 Does migration make you happy? The influence of migration on subjective well-being Malmberg, 2006). Problems arise from both of these designs. In the first type of research, it is not possible to distinguish whether a deviation in the levels of SWB is caused by migration or by general differences in the level of SWB of migrants and natives. In the second type of research design, we cannot be certain that the indicated improvements are indeed objective. These problems emphasize the importance of using longitudinal data to conduct research on SWB, as these issues raise doubts regarding the reliability of cross-sectional samples for this subject. Based on longitudinal data from the German Socio-Economic Panel Study (SOEP), this study investigates the influence of migration on SWB with regard to relocation from Eastern to Western Germany after the fall of the wall. After the collapse of the Eastern Bloc and the fall of the Berlin Wall on November 9, 1989, the former German Democratic Republic (GDR) found itself in a unique position (Mayer, 2006). After the first free election in March 1990, the reconstruction of the nation was controlled by the government of the Federal Republic of Germany (FRG). Despite sharing a common past and language, East and West Germany developed in different directions after the Second World War, and by 1989, the two former nations had as many differences as similarities. West Germany developed a market economy and a conservative-corporatist welfare regime, whereas East Germany adopted a socialist system with a planned economy. The socialist system was never able to compete with its capitalist counterpart, and per capita income in the East lagged behind that of Western standards. Even today, 20 years after reunification, the Eastern German labor market demonstrates weaker performance than the Western market, and Eastern Germans face large incentives to migrate (Melzer, 2011). The consequence of these conditions was a substantial and permanent migration from Eastern to Western Germany. Compared with the population level in 19883, the former GDR had lost 4.3 percent of its population by 1992, 7.9 percent by 1995, 10.7 percent by 2000 and 14.1 percent by 2006. The reunification of Germany, which several economists have called a “natural” experiment, provides a unique opportunity to study the influence of migration on SWB based on longitudinal data containing information for the periods before and after moves. Although previous research on the effects of migration or regional mobility on SWB provides initial insight into the topic and deepens our understanding of the process, the existing literature is limited in several aspects. 3 For the population levels of the GDR, see Staatliche Zentralverwaltung für Statistik (1989, p. 335). For more recent figures, see the Federal Statistical Office ( Statistisches Bundesamt 2011, p. 53 ff). The figures presented do not include East Berlin, as it is not possible to differentiate between East and West Berlin after 2000. By that year, eastern Germany including Berlin had lost 10.1 percent of its former population. 115 Introduction Chapter 4 First, previous studies are based on cross-sectional data. When analyzing crosssectional data, one cannot distinguish between the effects of unobserved heterogeneity, such as varying personality traits, and the effects of migration on SWB. However, previous research indicates that personality influences SWB (e.g., Diener et al., 1999).4 Second, when asked to compare two situations directly, most people report that their lives are improved after migration (Hagerty, 2003), and migrants do not differ from the general population in this regard (Scott and Scott, 1989). Therefore, migrants may report higher SWB after their moves to avoid acknowledging any cognitive dissonance (Festinger, 1957). Using cross-sectional data, one cannot clearly determine the causality between the described factors and SWB (Frey and Stutzer, 2005). Although it is for example clear that gender influences satisfaction, other factors, such as marriage or migration, may show a reverse causality. Therefore, cross-sectional studies are unable to determine whether migrants are more satisfied than the general population, whether the characteristics that make these individuals more likely to relocate also make them happier, or whether their greater satisfaction actually results from their relocation. These problems emphasize the importance of using longitudinal data to conduct research on SWB, as these issues raise doubts regarding the reliability of cross-sectional samples for this subject. Third, few studies link the effects of migration on SWB within an explanatory theoretical framework. The hypotheses that are tested are primarily derived ad hoc from previous findings (e.g., Lundholm and Malmberg, 2006). The few contributions to the literature that do provide a theoretical background concentrate on specific aspects. In some cases, the theoretical framework aims to describe the integration of migrants and to compare migrants and natives using a variety of assimilation models (Safi, 2010) or to discuss the integration process using concepts such as the social capital framework of Bourdieu (1986) (see: Amit, 2010; Amit and Litwin, 2009). Other authors have used theoretical concepts to explain the situations of migrants before and after migration. For example, Lu (2002) followed the housing career thesis in analyzing residential mobility. Only De Jong et al. (2002) integrated the question regarding the influence of migration on SWB with a theory that is typically used to analyze migration. In accordance with Sjaastad (1962), De Jong et al. (2002) treat migrations as investments in the productivity of individuals. Fourth, some features of migration have not been addressed at all. Questions regarding the influence of regional characteristics or the length of the stay in a new host region on SWB remain unanswered. However, studies that have been conducted at the macro level show the importance of regional income or unemployment levels 4 For example, neurotic individuals may report lower satisfaction than those who are not neurotic (Diener et al., 1999). IAB-Bibliothek 358 116 Does migration make you happy? The influence of migration on subjective well-being for individual satisfaction (for the USA, see Alesina et al., 2004; for Europe, see Di Tella et al., 2001; for Germany, see Easterlin and Plagnol, 2008). This study attempts to fill these gaps and to analyze the influence of migration from Eastern to Western Germany on SWB. Hypotheses that describe the relationship between migration and subjective well-being (SWB) are derived from human capital theory and psychology approaches. More precisely, this study aims to answer the following questions: How does migration influence SWB? How do changes in SWB after migration (if such changes exist) develop over time? Are there differences in SWB changes among different migrating groups or between men and women? What is the influence of the conditions of the regional labor markets on SWB? The reunification of Germany, which several economists have called a “natural” experiment, provides a unique opportunity to study the influence of migration on SWB based on longitudinal data containing information from before and after migration. The empirical investigations that are presented in this paper are based on the German Socio-Economic Panel Study (SOEP), and on waves from 1992 to 2006. Hereby information on all individuals who migrated from Eastern to Western Germany between 1990 and 2007 are included. Fixed-effects models are used to determine the effects of unobserved heterogeneity and migration on SWB, based on the assumption that unobserved characteristics tend to be stable over time. The use of fixed-effects models ensures that the effect of migration on SWB is causal rather than based on selection. This approach verifies that “happy” individuals are not those who typically migrate but that migration does indeed affect SWB. In the analyses, I control for individual and regional labor market characteristics in Eastern and Western Germany. I distinguish between migrants and persons who returned to East Germany after relocation to the West. Variables that indicate the amount of time that individuals have lived in the West account for the influence of time on changes in SWB. The analyses are conducted separately for men and women, as previous studies have found gender-specific differences in the influence of migration (e.g., Frijters et al., 2004). 2 Migration and subjective well-being Research on the SWB of migrants has different goals in sociology and economic contexts.5 Sociological research in this field focuses on the integration process of migrants and dates back to 1928 to the research of Park (1928) on ‘marginal man’ and the uprooted (Handlin, 1951). In contrast, the first economic research 5 For a general overview of the factors that influence SWB, see the work of Dolan et al. (2008). For an overview validating the theoretical importance and the measurement of SWB, see the studies of Blanchflower and Oswald (2004a), Di Talla and MacCulloch (2006), Frey and Stutzer (2002), and Kahneman and Krueger (2006). 117 Migration and subjective well-being Chapter 4 to address SWB in the context of migration was motivated by differences in SWB between countries.6 In fact, the average happiness of individuals differs among countries; individuals from Western Europe and the USA score higher on wellbeing scales than those from Eastern Europe (Blanchflower and Oswald, 2008).7 However, as Bartram (2010) indicated, it would be an ecological fallacy to conclude that migration from countries with lower levels of SWB to those with higher levels of SWB would increase happiness. Analyzing five Nordic countries (Denmark, Finland, Iceland, Norway, and Sweden), Lundholm and Malmberg (2006) revealed a positive relationship between residential mobility and SWB. Among the few people who were less satisfied after a move (8 percent), singles were overrepresented. In turn, individuals who reported greater satisfaction with their social lives after a move showed the highest level of general satisfaction, as social life has the largest effect on overall satisfaction. The authors concluded that migration in Nordic countries is not a trade-off situation in which social and environmental cuts are accepted in return for higher incomes; rather, they found that relocations serve as opportunities to obtain preferred types of housing, as found by Lu (2002) in an analysis of the USA. Other studies, such as a study of repeated, temporary and permanent migration in Thailand, reveal a mixed influence of migration on life satisfaction (De Jong et al., 2002). Mobility was found to increase, decrease and not to influence life satisfaction; approximately one-third of the migrants accounted for each group. However, job satisfaction increased after migration. The study reveals a negative relationship between life satisfaction and education, and this relationship was explained by the unrealistic expectations of highly educated individuals regarding living conditions after their moves. Finally, previous research in this field that concentrated on depression, which could be understood as the opposite of SWB, reveals that residential mobility increases the likelihood of depression, especially for women (Magdol, 2002). The most recent study that has applied a different methodology focuses on the life satisfaction of immigrants and natives in the USA. The results that were obtained by Bartram (2010) are consistent with those of other studies that have compared the SWB of immigrants and natives and found lower life satisfaction among immigrants than natives (e.g., Amit, 2010; Amit and Litwin, 2009; Ba�lt¸a�escu, 2007; Neto, 1995; Safi, 2010). In comparisons between countries of origin, immigrants from poorer countries show lower levels of life satisfaction, whereas immigrants from Europe or Canada do not differ significantly from natives. Moreover, the 6 In general, economic research on SWB dates back to Richard Easterlin (1974) who found that individuals do not report increased SWB with increased personal income. 7 Moreover, at the macro level, life satisfaction is an excellent predictor of international migration (Blanchflower and Oswald, 2008). IAB-Bibliothek 358 118 Does migration make you happy? The influence of migration on subjective well-being satisfaction of immigrants from poorer countries is defined to a greater degree by absolute income. These immigrants constitute a group with modest earnings and are therefore more frustrated than natives with regard to their inability to obtain higher incomes (Bartram, 2010). Thus far, only one study has included a variable measuring life satisfaction before and after migration based on longitudinal data. Frijters et al. (2004) investigated determinants of life satisfaction in Eastern and Western Germany and also included measurements of the influence of migration on life satisfaction. Using ordered logit fixed-effects models, the authors found a positive effect of migration from Eastern to Western Germany and a negative effect of relocation from West to East for men only (Frijters et al., 2004). This study provides a general overview of the influence of major life events on SWB rather than on the effects of migration. Differences in SWB between groups of migrants or over time were not examined. Moreover, this study was based on a rather short period during which levels of life satisfaction in Eastern and Western Germany were still converging. 3 Theoretical considerations 3.1 Subjective well-being of migrants According to neoclassical economics, migration represents a risky investment through the allocation of human capital to increase productivity. Individuals maximize their utility by choosing the most beneficial location. In this respect, migrants must be willing to tolerate present costs to obtain future benefits Borjas, 1987; Chiswick, 1999; Chiswick, 1978; Sjaastad, 1962). Individuals compare the costs and benefits of migration. In this sense, migration is equivalent to any other investment in human capital, such as schooling or on-the-job training. The costs of migration are both monetary and non-monetary (Sjaasjad, 1962). The monetary costs of migration are primarily transportation costs, whereas the non-monetary costs are more substantial and include the loss of location-specific human capital, such as the loss of family and friendship networks (DaVanzo, 1983), and opportunity costs (Sjaasjad, 1962). However, the benefits of migration can also be non-monetary. Therefore, individuals may migrate to a better climate, for family-related reasons or to improve life quality and they may accept the financial disadvantages in order to live in a new location (see, e.g., Benson and O’Reilly, 2009). When people make their decisions with sufficient information and without unrealistic expectations by considering both monetary and non-monetary costs and benefits, only those who profit from a migration in a subjective sense will migrate (Ziegler and Britton, 1981). Thus, as De Jong et al. (2002) claimed, 119 Theoretical considerations Chapter 4 migrants are likely to report higher subjective well-being (SWB) after a move than before a move (hypothesis 1). Moreover, the increase in SWB after a move should be enduring8 (hypothesis 2) because migration (similar to other career investments) is a long-term investment. An alternative view of the long-term development of SWB is provided by the psychological literature: following an initial increase in SWB in the period immediately after migration, mechanisms that include adaptation, aspiration and comparison reduce SWB in later periods. First, as individuals adapt to repeated stimuli (Scitovsky, 1992), they should adapt quickly to their improved living standards after migration and thus experience a decline in SWB. Second, obtaining higher incomes may trigger even higher aspirations regarding earnings and economic status (Stutzer, 2003; van Praag, 1993). Third, individuals change their reference categories after receiving an increase in income and thereafter compare themselves to even wealthier persons (Venhoven, 1991). Therefore, migrants who relocated from Eastern to Western Germany should change their comparison group from Eastern to wealthier Western Germans. Alternative hypothesis 2a states as follows: Following an initial increase in SWB after migration, migrants are likely to report decreasing SWB in later periods. 3.2 Group differences The costs and benefits of migration depend on the education of a migrant and the income distributions in the locations of origin and destination (Borjas, 1987; Chiswick, 1999; Chiswick, 1978). Highly educated individuals should profit the most from relocations from Eastern to Western Germany because of the higher marginal value of their education in the West (c.f. Melzer, 2011). Moreover, highly educated individuals also have regionally broader networks (Massey et al., 1998). These networks render these individuals as less regionally dependent and reduce their migration costs. Low costs combined with high gains should both motivate more highly educated individuals to migrate and increase their profits and SWB after such a move. In addition, highly educated individuals are more likely to be able to gather the necessary information for migration and to weigh the gains and losses appropriately. Hence, these individuals are more likely to make wellconsidered decisions and to avoid disappointment. Therefore, highly educated individuals should report more positive changes in SWB than migrants with lower educational levels after a move (hypothesis 3). 8 According to neoclassical economics, such decisions are based on the unrealistic assumption that individuals maximize their utility for life. Therefore, “enduring” indicates an improvement in SWB over a lifetime. IAB-Bibliothek 358 126 Does migration make you happy? The influence of migration on subjective well-being Table 1: Consequences of migration from Eastern to Western Germany on the SWB of men: fixed-effects regressions based on SOEP data for the 1992–2006 period men I II III IV V VI individual characteristics migrated from Eastern to Western Germany 0.401*** 0.291** 0.294* 0.092 (0.100) (0.097) (0.124) (0.175) returned to Eastern Germany 0.071 0.052 -0.266 0.059 0.039 -0.081 (0.242) (0.230) (0.228) (0.230) (0.233) (0.259) migrated less than two years ago 0.315** 0.142 (0.106) (0.153) migrated two to four years ago 0.323** 0.136 (0.114) (0.163) migrated four to six years ago 0.395** 0.199 (0.139) (0.200) migrated six to nine years ago 0.181 -0.029 (0.168) (0.221) migrated more than ten years ago 0.389* 0.155 (0.174) (0.232) age -0.061*** -0.123*** -0.124*** -0.123*** -0.132*** -0.123*** (0.013) (0.014) (0.014) (0.014) (0.017) (0.017) age squared 0.001*** 0.002*** 0.002*** 0.002*** 0.002*** 0.002*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) health 0.423*** 0.409*** 0.409*** 0.408*** 0.408*** 0.408*** (0.018) (0.018) (0.018) (0.018) (0.018) (0.018) found a partner in the past year 0.206** 0.075 0.071 0.088 0.085 0.077 (0.070) (0.072) (0.072) (0.074) (0.074) (0.072) married in the past year 0.245** 0.113 0.110 0.098 0.095 0.112 (0.092) (0.093) (0.093) (0.095) (0.095) (0.093) divorced in the past year -0.176 -0.164 -0.167 -0.169 -0.170 -0.168 (0.121) (0.116) (0.116) (0.116) (0.116) (0.116) spouse died in the past year 0.038 -0.016 -0.019 -0.025 -0.034 -0.025 (0.228) (0.232) (0.232) (0.233) (0.232) (0.232) has children 0.136** 0.177*** 0.177*** 0.177*** 0.179*** 0.179*** (0.042) (0.041) (0.041) (0.041) (0.041) (0.041) became unemployed -0.390*** -0.391*** -0.390*** -0.389*** -0.395*** (0.077) (0.077) (0.077) (0.077) (0.077) began part-time work -0.382*** -0.382*** -0.380*** -0.375*** -0.376*** (0.099) (0.099) (0.098) (0.098) (0.099) began vocational training -0.236*** -0.237*** -0.235*** -0.234*** -0.211*** (0.062) (0.062) (0.062) (0.062) (0.063) household income 0.023*** 0.023*** 0.023*** 0.022*** 0.022*** (0.004) (0.004) (0.004) (0.004) (0.004) hours worked 0.003 0.003 0.003 0.003 0.003 (0.001) (0.001) (0.001) (0.001) (0.001) 127Chapter 4 Results men I II III IV V VI interaction terms tertiary education* migrated -0.120 -0.102 (0.189) (0.189) partner* migrated -0.101 -0.101 (0.152) (0.150) married* migrated 0.148 0.133 (0.157) (0.161) regional characteristics income level 0.007 0.006 (0.005) (0.005) unemployment rate -0.004 -0.004 (0.005) (0.005) city with more than 100,000 residents -0.039 -0.032 (0.078) (0.079) person-years 27457 27457 27457 27457 27457 27457 persons 4099 4099 4099 4099 4099 4099 r-square overall 0.145 0.171 0.172 0.171 0.188 0.184 r-square within 0.060 0.085 0.085 0.085 0.085 0.086 r-square between 0.175 0.204 0.204 0.204 0.230 0.225 rho 0.525 0.520 0.520 0.520 0.513 0.515 Robust standard errors are used; year dummies are included; * p < 0.05, ** p < 0.01, *** p < 0.001. More importantly, the change in SWB was enduring (Model 3). Male migrants reported higher levels of life satisfaction even six years after migration. However, after labor market characteristics are controlled in Models 5 and 6, the effect of migration is no longer significant for men. Therefore, the increased SWB after migration can be explained by the superior labor market characteristics of Western Germany. Men appear to have benefited, in terms of SWB, from higher income levels and more secure employment situations in Western Germany. Neither highly educated men nor those who moved to Western Germany with a partner (married or otherwise) differ from other groups of migrants (Model 4). The SWB patterns that were found in this study are generally similar for men and women. Migration increased SWB for both sexes. The effect endured for at least 6 years, and no group differences can be found, although the positive effect of migration on SWB appears to have been slightly larger and more stable for females. In addition, the effect of return migration was positive but never significant. Few differences between men and women can be found. The most important difference is that the influence of migration on the SWB of women remains stable even when regional features are controlled (see Table 2 and Models 5 and 6). There were differences between men and women with regard to employment status. Both men and women who became unemployed, began apprenticeships or began part-time work were less satisfied. Initially, this pattern showed no gender-specific differences. However, a comparison of the effects of migration and employment IAB-Bibliothek 358 128 Does migration make you happy? The influence of migration on subjective well-being status on SWB reveals differences between men and women. Next to the effect of health, unemployment or part-time work shows the strongest effect on SWB for men, whereas migration shows the strongest effect on SWB for women. Table 2: Consequences of migration from Eastern to Western Germany on the SWB of women: fixed-effects regressions based on SOEP data for the 1992–2006 period women I II III IV V VI individual characteristics migrated from Eastern to Western Germany 0.472*** 0.416*** 0.412** 0.500*** (0.095) (0.093) (0.135) (0.142) returned to Eastern Germany 0.335 0.348 -0.024 0.339 0.355 -0.094 (0.206) (0.199) (0.202) (0.197) (0.199) (0.206) migrated less than two years ago 0.472*** 0.562*** (0.094) (0.106) migrated two to four years ago 0.351** 0.450*** (0.123) (0.135) migrated four to six years ago 0.475*** 0.574*** (0.139) (0.146) migrated six to nine years ago 0.190 0.286 (0.169) (0.173) migrated more than ten years ago 0.386* 0.479** (0.178) (0.183) age -0.040** -0.095*** -0.094*** -0.095*** -0.117*** -0.109*** (0.014) (0.014) (0.014) (0.014) (0.018) (0.018) age squared 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) health 0.402*** 0.403*** 0.402*** 0.403*** 0.403*** 0.403*** (0.017) (0.017) (0.017) (0.017) (0.017) (0.017) found a partner in the past year 0.234*** 0.112 0.106 0.118 0.122 0.123 (0.062) (0.065) (0.066) (0.069) (0.069) (0.066) married in the past year 0.158 0.061 0.058 0.068 0.074 0.073 (0.084) (0.087) (0.086) (0.088) (0.088) (0.086) divorced in the past year -0.117 -0.085 -0.087 -0.083 -0.078 -0.079 (0.121) (0.119) (0.119) (0.119) (0.119) (0.119) spouse died in the past year -0.001 -0.021 -0.025 -0.015 -0.009 -0.014 (0.160) (0.159) (0.160) (0.160) (0.160) (0.160) has children 0.113** 0.182*** 0.182*** 0.182*** 0.183*** 0.188*** (0.039) (0.039) (0.039) (0.039) (0.039) (0.039) became unemployed -0.298*** -0.300*** -0.298*** -0.306*** -0.313*** (0.071) (0.071) (0.071) (0.071) (0.071) began part-time work -0.135** -0.134** -0.135** -0.135** -0.140** (0.043) (0.043) (0.043) (0.043) (0.043) began vocational training -0.258*** -0.259*** -0.258*** -0.260*** -0.239*** (0.062) (0.062) (0.062) (0.062) (0.064) household income 0.028*** 0.028*** 0.028*** 0.028*** 0.028*** (0.004) (0.005) (0.005) (0.005) (0.005) hours worked 0.001 0.001 0.001 0.001 0.001 (0.002) (0.002) (0.002) (0.002) (0.002) 129Chapter 4 Results women I II III IV V VI interaction terms tertiary education* migrated 0.134 0.105 (0.214) (0.213) partner* migrated -0.032 -0.040 (0.160) (0.159) married* migrated -0.050 -0.035 (0.175) (0.176) regional characteristics income level 0.009 0.008 (0.006) (0.006) unemployment rate 0.013** 0.013** (0.005) (0.005) city with more than 100,000 residents -0.179* -0.173* (0.087) (0.087) person-years 28908 28908 28908 28908 28908 28908 persons 4134 4134 4134 4134 4134 4134 r-square overall 0.130 0.154 0.154 0.154 0.159 0.157 r-square within 0.051 0.067 0.067 0.067 0.067 0.068 r-square between 0.180 0.199 0.197 0.199 0.201 0.199 rho 0.523 0.518 0.518 0.518 0.511 0.511 Robust standard errors are used; year dummies are included; * p < 0.05, ** p < 0.01, *** p < 0.001. I conducted several tests to account for possible sources of selectivity. First, I tested whether persons who are removed from the data are indeed a random group and not selected among those who are less satisfied (see the additional variable test used by Wooldridge, 2001: 581). Two of the three constructed tests support the view that individuals are removed from the data randomly. Second, I constructed various robustness tests to analyze whether the various methods of accounting for migrants and returnees influence the analyses.12 In this situation, the main problem is that returnees might be selected according to their satisfaction. East- West migrants who are particularly dissatisfied in the West might return to the East. The effect of migration remains stable regardless of the specification used (the results are not presented here). Finally, I estimated the Heckman selection controls with a reduced set of variables following the method of Heckman and Smith (1996). The results indicate that the unobserved characteristics based on which persons are selected for migration account for approximately one-sixth of the effects of migration on SWB (the results are presented in the appendix). The results indicate the importance of using fixed-effects estimations and controlling for unobserved characteristics. 12 Possible influences were investigated, for example, by including previously excluded returnees in the sample, adding returnees to migrants in West Germany or treating these migrants as a separate group in the final models. IAB-Bibliothek 358 130 Does migration make you happy? The influence of migration on subjective well-being 6 Conclusion The purpose of this study was to investigate the influence of migration on SWB. The hypotheses in this study regarding the effect of migration on SWB were derived from neoclassical economic concepts such as the human capital theory and from concepts developed in the psychology literature. The main prediction derived from the human capital framework is that migrants should be more satisfied after a move. The estimations are based on longitudinal data (SOEP 1990–2007), which include information pertaining to individuals living in, moving from, and returning to Eastern Germany. Fixed-effects hierarchical models were used to distinguish the effects of personality and mobility on SWB. East-West migration in Germany has had a positive influence on SWB. Therefore, hypothesis 1 (migrants should report higher SWB after their moves) is verified. Although the favorable conditions in the Western German labor market, such as higher regional income levels, account for the increase in the reported SWB of men, the same result was not observed for women. Previous studies have shown that men and women experience increased SWB as a result of different factors. For example, Fandrem et al. (2009) showed that young women gain greater satisfaction from housing. For example, the quality of housing may differ between Eastern and Western Germany, and women may have experienced greater SWB in the West because of the superior quality of housing. Other reasons for the high and stable increases in the SWB of women after migration might have been associated with the structure of the labor market in West Germany, which may have been especially beneficial for women. Research addressing mobility in urban municipalities states that women are more likely to relocate to larger cities because the men in these cities have higher education levels and earnings and are thus more attractive to women (Edlund, 2005). Wages in Western Germany tend to be higher than wages in Eastern Germany. Therefore, westward migration may be driven by a mechanism that is similar to that which motivates migration to cities. Single women who migrate to Western Germany or to larger cities may profit from increases in their own wages and those of their potential partners (cf. Edlund, 2005). For female migrants who live in partnerships, a different mechanism could account for the increased SWB following migration. In this context, the research on over-qualification might be insightful. For example, Büchel (2000) showed that married women who live in more highly populated municipalities are less likely to work in jobs for which they are overqualified (Büchel, 2000). If Eastern and Western Germany differ regarding the density of population and the density or quality of jobs that are available (e.g., high-quality jobs in East Germany are more scarce) or if couples migrate from rural areas to urban areas, then women who migrate with partners or spouses may be more satisfied because of the 131Chapter 4 Conclusion superior job opportunities that are available. If this explanation holds true, then the regional control variables that were used may not capture the complete effect of the superior job opportunities that are available in West Germany. The positive effect of migration on SWB was found a maximum of six years after relocation for men and a maximum of ten years after relocation for women. Hypothesis 2 (the increases in SWB after a move are enduring) can be confirmed only for women. For men, the situation is more complex. On the one hand the effect can be interpreted as enduring; the SWB of men remained higher after six years, which is longer than individuals usually need to adapt to new situations. For example, the existing research shows that individuals typically need approximately 3 years to adapt to widowhood or marriage, two years to recover from layoffs and one year to adapt to a divorce (e.g., Clark et al., 2008). On the other hand, the effect of migration on SWB for East German male migrants who relocated to West Germany declines in magnitude after six years and does not remain significant. This result may indicate that an adaptation process was occurring and that male migrants were affected by the “hedonic treadmill”; and that even a significant change in living conditions, such as the changes associated with relocation, may not increase SWB indefinitely. Therefore, neither hypothesis 2 nor the alternative hypothesis 2a can be confirmed for men. Further research best one based on international data is necessary to clarify the process that affects the SWB of men after relocation. Interestingly, the results for women contradict the predictions of psychologists regarding adaptation, aspiration and the comparisons used in alternative hypothesis 2a (migrants should report decreasing SWB in periods following migration) and earlier research (cf. Brickman et al., 1978). However, a comparison of the results to more recent research provides a more harmonious view of the situation. According to recent research, individuals do not completely adapt to non-monetary life events, such as marriage, divorce, disability (Di Tella et al., 2007; Easterlin, 2003; Lucas et al., 2003) or migration. Neither highly educated migrants nor those who move with a partner reveal different levels of SWB (thus, hypothesis 3 must be rejected, but hypothesis 4 is confirmed). This study provided new information on migration using longitudinal data. However, new questions also arise. One of the most important questions concerns whether the positive, long-term effect that was found in this study can be confirmed for international migration. For example, does this pattern apply to people who relocate to a society with an entirely different culture or economic situation or to countries in which other languages are spoken? Moreover, it would be interesting to determine whether people who relocate to a society with a different ethnic majority also show such high increases in SWB. However, new data sources will be required to answer such questions. Finally, this study compared the SWB of IAB-Bibliothek 358 132 Does migration make you happy? The influence of migration on subjective well-being East-West migrants before and after their moves with the remaining population of the country of origin; the link between migrants and individuals from a country of destination is still absent and should be analyzed in future research. Acknowledgements I thank my supervisors Herbert Brücker and Hans-Peter Blossfeld for their advice and support. I am grateful to Michael Tåhlin, Ruud Muffles, Jennifer Hunt, Thomas Rhein, Dominik Morbitzer, Journt Mandemakers for their comments. This research was partly concluded during a research visit at the Swedish Institute for Social Research (SOFI) and partly during a research visit at the Department of Sociology at the University of Tilburg, whose hospitality is grateful acknowledged. I also thank the TransEurope Research Network for their generous financial support during the research visit at the SOFI. 7 References Alesina, A., R. Di Tella, and R. MacCulloch. 2004. “Inequality and happiness: Are Europeans and Americans different?” Journal of Public Economics 88: 2009–42. Amit, K. 2010. “Determinants of Life Satisfaction Among Immigrants from Western Countries and from the FSU in Israel.” Social Indicators Research 96: 515–534. Amit, K., and H. Litwin. 2009. “The Subjective Well-Being of Immigrants Aged 50 and Older in Israel.” Social indicators Research Online first DOI 10.1007/s11205- 009-9519-5. Ba�lt¸a�escu, S. 2007. “Central and Eastern Europeans Migration’ Subjective Quality of Life. A Comparative Study.” Journal of Identity and Migration Studies 1: 67–81. Bartram, D. 2010. “Economic Migration and Happiness: Comparing Immigrants’ and Natives’ Happiness Gains from Income.” Available at SSRN: http://ssrn.com/ abstract=1590265. Benson, Michaela and Karen O’Reilly. 2009. “Migration and the search for a better way of life: a critical exploration of lifestyle migration.” The Sociological Review 57: 609–625. Bielby, W. T., and D. D. Bielby. 1992. “I Will Follow Him – Family Ties, Gender- Role Beliefs, and Reluctance to Relocate for a Better Job.” American Journal of Sociology 97: 1241–1267. Blanchflower, D. G., and A. J. Oswald. 2004a. “Money, Sex and Happiness: An Empirical Study.” Scandinavian Journal of Economics 106: 393–415. Blanchflower, D. G., and A. J. Oswald. 2004b. “Well-being over time in Britain and the USA.” Journal of Public Economics 88: 1359–1386. 133Chapter 4 References Blanchflower, D. G., and A. J. Oswald. 2008. “Is well-being U-shaped over the life cycle?” Social Science & Medicine 66: 1733–1749. Bourdieu, P. 1986. “The forms of capital.” Pp. 241–258 in Handbook of theory and research for the sociology of education, edited by J. Richardson. New York: Greenwood Press. Borjas, G. J. 1987. “Self-Selection and the Earnings of Immigrants.” The American Economic Review 77: 531553. Brickman, P., D. Coates, and R. Janoff-Bullman. 1978. “Lottery Winners and Accident Victims: Is Happiness Relative?” Journal of Personality and Social Psychology 36: 917–27. Büchel, F. 2000. “Tied Movers, Tied Stayers: The Higher Risk of Overqualification among Married Women in West Germany.” in Gender and the Labour Market. Econometric Evidence of Obstacles to Achieving Gender Equality edited by S. S. Gustafsson and D. E. Meulders. New York Macmillian. Chiswick, B. R. 1999. “Are Immigrants Favorably Self-Selected?” The American Economic Review 89: 181–185. Chiswick, B. R. 1978. “The Effect of Americanization on the Earnings of Foreignborn Men.” Journal of Political Economy 86: 897–921. Clark, A. E., E. Diener, Y. Georgellis, and R. E. Lucas. 2008. “Lags and Leads in Life Satisfaction: A Test of the Baseline Hypothesis.” The Economic Journal 118: 222–243. DaVanzo, J. 1983. “Repeat Migration in the United States: Who Moves Back and Who Moves On?” The Review of Economics and Statistics 65: 552–559. De Jong, G., A. Chamratrithirong, and Q.-G. Tran. 2002. “For Better, for Worse: Life Satisfaction Consequences of Migration.” International Migration Review 36: 838–863. Di Tella, R., J. P. Haisken-DeNew, and R. MacCulloch. 2007. “Happiness Adaption to Income and to Status in an Individual Panel.” NBER Working Paper No. 13159. Di Tella, R., and R. MacCulloch. 2006. “Some Uses of Happiness Data in Economics.” Journal of Economic Perspectives 20: 25–46. Di Tella, R., R. MacCulloch, and A. J. Oswald. 2001. “Preferences over inflation and unemployment: Evidence from surveys of happiness.” The American Economic Review 91: 335–341. Diener, E., Suh, E.M., Lucas, R.E., & Smith, H. E. 1999. Subjective well-being: three decades of progress. Psychological Bulletin, 125(2), pp, 276-302. Dolan, P., T. Peasgood, and M. White. 2008. “Do we really know what makes us happy? A review of the economic literature on the factors associated with subjective well-being.” Journal of Economic Psychology 29: 94–122. IAB-Bibliothek 358 134 Does migration make you happy? The influence of migration on subjective well-being Easterlin, R. A. 1974. “Does economic growth improve the human lot?” Pp. 89–125 in Nations and Households in Economic Growth: Essays in honor of Moses Abramowitz, edited by P. A. David and M. W. Reder. New York: Academic Press. Easterlin, R. A. 2003. “Explaining Happiness.” Proceedings of the national academy of Sciences of the United States of America 100: 11176–11183. Easterlin, R. A., and A. C. Plagnol. 2008. “Life satisfaction and economic conditions in East and West Germany pre- and post-unification.” Journal of Economic Behaviour and Organisation 68: 433–444. Edlund, L. 2005. “Sex and the City.” Scandinavian Journal of Economics 107: 25–44. Fandrem, H., D. Sam, and E. Roland. 2009. “Depressive Symptoms Among Native and Immigrant Adolescents in Norway: The Role of Gender and Urbanization.” Social Indicators Research 92: 91–109. Ferrer-i-Carbonell, A., and P. Frijters. 2004. “How important is methodology for the estimates of the determinants of happiness.” The Economic Journal 114: 641–659. Festinger, L. 1957. A Theory of Cognitive Dissonance. Stanford, California: Stanford University Press. Frey, B. S., and A, Stutzer. 2002. Happiness and Economics. Princeton: Princeton University Press. Frey, B., and A. Stutzer. 2005. “Happiness Research: State and Prospects.” Review of Social Economy 112: 207–228. Frijters, P., J. P. Haisken-DeNew, and M. A. Shields. 2004. “Investigating the Patterns and Determinants of Life Satisfaction in Germany Following Reunification.” The Journal of Human Resources 39: 649–674. Hagerty, M. R. 2003. “Was Life Better in the ‘Good Old Days’? Intertemporal Judgments of Life Satisfaction.” Journal of Happiness Research 4: 115–39. Handlin, O. 1951. The Uprooted. Boston: Little, Brow and Company. Heckman, J. J., and J. A. Smith. 1996. “Experimental and Nonexperimental Evaluation.” Pp. 37–88 in International Handbook of Labour Market Policy and Evaluation, edited by G. Schmid, J. O’Reilly, and K. Schönmann. Aldershot: Edward Elgar. Hunt, J. 2006. “Staunching Emigration from East Germany: Age and the Determinants of Migration.” Journal of the European Economic Association 4: 1014–1037. Kahneman, D., and A. B. Krueger. 2006. “Developments in the Measurement of Subjective Well-Being.” Journal of Economic Perspectives 20: 3–24. Kalter, F. 1998. „Partnerschaft und Migration: Zur theoretischen Erklärung eines empirischen Effekts.“ Kölner Zeitschrift für Soziologie und Sozialpsychologie 50: 283–309. Lu, M. 2002. „Are Pastures Greener? Residential Consequences of Migration?” International Journal of Population Geography 8: 201–216. 135Chapter 4 References Lucas, R. E., A. E. Clark, Y. Georgellis, and E. Diener. 2003. “Reexaming adaption and the set point model of happiness: Reactions to changes in marital status.” Journal of Personality and Social Psychology 84: 527–539. Lundholm, E., and G. Malmberg. 2006. “Gains and Losses, Outcomes of Interregional Migration in the Five Nordic Countries.” Geografiska Annaler. Series B, Human Geography 88: 35–48. Magdol, L. 2002. “Is moving gendered? The effect of psychological well-being among Australian young people.” Sex Roles 47: 553–560. Massey, D. S., J. Arango, G. Hugo, A. Kouaouci, A. Pellegrino, and J. E. Taylor. 1998. Words in Motion: Understanding International Migration at the End of the Millennium. Mayer, K.U. 2006. “Society of departure: The German Democratic Republic” Pp. 29–43 in After the fall of the wall: Life courses in the transformation of East Germany, edited by M. Diewald, A. Goedicke and K.U. Mayer. Stanford, California: Stanford University Press. Melzer, S. M. 2011. “Reconsidering the Effect of Education on East-West Migration in Germany.” European Sociological Review: online DOI: 10.1093/esr/jcr056. Mincer, J. 1978. “Family Migration Decisions.” Journal of Political Economy 86: 749–773. Neto, F. 1995. “Predictors of Satisfaction with Life Among Second Generation Migrants.” Social Indicators Research 35: 93–116. Park, R. E. 1928. “Human Migration and the Marginal Man.” The American Journal of Sociology 33: 881–893. Safi, M. 2010. “Immigrants’ Life Satisfaction in Europe: Between Assimilation and Discrimination.” European Sociological Review 26: 159–176. Scitovsky, T. 1992. The joyless economy: The psychology of human satisfaction. Oxford: Oxford University Press. Scott, W. A., and R. Scott. 1989. Adaptation of immigrants: Individual differences and determinants. Oxford: Pergamon Press. Shihadeh, E. 1991. “The Prevalence of Husband-Centered Migration: Employment Consequences of Married Mothers.” Journal of Marriage and the Family 53: 432–444. Sjaastad, L. A. 1962. “The Costs and Returns of Human Migration.” The Journal of Political Economy 70: 80–93. Staatliche Zentralverwaltung für Statistik. 1989. Statistisches Jahrbuch der Deutschen Demokratischen Republik, edited by Jahrgang. Berlin: Staatsverlag der Deutschen Demokratischen Republik. Statistisches Bundesamt. 2011. Bevölkerung und Erwerbstätigkeit. Wiesbaden: Statistisches Bundesamt. IAB-Bibliothek 358 142 Reconsidering gender-specific commuting and migration between East and West Germany In OECD (2005, p. 92) countries, on average, more than one-tenth of the employed population commutes to work.2 With 16 percent of the employed population involved in commuting, Germany and the UK show the highest commuting rates.3 Hereby, the former border between East (the former German Democratic Republic) and West Germany (the former Federal Republic of Germany) plays an outstanding role in determining the mobility patterns within Germany. The East- West German divide represents a unique example of regional wage gaps within Western European countries. Indeed, as of 2009, wages in East and West Germany differed up to 30 percent, a difference that is comparable to income gaps among two countries rather than two regions (Statistisches Bundesamt 2010). These high income gaps create large incentives for East Germans to work in West Germany. Therefore, it is not surprising that since the reunification, 1.1 million more people migrated from East to West Germany than the other way around (Statistisches Bundesamt 2010)4 and an additional 300 thousand more people commute5 every year to work in West Germany (Haas 2008).6 East-West commuting and migration are characterized by high mobility between economically divergent but institutionally identical regions. Previous research has focused mainly on migration, neglecting other mobility patterns (van Ommeren et al. 1997). Other mobility patterns, such as commuting, are typically investigated separately from migration (Eliasson et al. 2003, p. 828; Nivalainen 2010, p. 146), even though the mobility patterns are interrelated (Zax 1991; Zax and Kain 1991) and despite the fact that migration and commuting can be used as alternatives (Green et al. 1999; Nivalainen 2010) or complements (Lundholm 2010; Sandow and Westin 2010). By ignoring commuting, either the population under risk of migration is incomplete because commuters are excluded from the sample or the analyzed categories are blurry because commuters and stayers, i.e. non-migrants, are combined within one group. In the worst case, this makes the investigation incomplete (Romaní et al. 2003), the interpretation of the effects more difficult, and may lead to biased results. 2 The OECD defines commuting based on the NUT2 levels and distinguishes between 36 administrative districts. People, who work in one of the administrative districts and live in another, are commuters. 3 In Germany and the UK, 16 percent of the employed population commutes to work between regions, while in countries like France (12 percent) the Netherlands (12 percent) and Italy (10 percent), the commuting rates are slightly lower (OECD 2005). 4 Most of the western federal states experienced population gains first of all Bavaria and Baden-Wurttemberg, which experience a population growth up to 9.4 percent in the period from 1990 to 2008. Meanwhile the population declined in the new federal states. In Sachsen-Anhalt the population losses were as high as 17.1 percent and in Mecklenburg-West Pomerania they still reached 13.5 percent (Statistisches Bundesamt 2010). 5 In this article, we use the term “commuter” for an individual who works in West Germany while residing in East Germany. East Berlin is defined as part of eastern Germany, while West Berlin is defined as part of western Germany. 6 Additional information on the net commuter ratio was provided by Anette Haas at the basis of pallas-data from the Institute of Employment Research. 143 Introduction Chapter 5 Second, the focus in existing research combining commuting and migration differs from the question analyzed here. The few existing works combining the analysis of migration with that of commuting focus on the impact of commuting distance between work and the place of residence on the decision to migrate (Clark et al. 2003; Zax and Kain 1991), the impact of recent migration on the commuting distance (e.g. Champion et al. 2009; Green et al. 1999), or varying work-living constellations (e.g. Kalter 1994; van Ommereren et al. 2000). Thus far, the literature is silent about the connection and interrelation between commuting and migration (Lundholm 2010), which is the focus of the present research. Third, to the best of our knowledge, no study has investigated the connection between commuting and migration while taking gender-specific differences into account. Previous research does not provide insight on the gender-specific usage and interrelations between the mobility forms or on the gender-specific impact of factors such as education on the choice of mobility forms. Recent research on migration on the micro and the macro level indicates gender-specific migration patterns (Dumount et al. 2007; Faggian et al. 2007; Hunt 2006; Melzer 2013) and higher self-selection on education among women than men (Dumount et al. 2007; Feliciano 2008; Melzer 2013). The assumption that such gender-specific patterns can be found for commuting stands to reason, particularly as theoretical considerations and previous empirical studies imply such differences (cf CORDIS 2008; Crane 2007; MacDonald 1999; Sandow 2008; Sandow and Westin 2010). We expect gender to have a substantial impact on the mobility form chosen and the functions assigned to them for several reasons. First, there is a gender-specific impact of the stress associated with commuting (Roberts et al. 2011). Second, women’s incomes are lower; thus, women have lower means available for commuting (MacDonald 1999). Third, the tasks that men and women perform within the household differ, with women engaged more (Johnston-Anumonwo 1992; Turner and Niemeier 1997). Based on these differences, we expect lower commuting attendance among women, especially those in partnerships, and that women commute only temporarily and as a stepping stone to migration rather than as an alternative to it. The investigation of gender-specific differences in commuting and migration behavior should deepen our knowledge on mobility because gender it has been neglected thus far, and gender, like race, class, and nationality, is an important dimension of social inequality (cf Portes 1997, p. 816). Finally, the few previous studies investigating commuting and migration used multinomial logit regressions, ignoring the nested structure of the data and the fact that persons are questioned on an annual basis. We use multinomial logit regressions with random effects to account for the structure of the data and the IAB-Bibliothek 358 144 Reconsidering gender-specific commuting and migration between East and West Germany fact that persons are more similar to themselves than to other people and that the likelihood of migration is non-independent from the individual. This article aims to fill some of the gaps in research on commuting and migration between East and West Germany. It addresses the following questions based on the neoclassical migration theory: Which people decide to commute; which people decide to migrate? Does commuting serve as a stepping stone or as an alternative to migration? Are there gender-specific differences in the choice and usage of the mobility patterns? Using the German Socioeconomic Panel Study (SOEP) data from 1990 to 2009, we first estimate multinomial logit regressions with random effects to identify differences between non-migrants, commuters and migrants. Second, random effects logit regressions are estimated to determine whether commuting serves as a stepping stone or a substitute for migration. This research should not only provide information on the connections and interrelations between migration and commuting but also show whether it is justifiable to reduce mobility decisions to the choice between migrating permanently and staying. 2 Previous research The existing research provides mixed evidence on the connection between migration and commuting. Research from Sweden (Eliasson et al. 2003; Lundholm 2010; Sandow and Westin 2010) indicates that commuting is used as a stable long-term solution and thus can be interpreted as an alternative to migration. Using binary probit models in the context of Spain, Romaní et al. (2003) report that commuting increases the likelihood of migrating, indicating that commuting serves as a stepping stone to migration. However, Romaní et al. (2003) find, in line with the findings of Champion et al. (2009) for England, that migration increases the probability of commuting, indicating that relocating (e.g., to a more rural area) might increase the travelling distance to work and thus have a positive influence on commuting. Regarding individual characteristics, previous research agrees that commuters tend to be slightly younger than stayers and have relatively even age patterns (Lundholm 2010; Nivalainen 2010; Sandow and Westin 2010), whereas migrants are much younger than both commuters and stayers (Nivalainen 2010; Romaní et al. 2003). Most research finds that both migrants and commuters have higher education (Eliasson et al. 2003; Lundholm 2010; Nivalainen 2010; Romaní et al. 2003); however, some studies find that commuters are middle educated (Sandow and Westin 2010). To the best of our knowledge, only one study has simultaneously investigated commuting and migration between East and West Germany. Estimating a multinomial logit model, Hunt (2006) shows that commuters tend to be younger 145 Theoretical framework Chapter 5 than non-migrants but not as young as migrants. In line with international research, the author shows that women are less likely to commute. However, contrary to the mainstream migration research, the author finds that East German women are slightly more likely to migrate than men. This finding was also supported by Melzer (2013) and Windzio (2007). 3 Theoretical framework People rationally decide whether to remain in their place of origin or to become mobile by comparing the costs and gains related to the different forms of mobility, such as commuting and migration, and staying (van Ommeren et al. 1997; van Ommereren et al. 2000; Zax 1991). Both forms of mobility bear monetary and non-monetary costs (DaVanzo 1981; Massey et al. 1993; Stutzer and Frey 2008; Zax 1991), but they differ in regards to timing of the costs accrual and the risk involved. Therefore, the investigation of mobility decisions requires the consideration of how such factors influence peoples’ mobility choices, for example, by distinguishing between short- and long-term costs. 3.1 Mobility costs People who migrate must make investments mostly before the relocation occurs, placing uncertainty and risk into the decision (Sjaastad 1962). Transportation and a new place of residence must be arranged, and the old accommodation must be canceled or sold. After relocation, there are no additional monetary long-term costs and the total costs should not exceed the costs incurred in the short term. Also the total non-monetary costs should not exceed the short-term costs, as migrants lose their location-specific human capital and are separated from their social networks immediately after relocation. The costs of commuting differ considerably from those of migration. The short-term costs are much lower, as people stay at their place of residence and profit from their social networks and location-specific human capital. Most importantly, commuting does not require high monetary investments before the event and is much less risky than migration. In contrast to migration costs, which do not rise further once the relocation occurs, the costs of commuting build over time. Every day that an individual commutes, new monetary costs are added. In addition, the non-monetary costs also accumulate, as commuting is one of the least pleasant daily activities (Kahneman et al. 2003) and embodies high mental and physical burden for both the commuter and the family (Roberts et al. 2011; Stutzer and Frey 2008). Commuting, especially over long distances, is very time- IAB-Bibliothek 358 146 Reconsidering gender-specific commuting and migration between East and West Germany consuming and reduces the time available for leisure and family activities, which might even create partnership problems (Kley 2012; van der Klis and Mulder 2008). As such, the costs of commuting and migration differ considerably over time. Although the costs of migration exceed commuting costs and are connected to higher risk in the short term, migration costs should be lower than commuting costs in the long term. Short term: commuting < migration Long term: migration < commuting The low costs and risk involved in commuting makes is more attractive as a shortterm solution than migration and should result in low selectivity among commuters. However, both commuting and migration costs, and thus the gains obtained, depend on individual characteristics and circumstances (Sjaastad 1962; van Ommeren et al. 1997; van Ommereren et al. 2000; Zax 1991), which also define peoples’ mobility choices. Highly educated people should not only be able to rely on better and more effective search strategies (Chiswick 1999), but are also more likely to profit from broader social networks, reducing both the monetary and non-monetary costs of migration (Brücker and Defoort 2009). In addition, mobility might be more important for their careers, as the labor market for highly educated and specialized individuals is less dense (Büchel and van Ham 2003) and mobility might become inevitable for promotion or to avoid overqualification for highly educated. Assuming that their good qualifications and high productivity is reflected in their earnings, highly educated should have few problems in raising the necessary financial means. The migration costs should be relatively lower than their earning potential, resulting in low risk in the decision to migrate. Lower migration costs increase the attractiveness of migration and lowers the incentives to commute or the duration people can spend commuting before the commuting costs exceed the migration costs. Hypothesis 1: When highly educated people (who have very low migration costs) become mobile, they are more likely to migrate than to commute in comparison to lower educated people. Hypothesis 2: When highly educated people begin commuting, this is only a temporary solution (which reduces the risk of migration even further) and they are more likely to use commuting as a stepping stone than as an alternative to migration compared with lower educated people. The mobility decisions should differ for people in partnerships. For this group, the migratory gains must exceed not only their own migration costs but also those 147 Theoretical framework Chapter 5 of their partners, which is especially difficult for double-career couples (Mincer 1978). As migration is shown to reduce earnings, working hours or to cause interruptions in the labor market participation of one partner, which is typically the woman (Boyle et al. 2001; Boyle et al. 2008; Rabe 2011; Shauman and Noonan 2007), commuting can serve as a compromise that provides the possibility to pursue a career while the partner can do the same at the place of origin (Green 1997; van der Klis and Mulder 2008). Hypothesis 3: When people in partnerships become mobile, they are more likely to commute than to migrate compared to singles. Hypothesis 4: People in partnerships are more likely to use commuting as an alternative rather than as a stepping stone to migration compared to singles. 3.2 Gender-specific differences In line with theoretical and empirical research on commuting (cf CORDIS 2008; Crane 2007; MacDonald 1999; Sandow 2008; Sandow and Westin 2010) previous research on the East-West mobility in Germany indicates lower commuting rates for women (Hunt 2006). Contrary, despite the general exception that women are less mobile (e.g., Le Grand and Tåhlin 2002; Lehmer and Ludsteck 2009), previous research in Germany indicates higher migration rates from East to West Germany among women (Hunt 2006; Melzer 2013; Windzio 2007). This might indicate a substitution effect, in which the higher migration rate of women indicates different choices regarding the mobility forms among men and women. It is possible that while women choose to migrate to the West, men choose to commute. Hypothesis 5: When women become mobile, they are more likely to migrate than to commute compared to men. Hypothesis 6: When women begin commuting, they are more likely to use commuting as a stepping stone rather than as an alternative to migration than are men. People who migrate profit from regional wage differences. The higher the income gap, the more individuals, independent from their personal characteristics, can realize gains from the relocation (Sjaastad 1962). In 2009, for example the wage gap between East and West Germany was 30.3 percent for men. For women, the average West German wages were only 12.9 percent higher than the East German wages (Statistisches Bundesamt 2010). The lower income gap should enable IAB-Bibliothek 358 148 Reconsidering gender-specific commuting and migration between East and West Germany only a specific part of the female population with higher education to gain from migration. The additional 17.4 percent of income increase connected to migration should, in turn, enable even some men with higher migration costs, such as those with slightly lower education or those in partnerships, to profit. Previous research confirms that only women with at least upper secondary degrees seem to gain from the relocation from East to West Germany, whereas men are able to gain from the relocation with lower educational degrees (Melzer 2013). As the incentives should be similar for all types of mobility, we expect commuting to be profitable only for women with higher education. Hypothesis 7: Women’s mobility choices are more dependent on higher education than are those of men. The typical difficulties for men and women in partnerships to compensate their own and their partner’s migration costs are enforced further for women because the gains from working in West Germany are, on average, smaller for women than for men. This, in connection with the tendency of couples to place the men’s career first (e.g. Bielby and Bielby 1992; Boyle et al. 2001; Nivalainen 2004), makes it difficult for women in partnerships to initiate a relocation (Nisic and Melzer 2013). At the same time, due to the gender-specific division of the household tasks, it should be also more difficult for women in partnerships to commute to work than for men in partnerships (cf CORDIS 2008; Crane 2007; Johnston- Anumonwo 1992; MacDonald 1999; Sandow 2008; Sandow and Westin 2010; Turner and Niemeier 1997). Therefore, it is expected that: Hypothesis 8: Women’s mobility choices are more restricted by their partnership than are the mobility choices of men. 4 Data and methods 4.1 Data German Socio-Economic Panel Study (SOEP) from 1992 to 2009, a representative longitudinal survey of private households that began in 1984 in West Germany and West Berlin and six years later in East Germany and East Berlin (Wagner et al. 2007), is used. The sampling procedure is based on a random selection of households; every member over 16 years old is interviewed. Unbalanced samples restricted to members of the working age population between 16 and 63 who spend at least one year in East Germany are used for the 149 Data and methods Chapter 5 estimations.7 The dataset is balanced on gender (51.2 percent women). Commuting is more common among men; 61.7 percent of commuters are men, whereas they only comprise 43.7 percent of migrants. The presented estimations are based on two slightly different samples. The first analyses, which investigate the determinants that lead to the decision to begin commuting or to migrate (presented in Table 2), include only information from the year immediately before people begin to commute or migrate. The periods following the choice of mobility are censored to ensure that the causes and consequences of mobility decisions are not mixed and that commuting and migration are treated analog. We distinguish people who are immobile, who migrate, and who begin commuting. The first dataset (1992–2008)8 contains 49,533 observations for 4,010 men and 4,067 women, with 663 observations for people who migrated and 1358 observations for those who commuted. In the first set of estimations, the dependent variable, mobility form, takes the value of one for people who commuted (1), two for people who migrated (2) and zero for stayers. In the second set of estimation, which investigates how commuting influences the migration decision, we no longer use the variable that indicates the distance to the West German boarder and include all commuting episodes, which extends the data to 62,329 observations for 8,951 persons. The dependent variable, migration, takes the value of one if a person migrates and zero, otherwise. Independent Variables To analyze the various forms of mobility, we control for age using six age groups, gender (female = 1) and if people have a partner (married or in partnership = 1). Moreover, we use two dummy variables to indicate whether people have children and the children’s age (i.e., younger than 6 versus between 6 and 18 years). To account for individual-level qualifications, we measure education (centered at nine years), duration spent in the current company and cumulative duration of unemployment in years. We account for employment status by distinguishing between full-time employment, part-time employment, apprenticeship (employment = 0) and unemployment (= 1). Income is deflated to the year 1992. Moreover, we use the logarithm of income to account for the fact that an increase by one unit might have different consequences for 7 We do not control for place of residence prior to the fall of the Berlin Wall, as is sometimes done (see e.g., Hunt 2006). Although this procedure might have been warranted for the first years after reunification, particularly in research specifically focused on the migration patterns of East Germans, the populations have mixed 20 years after reunification and former West Germans who have lived in the eastern part since the reunification would have been excluded from the sample if we continued to follow this procedure. Moreover, people born in 1990, who are now able to make their own migration decisions, are not documented by the question, “where did you live in 1989?” This question is typically used to distinguish between East and West Germans. 8 As migration can only be observed between two waves, the last year in the dataset is excluded from the estimations. IAB-Bibliothek 358 150 Reconsidering gender-specific commuting and migration between East and West Germany people at the lower and the higher end of the income distribution. To account for monetary migration costs, we control for homeownership using a dummy variable that takes the value of one if a person owns real estate and zero, otherwise and distance to the nearest West German town or to West Berlin is measured in 10 km steps. In the second set of estimations, in addition to the previously listed variables, we include the variable commuting, which is one (1) when persons who live in East Germany commute to work in western Germany (1) and is otherwise zero (0). To measure the mediating effect of gender on the influence that education and partnership has on mobility decisions, two interaction terms with the variable female were included in the first set of estimations. In the second set of estimations, interaction terms with the variable commuting indicate whether various groups of the population, such as women or the highly educated, use commuting differently than the main population. 4.2 Methods We use multinomial logit and logit regressions with random effects (RE), which take into account the clustered structure of the dataset and the fact that individuals were questioned on several occasions. Moreover, models with RE, in contrast to simple multinomial logit and logit models, are able to account for the non-independence of the likelihood of migration or commuting from individuals and, thus, the fact that individuals are more similar to themselves over time than to others. We estimate RE regressions because education, one of the main variables of interest, does not change much over a person’s lifetime once schooling is completed, which is the case for the main part of the sampled population (cf Melzer 2013). Fixed effects (FE) regressions are not able to estimate the influence of education on mobility decision in such a case, as the sample barely shows any within-subject variance. Contrary to FE regressions, which are based entirely on the within-subject variance, RE models estimate the likelihood of migration relying on between-subject and within-subject variance simultaneously and are able to provide information on differences in the mobility patterns of people with varying education. One drawback of RE models in comparison to the FE models is that they require an additional assumption regarding the structure of unobserved heterogeneity involved, assuming that the unobserved factors are not correlated with the explanatory factors and this assumption might be violated (Wooldridge 2009, p. 496). However, if the assumption holds, RE panel regression models are not only consistent, and the Hausman test verifies this, but also more efficient than FE models. 151 Results Chapter 5 To simultaneously analyze the decision to stay, commute or migrate, we use multinomial logit regressions with random effects that account for the fact that individuals face a set of opportunities simultaneously. Multinomial logit regressions estimate J-1 logit models at the same time, comparing J outcomes simultaneously (Long 1997). The presented models employ non-migrants as the reference group. To use multinomial regressions, the independence of irrelevant alternatives (IIA) assumption must be fulfilled. In other words, the odds of any two forms of mobility must not change when an additional form of mobility is included or removed from the estimation. We tested the independence of irrelevant alternatives (IIA) assumption; all three computed tests (i.e., the Hausman, the Suest-based Hausman and the Small-Hsiao tests) support the IIA and indicate that multinomial logit regressions are suitable to investigate the choices of mobility forms. For the estimation of multinomial logit models with random effects, the IIA is relaxed. In addition, the correlation between the alternatives is allowed by introducing random effects. In both the multinomial and the logit models, we estimate person-specific intercepts, which account for persons’ variability in their willingness to become mobile and determine the persons’ behavior over time. Separate analyses for men and women are estimated in addition to pooled estimations with gender-specific interaction terms, as previous research and theoretical considerations suggest gender-specific mobility patterns. 5 Results 5.1 Descriptive results Table 1 provides a description of the characteristics of male and female stayers, commuters and migrants. People from the mobile groups are younger. Migrants are less likely to be in partnerships and have spent less time either unemployed or in employment in their last company than stayers and commuters. Generally, commuters resemble stayers to a greater degree than migrants. Among commuters, those who migrate after spending a period commuting seem to be younger and better educated. The high incomes that commuters earn indicate the higher income levels of West Germany. Finally, the low rate of unemployment among commuters also points to the nature of commuting, as persons work in West Germany, which automatically excludes unemployed persons. The few percent of commuters who are listed as unemployed spent the majority of the analysis year unemployed (i.g. seven months) before they began to commute.