English title: Decreasing Social Mobility? The Swiss Case in Western European Comparison
Author(s): Simon Seiler -
Type: Conference paper/poster
Resent research on effects of social origin on class and education in 20th century Switzerland has not only demonstrated the persistence of these effects (Pfeffer, 2008; Falcon, 2012; Falcon, 2013) but some results suggest even increasing effects of social origin for the youngest cohorts (Jann and Combet, 2012; Jann and Seiler, 2014), especially on education (for a short overview, cf., Falcon and Joye, 2015). This empirical evidence directly contradict the modernization thesis which predicts constant reductions of effects of social origin over the course of processes which are summarized under the label of modernization (eg., Lipset and Zetterberg, 1959; Kerr, 1962; Blau and Duncan, 1967). The importance of these results from Switzerland is, however, difficult to assess, because, to my knowledge, occupational mobility in Switzerland has never and educational mobility has only once (Pfeffer, 2008) been studied comparatively. The two basic research questions of this paper are therefore: How do the effects of social origin on education and class found in Switzerland compare to those from other Western European countries? And: can the analyses in this paper confirm the increasing effects of social origin suggested by some results for the youngest birth-cohorts in Switzerland? Data and Methodological Approach: To investigate these questions, a harmonized dataset is used with data on nine countries: Switzerland (CH), Germany (DE), Denmark (DK), Spain (ES), France (FR), Great Britain (GB), Ireland (IE), Netherlands (NL), and Norway (NO). Data is drawn from the 2011 EU Statistics on Income and Living Conditions (Eurostat, 2015) which includes a module on intergenerational transmission of disadvantages, and from the rounds 4 7 of the European Social Survey (ESS, 2008; ESS, 2010; ESS, 2012; ESS, 2014). For 63 897 respondents aged 35 60 the necessary information on their own and theirs parents’ occupation and education is available. These observations are grouped into three birth cohorts (the year used as label in parentheses): 1950-59 (1955), 1960-69 (1965), and 1970-79 (1975). The two main dimensions used in the analyses are class and education. To avoid empty cells, both of them are measured with only three categories. For class, we collapse the original EGP-scheme (Erikson, Goldthorpe and Portocarero, 1983) into three classes: The service class (I & II) becomes the upper class, non-manual employees (III), self-employed (IVa,c) and skilled workers (V & VI) the middle class, and semi- and unskilled workers (VIIa) as well as farmers (IVc) and agricultural workers (VIIb) the lower class. In order to fully utilize the full richness of the data, homemaker is included as a forth class to the class-scheme of the mothers. Similarly, the ISCED categories for education have been collapsed into categories for low, intermediate, and tertiary education. The applied methodological approach to estimate the effects of social origin builds on an approach first used by Jann and Combet (2012) and evaluated and described in more detail by Jann and Seiler (2014). The basic idea is simple: If origin strongly affects destination, the knowledge of origin strongly improves the prediction of destination. In turn, the improvement of the prediction of destination by knowing both parents’ education or class, measured by the Proportional Reduction of Errors (PRE), can be seen as a measure for the effect of origin on class or education of the respondent. In practice, this involves three steps: predict destination using a model without origin, predict destination using a model with origin and assess the improvement of the prediction by calculating the PRE conditional on each country and birth cohort. Contrary to the procedure described by Jann and Seiler (2014), a GMM-estimator and not a bootstrap procedure is used in to obtain valid standard errors. Preliminary Results and Outlook: Figure 1 shows the estimated effect of social origin on class and education over the three birth cohorts, separately for men and women and for each of the nine countries. Additionally, each plot shows the estimated differences between the three birth cohorts. Estimates and differences whose spikes do not cross the zero-line can be said to be statistically significant different from zero. For class, the general picture depicts persistence of effect across birth cohorts. There are two exceptions: Effects of parents’ classes on German women’s class have risen significantly from the first to the later birth cohorts; the same is true for French men. Apart from that, no increasing (but also: no decreasing) effects of social origin on class can be found. The results are more diverse for effects of parents’ on respondent’s educational attainment. On the left side of the figure, we can see that effects of social origin are relatively high in Switzerland and Germany. Furthermore, the results show significantly increasing effects for German women and Swiss men. Rising effects of social origin can also be found for Danish women when we compare the youngest with the middle cohort. The opposite is only true for Norwegian women: Here, the effects decrease strongly and significantly from the first to the second before it insignificantly increases again. In summary, the present paper mainly draws a picture that is in line with the research on educational inequality and social mobility (cf., Breen and Jonsson, 2005). I.e., it depicts persisting effects of social origin. However, results deviating from this general picture do not exhibit decreasing effects of social origin, but increasing trends. The analyses presented in this paper support – at least for men – the findings from Switzerland suggesting increasing effects of social origin on educational attainment. They also show that this is not the normal case in Western European countries but that it is not the only exception either. Comparative research needs comparable data and comparable data always come to the prize that they are badly adapted to some local specificities. The present analyses should, therefore, be repeated with local data but comparable method. Comparable data are, on the other hand, the only way to see a specific result within a broader picture. And they can point to phenomena that need closer attainted – and increasing effects of social origin is definitely such a case.
Conference name: ISA RC28 Summer Meeting -
Start date: Aug 29, 2016