Cultural Influences on the Fertility Behaviour of First- and Second-Generation Immigrants in Germany Holger Stichnoth Mustafa Yeter ZEW Mannheim 8. Nutzerkonferenz Forschen mit dem Mikrozensus Mannheim November 12, 2014
I - Motivation Low fertility rates in Germany Total fertility rate circa 1.4 (France, UK, US: circa 2) Despite circa 180 billion euros p.a. spent on family-related measures Heated public discussion about immigration: Beneficial for the pension system? Detrimental for the composition of the society? Research questions: Does the exposure to a certain fertility environment affect one s own fertility behaviour even under a common institutional setting? To what extent is this exposure effect transmitted across generations?
Source: 1950s - UN Population Division (2011); post-1960 - World Bank (2012)
I - Challenges - Identification Norms, beliefs and institutions are likely to be intertwined Institutions influence, but also reflect norms and beliefs Challenge: disentangle the cultural effect from the institutional Epidemiological approach (surveys: Fernández 2007, 2008) Immigrants in the same host country Different fertility environments pre-migration, but (most of) fertile period in the same institutional setting 2 nd generation: no own exposure, cultural norms transferred from 1 st generation holding the institutional background constant
I - Challenges - 'Culture' and proxies 'Culture' is multidimensional: Norms transmitted through parents or religion Personal experience Conceptualising and measuring culture perfectly is not possible. We have to rely on proxies Proxies: Fertility norms in the country of origin Total fertility rate theoretical no. of children born to woman over her lifetime Cohort fertility rate no. of children actually born for specific birth cohorts Direct measures of fertility norms (World Values Survey)
I - Related studies Epidemiological approach for fertility outcomes: 2nd generation, US: Fernández/Fogli (2006, 2009), Blau et al. (2013) 1st generation, Germany: Cygan-Rehm (2011) Other outcomes: Female labour supply: Blau et al. (2011, US), Blau/Kahn (2011, US), Gevrek et al. (2011, Canada), Kok et al. (2011, Netherlands) Preferences for redistribution: Luttmer/Singhal (2011, Europe) Divorce: Furtado et al. (2011, US)
I - Contributions Scope: Cygan-Rehm (2011) pools SOEP 1991, 1999, 2007, only 1 st generation We study also the 2 nd generation, have a much larger sample size Method: Exploit within-country variation (1 st Gen.: different arrival years, 2 nd Gen.: different birth years) New (additional) proxy for culture: cohort fertility rates Examine sensitivity of results with respect to the upper limit for age at migration Controlling for mixed origins
I - Main results Fertility rates in country of origin are positively related to completed fertility of female immigrants in Germany Strength of effect: 1 st generation > generation 1.5 > 2 nd generation Results sensitive to upper limit for age at migration Generation TFR = 1 ( 1 s.d.) in completed fertility 1 Year of migration [0.28, 0.35] (pooled) [0.32, 1.10] (within) 2 Year of birth 0.06 (pooled) Stronger on the intensive than the extensive margin and for low education, partner or parents from the same origin
II - Data Mikrozensus 2008 1% sample of German households (mandatory participation) Women of first and second generation Generation 1: immigrated at age 15 or older Generation "1.5": immigrated below age 15 Generation 2: born in Germany, but at least one parent immigrated 18 countries of origin Sample restrictions: Age 45 (completed fertility) First generation: migration age 18, as identification relies on fertility choices under common institutional setting
III - Identification strategy Institutions Fertility rate Turkey Fertility norms Institutions & fertility norms Germany 1 st Generation Turkey 2 nd Generation Institutions Fertility rate Italy Fertility norms 1 st Generation Italy Transmission of fertility norms 2 nd Generation Migration time
III - Main specification OOOOOOOOOOOOOOOOOOOOO iiiiiiii = ββ 0 + ββ 1 xx ii + ββ 2 FFFFFFFFFFFFFFFFFFFFFFFFFF cccc + ff cc + ff ss + εε iiiiiiii Indices: woman (i) who arrived from country (c) in year (t) and currently resides in federal state (s) OwnCChiiiiiiiiiiii iiccst : completed fertility FFFFFFFFFFFFFFFFFFFFFFFFFF cct : total fertility rate in country of origin in year of migration 2 nd generation: year of birth instead of year of migration ff c and ff ss : indicator variables for country of origin and federal state
IV - Robustness checks Count data and Tobit models instead of OLS Cut-off point separating generations 1 and 1.5 at age 6 or 12 Omitting years pre-1960 Omitting Turkey Omitting Russia/Poland (because of ethnic Germans )
V - Alternative specifications When controlling for marital status, labour force participation, foreign nationality, household income Coefficient on TFR almost unchanged in pooled models, slightly larger in within models Differences not statistically significant Variables lie on the causal pathway from norms to fertility outcomes, they should not be controlled for when interested in the total effect Coefficient on TFR gets smaller when controlling for GDP per capita Interaction models using education and origin of partner: 1 st generation: stronger with lower education & partner from the same country of origin 2 nd generation: stronger if parents from same country of origin
V - Fertility norms Direct evidence on fertility norms from the 1990 wave of the World Values Survey: "What do you think is the ideal size of the family?" "Do you think a woman has to have children in order to be fulfilled?" "A job is alright, but what most woman really want is children" Coefficients on these variables are statistically insignificant Possible explanations: long time lag between migration and 1990 fertility norms likely to have changed a lot by 1990 in countries of origin
VI - Conclusion Fertility rates in country of origin are related to completed fertility of female immigrants in Germany Strength of effect: 1 st generation > generation 1.5 > 2 nd generation Stronger on the intensive than on the extensive margin Strongest for women with low education, partner or parents from same origin Cultural effect does not persist indefinitely, no evidence for dramatic shifts in German society due to migration Demographic problems in Germany cannot be addressed by immigration alone