Quasi-natural experiments in socio-demographic analysis. A data issue. Roberto Impicciatore University of Bologna
Causal inference in socio-demographic research Standard approach: controlling for i.e. including control factors in multivariate analysis check for spurious correlations. Econometric approach: multiprocess modelling (e.g. Heckman, SUR, etc.); propensity score matching, etc. check for endogeneity/selectivity. Quasi-natural experiment: data are drawn from naturally occurring phenomena 1. subject who receive a treatment can be compared to those of a control non-treated group; 2. Subject are not randomly assigned to treatment and control groups 3. Experiment is not under control of the researcher 2
Quasi-natural experiment. Example 1 The relevance of local environment in the first years of life for future health and well-being (Bengtsson 1989, 1993, 2004; Bengtsson and Lindström 2003; Quaranta 2013) I ndividuals born in years with high infant mortality rates or with smallpox or whooping cough epidemics experienced lower socioeconomic performance and higher levels of adult and old-age mortality, while no effects were seen in relation to exposure to high prices (Sweden 1813-1968, Scanian Economic Demographic Database) 3
Quasi natural experiment. Example 2 Month of birth is largely random and exogenous to the process of educational attainment Nevertheless, being born just before the cutoff (November-December) for school enrollment can be conceived as a negative trait that affects the likelihood of future success (France: Bernardi F., 2014) and also timing in family formation (Sweden: Skirbekk, Kohler, Prskawetz 2004)
Quasi-natural experiments in socio-demographic analysis. Disentangling the role of welfare regime and persistent cultural traits on demographic behaviours.
Quasi natural experiment. Example 3 Family policies and the Western European fertility Divide: Insights from a natural experiment in Belgium (Klüsener, Neels, Kreyenfeld 2013) A distinct fertility divide: a clear dividing line seems to run through Western Europe and largely follows national borders. Specific aspect of political geography may help to disentangle the role of institutional context and social norms in understanding fertility trends.
Background: In the aftermath of World War I, Germany was forced to cede the territory of Eupen Malmedy to Belgium as compensation for its attack on the neutral Belgian state. After more than 90 years this community has retain its linguistic identity and social norms prevalent in the German society. Nevertheless, they have been subject to incentive structure created by the Belgian family policies. Results: their overall fertility outcomes resemble the Belgian pattern (institutional context plays the key role in explaining the fertility differences between Belgium and Germany)
Example 4. Focus on second generation migrants. Children of immigrant born and raised in the host country. They share the same welfare and institutional context as the children of non-immigrant parents, but they are exposed to a different cultural influence within their family. Thus, the comparison between patterns experienced by second generations and those with native parents can shed light on the role of the welfare regime and of specific cultural factors transmitted from parents to their children.
The transition to adulthood of the Italian Second (Impicciatore 2015) Generation in France Context: Lack of convergence in the patterns of transition to adulthood in Europe i.e. the Second Demographic Transition has not spread uniformly across Western Europe (postponement and lower diffusion of cohabitation and children out-ofwedlock in Italy and Spain) Two possible explanations: 1. Socio-economic context, institutional setting, welfare and public policies (e.g. Buchmann 1989; Esping-Andersen 1990; Mayer 1997; Vogel 2002; Saraceno 1994; Naldini 2003) 2. Cultural traits Anthropological differences rooted far in the past and persistent over time (Reher 1998): weak family ties (individualistic values) vs strong family ties (familism).
Familistic viewpoint (transmitted from parents to their children generation after generation) may influence the transition to adulthood Leaving parental home parents do not encourage their children to leave because they are reluctant to see their children suffer in material terms ( golden cage ) (Dalla Zuanna 2001, Castiglioni and Dalla Zuanna 1994). Non marital cohabitation low incidence of cohabitation and other behaviours that may openly clash with the values of parents (Rosina and Fraboni 2004; Axinn and Thornton 1993, Rosina and Micheli 2006; Viazzo 2003) Children out-of-wedlock societies with strong families tend to have greater social cohesion resulting in the low incidence of divorce and extramarital births (Reher, 1998). R. Impicciatore - Living arrangements of immigrants descendants in Europe
On the one hand. The extent to which the second generation differs from the majority reduces the relevance of the welfare regime and the institutional settings in shaping the transition to adulthood. On the other hand The evidence that the patterns followed by second generation are parallel to their counterparts in the country of origin reinforces the hypothesis that familistic cultural norms are at play in defining life course patterns. R. Impicciatore - Living arrangements of immigrants descendants in Europe
Why Italians in France? 1. A large number of people with Italian origin. 2. Crucial differences in the transition to adulthood between Italy and France. 3. The wide retrospective survey called ètude de l'histoire familiale (EHF) carried out during 1999 (360,000 cases): 7313 individual born in France with at least one parent born in Italy. Research strategy - Longitudinal approach - Long-term perspective (cohorts born over 8 decades) R. Impicciatore - Living arrangements of immigrants descendants in Europe
Median ages by cohort Exit from parental home (median age) Delay among G2 in the exit from parental home (in between Italians and French) Cohabitation as first union (% among who has experienced a union) Increasing gap in the decades born after 1950. In particular among women. First child out of marriage (% among parents)
Other covariates included in the models: a. socio-economic category of parents, birth cohort and birth order, number of siblings, living with both parents at 14 years of age, area of residence, urban dimension, level of education (time-varying), to be a student (time-varying), entered in the labor market (time-varying), b. and c. socio-economic category of parents, birth cohort and birth order, number of siblings, living with both parents at 14 years of age, level of education (at the interview), area of residence, urban dimension. Significance: *** 99%; ** 95%; * 90%. Results: multivariate analysis (hazard and logistic models) 1. Persistent delay for leaving parental home 2. Lower probs for non marital cohabitation and first child out-of-wedlock. 3. Stronger gap among women (also for the timing of first birth
Conclusion Key role of institutional settings and welfare (Italian G2 are more similar to French native than Italians) However, Clear signs of the effect of familistic perspective among immigrant families: slow diffusion of modern value orientation (deep-rooted cultural traits). Lack of convergence in Europe should not be entirely explained in terms of Welfare regime and the institutional setting but also in terms of persistent cultural norms
Thank you roberto.impicciatore@unibo.it