Lifecourse Patterns after Migration to Germany: The Interplay of Employment and Fertility Trajectories Since Arrival
Cristina Samper, Hertie School of Governance
Introduction: The interrelation between fertility and migration is a widely studiedtopic in demographic and sociological literature (Wolf2016, Kreyenfeld & Krapf2017, Milewski 2010, Schmidt & Kohls 2010, Mussino et al. 2015).In parallel, large amounts of literature have been developedon the dynamicwork-family link, mostly highlighting the negative correlation between employment and fertility trajectories (Budig 2003, Cramer 1980) in Germany and other contexts (Assve et al. 2007, Han& Moen 1999, Madero-Cabib& Fasang 2016, Schidt2012). However, few studies have done a combined analysis to understand howmigrant employment careers are affected by their fertilitycareers at arrival (Lundtrstm and Andersson 2012). Uncovering this relation is nonethelessimportant today given the attentionthe low labor market participation of female migrants has received in Germany. Against this background,this paper explores how labor market capital resources at migration, partnership, and fertility dynamics interact to determine differential life course trajectories.Data and Method
Datafor this analysis comes from the immigrant sampleof the German Socio EconomicPanel (GSOEP) which is organized as an annually repeated panel study. For thispreliminary analysis, we have selecteddata from women who have beenin the 2013 and 2014. As welimited our analytical sample to female respondentswho arrived first as adults ages 18-64, and those who arrived between1990-2005. Our final sampleconsists of 1090 female respondents.
Asa method, we employ multichannel sequence analysis (Gauthier et al. 2010, Pollock2007 ) the (main) process time is the time since migration. We havedefined individual alphabetsfor each channel (see figure 1) and a dynamic hamming distance (Lessnar 2006) to calculate a weighted dyadic distance matrix and account for themultidimensionality. Then we have useda ward method as a hierarchical clustering algorithm, which has as an aim toreduce residual variance and toevaluate the best possible clusteringsolution (Studer 2013).
Oncewe obtained a robust set of clusters a multinomial logistic analysis was done to compare the individual characteristics of cluster affiliation. Here we look at marital status, age groups, labor market capital and other socio-demographical characteristics.
Figure1 gives first descriptive results of the 5 group clustersolution . The graph displays the parallel employmentand birth histories of fivecluster types of female migrants. The right handcorresponds to the sequence indexplots of employment (Gabandinho et al. 2011), where every line corresponds to one lifehistory. On the left sidein are the correspondingsets of fertility frequencyplot (Fasang & Liao 2014). In C1 we can see a typology of women who are childlessat arrival and remain childless for at least the first5 years since arrival. These women have employmentand education patterns that could be in consonance with the disruption hypothesis (Wilson, 2013), where they have postponedtheir fertility decisions to firstintegrate into the host society.
Inthe other four clusters wecan identify two very distinct employmentpaths each attached to twodifferent types of birth histories. In C2 and C3 wecan see mothers who show mixed employment and housewives dynamics both attachedto women mostly with maximumone child or already with2 children at arrival. These women howa higher tendency to work thanwomen in C4 and C5, 20% of whichare permanently housewives for the 10 yearssince arrival in each group. Thetendency to stay in the householdis especially pronounced for the case of women from cluster 5 which have childrenvery close to arrival and continue to haveyoung children along the first10 years. These women we couldpossibly be identified by the interrelationof fertility and migration hypothesis (Ortensi, 2015).
Table 1 reportsthe results from the multinomiallogistic regression on the clusteraffiliation. We have used thechildless cluster as a baseline category. In each of the columns,we can see the relative risksof being in the corresponding cluster over the probabilityof being in the baseline category, in our case C1: the childless mixed work and education trajectories. Corroborating the descriptive findings women in the base cluster are those who weremost likely to be single at migration, additionally they are also the youngestand those to obtain the highestlevels of education. As for the affiliationof clusters C4 and C5 that consist of mostly housewife trajectories it is interestingto see theyare the groups who tend tohave the least local friends and are much more likely to have anIslamic religious affiliation in contrast to a Catholic one.