The role of higher education stratification in the reproduction of social inequality in the labor market

Available online at www.sciencedirect.com

Research in Social Stratification and Mobility 32 (2013) 45–63

The role of higher education stratification in the reproduction of social inequality in the labor market Moris Triventi ∗ Department of Sociology and Social Research, University of Milano-Bicocca, Milan, Italy Received 14 February 2012; received in revised form 5 December 2012; accepted 28 January 2013 Available online 8 February 2013

Abstract This paper analyses whether social origins affect labor market outcomes (wage and occupational status) of a recent cohort of graduates and whether the type of qualification obtained (program length, field of study and institutional quality) accounts for this relationship. We use data from the 2005 Reflex survey on European graduates in 4 countries (Germany, Norway, Italy, and Spain) which were selected on the basis of their institutional profiles. Results from binomial logistic regression models indicate that those with tertiary educated parents are more likely to have a highly rewarded occupation in all the countries except Germany. Moreover, the effect of parental education is greater on occupational status than on wages. The Karlson-Holm-Breen decomposition method shows that the type of qualification obtained contributes to the reproduction of social inequality in the labor market, but its mediating role is greater in Norway and smaller in Italy, with Spain in the middle. A discussion of the institutional differences between the countries tries to explain the sources of this variation. © 2013 International Sociological Association Research Committee 28 on Social Stratification and Mobility. Published by Elsevier Ltd. All rights reserved. Keywords: Higher education; Labor market returns; Social inequality; Stratification; Social origins

1. Introduction One of the major findings of social stratification research is that there is a significantly positive association between social origin and status attainment in modern societies and that this is mostly mediated by educational attainment (Blau & Duncan, 1967). Yet several studies showed that social origin had relevant effects over and above what was mediated by education (Breen & Jonsson, 2005; Breen, 2004; Erikson & Jonsson, 1998; Hansen, 2001). This is often referred to as ‘direct effects’

∗ Correspondence address: Via Bicocca degli Arcimboldi 8, 20126 Milan, Italy. Tel.: +39 026448 7504. E-mail address: [email protected]

(Mastekaasa, 2011). Furthermore, it has been shown that in the United States the direct relationship between social origin and the socioeconomic position attained varied across school levels. It is significant among low-educated people but not among college graduates (Hout, 1988). Research in Europe has also shown a weaker intergenerational association at higher levels of education, even if its strength varies across countries (Breen, 2004). These results were found by adopting an explorative approach and most refer to older cohorts. In this article, we aim to update and extend knowledge on the effect of social background on occupational outcomes among higher education graduates, adopting a comparative and theoretically-oriented approach. As recently suggested by Torche (2011), the focus on college graduates makes it possible to examine whether a higher education degree

0276-5624/$ – see front matter © 2013 International Sociological Association Research Committee 28 on Social Stratification and Mobility. Published by Elsevier Ltd. All rights reserved.

http://dx.doi.org/10.1016/j.rssm.2013.01.003

46

M. Triventi / Research in Social Stratification and Mobility 32 (2013) 45–63

represents a kind of ‘liberation’ from social background or, on the contrary, whether the resources associated with social origin continue to play a relevant role even after graduation. Moreover, the adoption of a comparative and a theoretically-oriented approach allows us to elaborate hypotheses on the cross-country variation in the role of social background in the labor market. In the second part of the article, we investigate the mechanisms whereby social inequalities are reproduced among tertiary graduates. In particular, we aim to understand whether the choice of different educational routes in higher education (what we refer to as ‘higher education institutional stratification’) may account for the heterogeneous occupational outcomes of those with a different social background. Since participation in higher education has rapidly expanded in the last decades, it is likely that differentiation of courses and degrees at this level becomes increasingly relevant for occupational outcomes. At the same time, people from better educated families could take advantage of institutional stratification to access the better quality and rewarded types of education, thus affecting the intergenerational reproduction of social inequality. In the next section we develop our theoretical framework, connecting and integrating several theories developed within the social stratification literature. In Section 3, the main institutional characteristics of the selected countries are described, whereas the research hypotheses are formulated in Section 4. The Section 5 describes data, variables and methods used in the empirical analysis, while Section 6 presents results from the analysis and the last section concludes. 2. Theoretical framework The starting point of our discussion is the negligible association between social background and occupational attainment among college graduates in the United States at the end of the 1980s (Hout, 1988). Several explanations were elaborated to account for such finding (Breen & Jonsson, 2005). The first is that meritocratic recruitment criteria prevail in the labor markets for tertiary graduates. According to this perspective, employers, when screening potential candidates for vacant positions, do not take into consideration ascriptive characteristics or other resources that are potentially associated with social background, such as cultural capital and social connections, but only their educational qualification and skills (Hout, 1988; Breen, 2004; Breen & Jonsson, 2005). This argument is in line with modernization and liberal industrialism theses, according to which the process of modernization should be followed by a general trend

from ascription to achievement as the main principle of allocation of individuals to different social positions (Treiman, 1970). According to this perspective, one crucial aspect that reduces reliance on ascriptive traits in the selection procedures is greater bureaucratization, which contributed to limit the scope of professions’ inheritance and the role of family ties in the job search. A second explanation of the null effect of social origins on occupational returns among college graduates refers to the social selection at earlier school stages (Boudon, 1974; Mare, 1980). At each successive educational transition, a smaller proportion of lower class students than upper class students decide to continue their studies. More specifically, those from socio-economically advantaged families are more likely to obtain a university degree than students from less advantaged families (Shavit & Blossfeld, 1993). The implication is that upper and lower class children passing through the various transition points become more and more similar in terms of unobserved characteristics, including ability, motivations and occupational aspirations, since only the more motivated and able lower class students reach higher school levels (Mare, 1980). If these unobserved attributes matter in the labor market, greater homogeneity of unmeasured factors at higher levels of schooling may reduce the effects of observed socioeconomic variables on occupational outcomes. At this point, it is useful to discuss whether and to what extent these two explanations may hold for the cohort analyzed in our study. The degree of selectivity of graduates is affected by overall participation in higher education and the degree of inequality in educational opportunities at lower education levels. The expansion of tertiary education in many European countries in recent decades had noticeable implications for the processes under discussion. First, there was an increase in the proportion of lower-class students who entered and completed higher education. This implies a reduction in the selectivity of tertiary studies and an increase in the heterogeneity of graduates in terms of unmeasured ability and aspirations. Second, it is likely that, in a context where a large number of students attend higher education, obtaining a tertiary degree is no longer sufficient to enter the best occupational positions. Conversely, it is likely that employers, when selecting potential candidates for vacant positions, evaluate additional ‘signals’, such as field of study, type of institution or program level (Gerber & Cheung, 2008; Torche, 2011; Triventi, 2011). Upper class families are particularly concerned about this issue because, since a greater number of people can acquire a higher education degree, this qualification

M. Triventi / Research in Social Stratification and Mobility 32 (2013) 45–63

is no longer sufficient to obtain a high-ranked social position, which is a key goal for these families (Breen & Goldthorpe, 1997).1 Furthermore, they may have more and better information on the relative occupational returns associated with different routes in higher education (Paulsen, 2001). Thus, it is reasonable to think that they would try to maintain their offspring’s advantages adopting strategic choices in the school system, as suggested by the effectively maintained inequality thesis (Lucas, 2001). According to this argument, when an education level is attended by a relatively small share of people, the socioeconomically advantaged use their advantages to secure that level of education. Once that education level becomes relatively widespread, they seek the qualitative differences in that education stage and use their advantages to secure quantitatively similar amounts of qualitatively better schooling. Thus, when a large share of people enter higher education, upper class families will choose the best educational options within this level for their children in order to maintain their relative advantages.2 This is in line with the ‘diversion hypothesis’ elaborated by Brint and Karabel (1989) which states that the existence of different routes in higher education (for example, universities and community colleges in the United States) produces a diversion of lower class students to the less prestigious and lower quality institutions and courses. Looking at higher education, two main axes of institutional stratification provide different resources and opportunities for their graduates (Charles & Bradley, 2002). Vertical stratification refers to distinct course levels or cycles which are arranged in a sequence; each cycle gives access to a higher degree and more years of education. Horizontal stratification includes at least two kinds of differentiation. The first is the different types of institutions or educational sectors that can be hierarchically classified on the basis of degree of selectivity, quality of instruction and academic prestige. The second refers to academic disciplines, fields of study or majors, which vary in their organization, type of knowledge, selectivity, 1 More precisely, according to the relative risk aversion hypothesis, the main goal in education decisions for all individuals is to avoid social demotion. Since people occupy different positions in the social structure, lower class children who make the transition to university have just reached this goal, while this is not the case for upper class children. Thus, if higher education is stratified in programmes and institutions that convey different occupational returns, the choice of the best educational routes becomes important for upper class children in order to reduce the risk of social demotion. 2 For a deeper discussion on the strategies adopted by socioeconomically advantaged people to secure the attainment of the best qualifications within higher education see Alon (2009).

47

academic prestige and economic payoffs (Bourdieu, 1996; Van de Werfhorst & Kraaykamp, 2001). Given the existence of several lines of institutional stratification in European higher education systems (Teichler, 1988) and differentiated occupational returns linked to different fields of study, course levels and types of institutions (Barone & Ortiz, 2011; Brunello & Checchi, 2007; Gerber & Cheung, 2008; Holmlund, 2009; Reimer, Noelke, & Kucel, 2008; Van de Werfhorst, 2004), those from upper social classes can take advantage of the best educational options to make the transition to the labor market with better rewarded credentials. The literature also identifies a second relevant mechanism whereby the upper class may try to maintain its advantages: social capital and social networks. On the one hand, it has been shown that social origin is related to available social resources, i.e. resources accessed in social networks (Lin, 1999). On the other, a large body of literature shows that social resources are associated with several outcomes, such as a high-ranked social position (Bourdieu, 1979; Granovetter, 1973). Because those from the upper classes have larger social networks and ties to more influential people, they may gain advantages when looking for a job or when in competition over promotion. For instance, they have more information on job opportunities and their parents can mobilize their social contacts in order to smooth the transition to the labor market or facilitate job change in order to obtain a better occupational position.3 In this work the use of social network is considered as the main alternative way (compared to the choice of higher education route) to reproduce social inequality in the labor market, even if its contribution will not be empirically assessed. 3. The main institutional features of the selected higher education systems Following the discussion in the previous section, we argue that several macro-factors are likely to affect the cross-country variability in the phenomena of our interest. The most important are: (i) social selectivity of

3 Another way by which graduates from the upper classes could maintain their advantages in the labor market is the development of cognitive and non-cognitive skills. There is a widespread consensus that both these types of skills have important effects on school attainment and on labor market outcomes and that they are affected by parental education (Brunello and Schlotter, 2011). Nevertheless, given the difficulty to accurately measure them and the lack of theories on their variation across institutional contexts, their effects over and above those incorporated in formal qualifications will not be considered in this article.

48

M. Triventi / Research in Social Stratification and Mobility 32 (2013) 45–63

education system, (ii) level of higher education expansion, (iii) degree of institutional connections between higher education and labor market, and (iv) degree of institutional stratification in higher education. The strength of social selectivity of the education system affects the proportion of lower class individuals who are able to pass each school transition and, at the end of the process, attain a tertiary education degree. The higher is the social selectivity of the education system, the lower is the proportion of graduates from the lower classes and, ceteris paribus, the smaller is the effect of social background on occupational outcomes. The second macro-characteristic is the level of higher education expansion, which can be measured as the proportion of graduates in a given cohort. All else being equal, countries with a mass higher education system may be characterized by stronger competition among graduates in the transition to the labor market and, thus, by a larger effect of social background on occupational returns. The third factor is the degree of institutionalization of the links between higher education and the labor market. It refers to the extent to which graduates rely on formal (employment agency, work placement, advertisement in newspaper, etc.) or non-formal channels (help of parents, family or friends) to find a job (Reyneri, 2005). Great reliance on formal channels distinguishes countries with highly-institutionalized links between higher education and the labor market, whereas the use of social networks characterizes systems with little institutionalization. We expect the role of social origin to be greater in the latter contexts because there is more room for family actions to help their children to find a job or obtain a promotion.4 At the end, stratification of higher education refers to the degree of variation in selectivity, quality/prestige and labor market value of different courses, fields of study and institutions. All else being equal, the higher is the stratification of higher education, the more important is the role of social background in the occupational attainment process. We chose to focus our attention on four European countries – Germany, Italy, Spain and Norway – since they have interesting profiles with regard to the institutional features that we have briefly presented. We believe

4 Assuming that social resources matter in the labor market, this could occur if at least one of these conditions holds: (1) upper class families have larger social networks than lower class families; (2) upper class and lower class families have similar amounts of social contacts, but the former have better social networks (connections to more influential people) than the latter, or (3) they are more able to mobilize their networks to obtain a given goal.

that the analysis of countries with different institutional profiles can be useful to illustrate and better understand the variation of social inequality reproduction in the labor market among recent graduates. Table 1 provides a synthetic description of the macro characteristics of the four countries considered. As reported in the note, it is based on information taken from existing national studies or from our own calculation on available data sources. Let us now discuss briefly the institutional profiles of the countries included in our study. Germany has a relatively low graduation rate and a highly selective school system: only a small proportion of students obtains a tertiary degree in recent cohorts (15%) and inequalities in tertiary degree attainment between social groups are extremely large. This could be due, at least partially, to some specific features of the German education system. There is a strongly stratified secondary school, the tracking of pupils occurs earlier than in other countries (around 11) and there is a binary higher education system made up of Universitäten (universities) and Fachhochschulen (universities of applied sciences) (Mayer, Müller, & Pollak, 2007). Furthermore, looking at the transition to occupation, there is a medium to high degree of institutionalization of the links between higher education and the labor market because only a small proportion of graduates finds their first job through social contacts (9%). Italy is characterized by a low graduation rate (15%) and a relatively selective school system: pupils are channeled in different tracks at the age of 14 and, even if all those with a 5-year diploma can access university, the large majority comes from the academic track. Social selection occurs not only in the transition from secondary to tertiary education, but also in the form of dropping out during the early years of tertiary studies (Argentin & Triventi, 2011) and only a small proportion of students obtains a tertiary degree (Recchi, 2007; Triventi & Trivellato, 2009). Inequalities in tertiary education attainment are rather great and much similar to what found in Germany (Triventi, 2012). Italian higher education can be considered as a ‘unitary’ and undifferentiated system (Shavit, Arum, & Gamoran, 2007) since all the institutions are universities and the proportion of students enrolled in short vocational programs was very small in the second half of the 1990s. Moreover, there is a ‘legal value’ of a university degree, according to which all graduates in the public sector are considered at the same level, irrespective of the university where they obtained their degree. The transition to the labor market is relatively long (even if slightly shorter than in Spain) and the degree of institutionalization is rather limited, since a noticeable percentage of graduates (more than

M. Triventi / Research in Social Stratification and Mobility 32 (2013) 45–63

49

Table 1 Synthetic description of the main institutional characteristics of the countries examined.

Germany Italy Norway Spain

Social selectivity in higher education attainment

Higher education attainment

Higher education stratification

Degree of institutionalization HE-LM

High High Low Medium

Low Low High Medium-High

Medium Low High Medium

Medium-High Low High Low

Note: Social selectivity in higher education attainment: Odds ratio tertiary education attainment vs less than tertiary, comparing people with tertiary educated parents and those with up to lower secondary education (people born between 1969 and 1979) (European Social Survey, 2002–2008). Additional sources: Kouck´y et al. (2009) and Triventi (2012). Higher education attainment: % of Isced 5A graduates among 25–34 years old (OECD, 2007). Higher education stratification: qualitative information from national studies. Degree of instituzionalization higher education-labor market linkage: 100-(% of graduates who found the first job through family and friends) (Reflex, 2005).

20%) finds their first job thanks to the help of family or friends (Reyneri, 2005). The Spanish educational system has experienced major changes since the 1980s: the age at which pupils are divided into different tracks has been raised from 14 to 16 (Brunello & Checchi, 2007) and there was a reduction of educational inequalities related to social background in the second half of the 20th century (Ballarino, Bernardi, Requena, & Schadee, 2009). Overall, access to higher education is fairly open and there are no strict limits for staying enrolled in the programs (Mora, Garcia-Montalvo, & Garcia-Aracil, 2000). Spain experienced a rapid higher education expansion in the last decades and in recent cohorts there is a relatively large percentage of graduates (27%). Social inequality in tertiary degree attainment is still high but lower than in Germany and Italy. The Spanish higher education system is almost exclusively made up of universities and there is no strong institutional differentiation. Nevertheless, there is a medium to high level of differentiation in university programs, which are characterized by different length, prestige and selectivity (Mora et al., 2000). Furthermore, institutionalization between higher education and the labor market is limited, since more than a quarter of graduates use their social resources to find a job. Spain also stands out among the European countries analyzed when it comes to the length of the transition to the labor market and the incidence of over-education (Teichler, 2002; Barone & Ortiz, 2011). Finally, Norway implements late tracking in secondary education (at 16) and the flexibility of the higher education system facilitates access for mature students and student mobility between study programs. Given these institutional arrangements, around 40% of those between the ages of 25 and 34 obtained a tertiary type A degree in 2005, a proportion that places Norway at the top of European countries. Furthermore, the strength of social inequalities in tertiary education attainment is

smaller than in the other countries considered in our research (Kouck´y, Bartuˇsek, & Kovaˇrovic, 2009). The higher education system was developed with a great variety of supply in terms of type and length of courses (Arnesen, 2000). It is divided into a university and a non-university sector: the first mainly offers long courses which last for four to seven years and the second mainly offers two- to three-year courses (with some exceptions). The transition to the labor market is shorter than in the Southern European countries and it is more institutionalized, since only a small proportion of graduates finds their first job through social networks (around 5%). Moreover, more than half the graduates is employed in the public sector, and more especially in the education, welfare and health related sectors. 4. Research hypotheses It is important to note that, since we focus on a limited number of countries, we cannot formally test the role of each macro-characteristic in modifying social inequalities in occupational returns. Nonetheless, some of the macro-factors are highly correlated and, therefore, even with a larger number of countries it would be difficult to discern the effect of each single characteristic. Our strategy is to rely on the rather distinct institutional profiles of the countries included in this study, in order to assess how the different combinations of macro characteristics lead to differentiated patterns of social inequalities in early labor market returns of tertiary graduates. Following the discussion of the theoretical framework and the description of the main institutional features of the countries under scrutiny, we are now able to formulate specific hypotheses. First, we focus on a recent cohort, where the graduation rates in tertiary education are higher than in the past. Hence, we expect there to be a significant heterogeneity among graduates in terms of social background.

50

M. Triventi / Research in Social Stratification and Mobility 32 (2013) 45–63

Furthermore, simply obtaining a tertiary degree could be no longer sufficient to find a high-ranked occupation. Therefore, it is likely that upper class families will activate their resources to help their children to obtain a prestigious and well-remunerated occupation, giving them an additional advantage over graduates from lower social backgrounds. Given these considerations, we hypothesize a significant total effect of social background on occupational returns among graduates (hypothesis 1). Nevertheless, we hypothesize a certain degree of heterogeneity in the strength of the relationship between social background and occupational returns across institutional settings. As anticipated, we expect that in countries with a larger participation in higher education there is stronger competition among graduates to access the best positions in the labor market and, consequently, upper class students would try to maintain their advantages entering the most remunerative and prestigious study programs and institutions. Moreover, it is likely that social background will play a more relevant role when finding a job through social contacts is relatively widespread and when higher education is stratified in programmes with different prestige/quality, because both mechanisms of social inequality reproduction (see Section 2) could be at work. Therefore, we expect the association between social background and occupational returns to be stronger in Spain, where there is a large number of graduates, more people rely on personal contacts to find a job and there is a considerable programmes differentiation in higher education. On the contrary, social inequality should be smaller in Germany, because of the high social selectivity in tertiary degree attainment and small graduation rate, and also because social contacts are less important in the job search. The other two countries, Italy and Norway, have more mixed characteristics in this respect and, therefore, should be placed somewhere in the middle (hypothesis 2). The second part of the analysis assesses whether the choice of routes in higher education accounts for differentiated labor market outcomes between graduates from highly educated and less educated families. Institutional stratification could be a relevant mediator if, at the same time, social background significantly affects graduation from the best educational courses and institutions, and higher education qualifications are differently rewarded in the labor market. Previous research showed that social origin affects the choice of field of study, program length and type of institution in several European countries (Berggren, 2008; Duru-Bellat, Kieffer, & Reimer, 2008; Reimer & Pollak, 2010; Triventi, 2011). Moreover, the type of education and qualification has been shown to

be related to both symbolic and monetary occupational returns (Brunello & Cappellari, 2008; Reimer et al., 2008). Therefore, we expect institutional stratification to play a relevant role in mediating the relationship between social origin and occupational outcomes (hypothesis 3). Yet it is likely that the importance of higher education differentiation in the reproduction of social inequality in the labor market among graduates will vary according to several institutional characteristics. First, following the effectively maintained inequality thesis and the diversion argument, the horizontal differences in tertiary education will become more relevant if access to higher education increases and there is heterogeneity in the prestige/quality of programmes and institutions. As previously argued, students from well-educated families could be more aware of the institutional differences and opt for the educational routes considered as more prestigious or remunerative. Second, the role of institutional stratification should be stronger when social contacts are of minor importance in the job search and other institutionalized channels prevail (competitive entrance examinations, public and private agencies, etc.). Indeed, it is likely that employers who rely on formal channels to recruit employees will use educational credentials and type of education as important screening devices. Therefore, we expect the role of institutional stratification as mediator of the relation between social origin and occupational outcomes to be greater in Norway because there is a large proportion of tertiary graduates, the higher education system is highly diverse, and the connections between tertiary education and labor market are more institutionalized. On the other hand, its role should be smaller in Italy, where graduation at tertiary level is still not widespread, social networks are important for finding a job and the institutional differentiation in the higher education system is limited, especially compared to the other countries considered here (hypothesis 4). 5. Data, variables and methods 5.1. 5.1 Data and variables Data from the ‘Research into Employment and professional FLEXibility’ survey are used. Reflex is a harmonised cross-section sample survey carried out in 2005/2006 on students from several European countries who graduated from tertiary institutions (Isced 5A) in the year 2000. The sample design is stratified and the strata include categories such as region and sector of higher education, depending on the national context. Survey weights that re-proportion the sample according to population figures are used in all estimations. Most of

M. Triventi / Research in Social Stratification and Mobility 32 (2013) 45–63

the graduates answered a written questionnaire (around 75%), while a minority was interviewed by telephone or filled in a web questionnaire.5 A comparison of Reflex with the European Labour Force survey showed no major discrepancies in the distribution of the basic variables (Barone, 2011). The Reflex dataset is used because it contains unique information on social origin, higher education qualifications and labor market outcomes in a comparative perspective, which are not all available in standard population surveys. Four countries are included in this analysis: Germany (DE), Spain (ES), Italy (IT), Norway (NO). Those over 35 years of age at the time of graduation were not included in the analysis to make the sample more homogeneous. Furthermore, the self-employed (with more than one ‘client’) were not included because the theoretical mechanisms discussed above were less applicable.6 The sample includes approximately 6700 cases. We used two alternative measures of socioeconomic standing as dependent variables: socioeconomic status and hourly wage related to the occupation held by graduates five years after graduation.7 Since in our theoretical framework we make reference to the process of competition among social groups in accessing the best occupational positions, in the empirical analysis we analyze being in the top part of the distribution of occupational returns as dependent variable. This allowed us to assess whether those with a higher social background were more likely to enter the most prestigious and remunerative occupations available to tertiary graduates (which can be considered a relatively good proxy of the best occupational positions in a country) in order to avoid downward mobility.8 We used the International SocioEconomic Index developed by Ganzeboom, De Graaf, and Treiman (1992) for occupational status and gross

5 Additional information is available on the project website (http://www.fdewb.unimaas.nl/roa/reflex/) and in Allen and van der Velden (2007). 6 Nevertheless, as a robustness check, we carried out the same analyses presented in this article on the whole sample of graduates, including the self-employed, and the results are very similar. 7 We focus on the occupation five years after graduation because it is a more ‘stable’ indicator of the position attained by graduates compared to their first occupation. However, the analyses of the first job produced substantially similar findings to those presented here. 8 Nevertheless, we repeated the analyses using the two dependent variables as metric variables. The main pattern of results remains unchanged, with some minor exceptions. This means that the basic mechanisms of reproduction of inequality does not only work at the top of the distribution of occupational returns, but also at their mean.

51

hourly wage to measure economic returns.9 For both variables, we created corresponding dummy variables which indicate if graduates fell in the highest quartile of the distribution in their country. The first variable accounting for institutional stratification is years of higher education attained, which is a synthetic measure of vertical differentiation and considers the length of the course and if the individual obtained additional post-graduate qualifications.10 The second variable is graduation from a top institution, defined on the basis of an overall measure of ‘institutional quality’ elaborated by Triventi (2011). It includes three components: the degree of selectivity at entrance, the quality of student intake and the quality of occupational outcomes.11 The final version of this variable indicates whether the respondent graduated from a university ranked in the highest quartile of the quality rank. The third variable is field of study, divided into seven categories: social sciences (psychology, sociology, political science), humanities (literature, arts, teacher training, foreign languages, education), economics (including business and administration), law, science (physics, mathematics, biology, chemistry, computing), technical disciplines (engineering, architecture), and health (medicine, other health-related courses). The main independent variable is a combination of the highest educational level attained by graduates’ parents, classified in four categories: (a) no more than one parent with an Isced 3–4 qualification; (b) both parents with an Isced 3–4 qualification; (c) one parent with an Isced 5–6 degree; (d) both parents with an Isced 5–6 degree. This classification accounts for the distribution of parental educational qualifications across countries in order to have a sufficient number of cases in each category.12 Unfortunately, in the Reflex dataset no variables of social 9 To use only reliable information we used a truncated version of this variable, which does not include outliers (around 1–2%). 10 We include this metric variable for simplicity and parsimony; using a categorical variable does not change substantially the results. 11 The first dimension is derived from the proportion of graduates who entered university after taking a special entrance exam and the average number of selection criteria applied to select students; the second dimension includes the proportion of graduates with a high final mark in secondary education, from the academic track and with tertiary graduated parents; the third dimension considers the proportion of graduates who enter a PhD course, average occupational status and monthly wage 5 years after graduation. For more information please refer to Triventi (2011), in particular to the ‘data and variables’ section and Table 1. 12 An alternative classification considering graduates whose both parents have less than Isced 3 as a separate category was used in previous analyses. Results are rather similar, but the percentage of cases in this additional category is too small for Germany (2%).

52

M. Triventi / Research in Social Stratification and Mobility 32 (2013) 45–63

class of origin or occupational background are available. For this reason, parental education is considered as a general indicator of social background. Additional control variables included in the models are gender, age, whether at least one parent is born abroad and whether graduates maintained the same job as before graduation. 5.2. Methods Binomial logistic regression models were used to estimate the partial association between parental education and the outcomes of interest (top quartile of occupational status and wage). Two models for each dependent variable were estimated: the first contained only parental education and the basic control variables (sex, age, parents born abroad, maintained the same job), the second added the three variables of institutional differentiation in higher education (years of higher education, top institution, field of study). The first model allowed us to estimate the total effect of social origin on the outcomes under scrutiny, while the second model estimated the residual effect of social origin, once controlled for type of education attended. Moreover, they gave information on the effects of higher education qualifications, net of relevant antecedent variables. These last results should be interpreted as mainly descriptive, because we did not formally address issues of endogeneity (students selfselect themselves into different higher education routes) and sample selection bias (not all graduates are employed five years after graduation).13 Since we compared results of logistic regression models in different countries, logit coefficients or odds ratios derived from them could lead to misleading conclusions if models widely differed in their residual variability (Allison, 1999; Mood, 2010). In order to partially tackle this issue, we used average partial effects as a measure of partial association, which indicates the average difference between two categories in the probability of interest, net of other control variables (Long, 1997). The aim of the second part of the analysis was to establish to what extent the association between parental education and occupation outcomes was mediated by the type of qualification acquired in tertiary education. In nonlinear regression models, decomposing the total effect of a variable of interest into direct and indirect effects is not as straightforward as in linear regression 13 Nevertheless, this last aspect should be mitigated by two facts: (1) the proportion of employed graduates five years after graduation is close to 90% in most of the countries examined; (2) the proportion of graduates who are not employed five years after graduation does not vary significantly across students from different social origins.

models. It is not possible to simply compare the estimated coefficient of the variable of interest between a reduced model without the mediator variables and a full model with these mediator variables (Karlson & Holm, 2011). In fact, the estimated coefficients of these models are not comparable between different models because of a rescaling of the model induced by a property of nonlinear regression models: the coefficients and the error variance are not identified separately (Allison, 1999; Mood, 2010). In order to tackle this issue, we applied the KHB-method, a recent technique which allows the decomposition of total effects into direct and indirect effects for several types of generalized linear models (Breen, Karlson, & Holm, 2010). This method compares the full model with a reduced model that replaces the mediators by the residuals of the mediators from a regression of the mediators on the key independent variable. Hence, it allows the separation of the change in the coefficient that is due to confounding (what is of substantive interest) and the change that is due to rescaling (Kohler, Karlson, & Holm, 2011). 6. Empirical results 6.1. Social origin and institutional effects on occupational outcomes Table 2 provides descriptive statistics for the main variables. First of all, the two dependent variables (top ISEI and top wage) and graduation from a top institution show little cross-country variation by design because they include approximately the highest quartile of graduates in each country.14 On the contrary, the distribution of parental education is highly differentiated across systems; the proportion of those whose parents have no more than one Isced 3 qualification exceeds 60% in Spain and 50% in Italy, while it is around 33% in Norway and less than 7% in Germany. Conversely, the proportion of graduates whose parents both have a tertiary degree exceeds 25% in Germany and Norway, while it is less than 12% in the two Mediterranean countries. This variation is due to both the distribution of educational qualifications in the cohorts of graduates’ parents and the social selection of students at earlier school stages. The average years of higher education attained are lower in Norway (3.8) and Spain (4.2), while they are higher in Italy and Germany (4.8), where most students graduated in long courses. Looking at the distribution of

14 Deviations from the precise quartile (25%) are due to rounding and to the idiosyncratic distribution of cases on specific values.

M. Triventi / Research in Social Stratification and Mobility 32 (2013) 45–63

53

Table 2 Distribution of the main dependent and independent variables according to country (column %). Germany

Italy

Norway

Spain

Total

ISEI Lower ISEI Top quarter ISEI

81.6 18.4

74.6 25.4

81.5 18.5

75.7 24.3

77.5 22.5

Wage Lower wage Top quarter wage

77.1 22.9

75.9 24.1

76.0 24.0

77.3 22.7

76.7 23.3

Social origin One secondary or less Both secondary One tertiary Both tertiary

6.7 26.1 35.1 32.2

55.0 23.0 12.9 9.1

32.8 14.6 27.0 25.5

63.0 8.1 17.0 11.9

46.2 15.9 20.8 17.1

Years of HE attained (mean)

4.84

4.78

3.81

4.20

4.36

74.3 25.7

75.2 24.8

80.4 19.6

79.5 20.5

77.8 22.2

9.9 22.1 16.2 5.2 14.1 24.2 8.3

12.4 16.3 21.3 7.6 11.7 19.0 11.6

13.7 28.1 10.2 2.8 10.3 12.0 22.9

12.6 21.6 20.5 6.4 14.5 14.5 10.0

12.4 21.7 17.9 5.8 12.9 16.6 12.8

1043

1703

1383

2635

6764

Institution quality Medium-Low level institutions Top institutions Field of study Social sciences Humanities Economics Law Natural sciences Technical Health Sample size

fields of study across countries, we note that the share of humanities is larger in Norway and lower in Italy, while the opposite is found for law, with Spain and Germany placed in the middle. Germany has the largest proportion of graduates in technical disciplines, while in Norway 23% graduated in health-related subjects, a very high proportion compared to that of the other countries (8–12%). The results of the multivariate analysis are shown in graphic form in order to facilitate the interpretation of the main findings, while results in tabular forms can be found in Appendix A. Fig. 1 presents the estimates of the average partial effects and 95% confidence intervals from binomial logistic regression models predicting the probability of entering a top ISEI occupation (upper graph) and a top wage occupation (lower graph) in each country. The estimates correspond to average differences in predicted probabilities between the categories represented in the graphs and the omitted reference category (individuals whose parents have no more than one upper secondary diploma). Filled dots represent the estimates obtained controlling only for socio-demographic variables (Model 1), while hollow circles refer to the estimates from models that also include the type of higher education (Model 2).

The main results are as follows. In three out of the four countries, there is a statistically significant and substantially relevant association between parental education and the likelihood of being in the top quartile of the status and wage distribution. Germany is the only country where there is no association between social background and labor market outcomes among recent graduates: no comparison between social categories is statistically significant or substantially relevant, considering both types of returns. In the other three countries, graduates with highly educated parents (at least one or both parents with a tertiary degree) are more likely to obtain a high-status and highly paid occupation than students with less educated parents. Despite the social selection at earlier school stages, the differences across categories are noteworthy: the largest average partial effects amount to 15–20 percentage points. Looking at the cross-country heterogeneity, the advantage linked to social origin is stronger in Spain, followed by Italy and Norway in the middle, while Germany comes last. Nevertheless, because of the great uncertainty, the confidence intervals partially overlap and the differences between countries are not significant. Furthermore, in all these countries, the association between social background and occupational status is stronger than the association

54

M. Triventi / Research in Social Stratification and Mobility 32 (2013) 45–63

Top ISEI DE

ES

IT

NO

−.1 0 .1 .2

−.1 0 .1 .2

−.1 0 .1 .2

−.1 0 .1 .2

DE

ES

IT

NO

−.1 0 .1 .2

−.1 0 .1 .2

−.1 0 .1 .2

−.1 0 .1 .2

Both secondary One tertiary Both tertiary

Top wage

Both secondary One tertiary Both tertiary

Fig. 1. Binomial logistic regression models predicting entrance into a top status occupation (upper graph) and in a top wage occupation (lower graph): average partial effects and 95% confidence intervals related to parental education. Filled dots indicate estimates from models with only socio-demographic controls; hollow circles indicate estimates from models which also control for types of higher education qualification.

with wage. This could be due to the use of parental education as unique indicator of social background. Indeed, it is likely that symbolic rewards in the labor market are more related to proxies of cultural capital, while monetary returns to proxies of parental class and income. This interpretation is in line with the notion of differentiated hierarchies of power outlined by Bourdieu (1979, 1996), in particular those related to the distinction between cultural and economic capital. The second models show that controlling for the detailed type of qualification attained in higher education reduces the magnitude of the average partial effects, making most of the comparisons between social categories not significant. This is a preliminary sign that differentiation in higher education could mediate, at least partially, the relation between social background and occupational outcomes, but we will formally test this hypothesis below. 6.2. The role of institutional stratification in mediating the relation between parental education and occupational outcomes Before investigating how far the relationships between parental education and occupational outcomes are mediated by the type of qualification attained, it is useful to study separately the two relationships involved

in this decomposition analysis. First, we analyze the association between parental education and the three variables that identify the type of qualification obtained. The first column in Table 3 shows that in all the countries there is a positive association between parental education and years of higher education attended, but this is greater in Norway and Spain, and smaller in Germany and, more especially, Italy. The probability of obtaining a degree from a top institution (Table 3, column 2) is also unevenly distributed across social groups, being greatest for graduates whose both parents have a tertiary degree. This pattern is found in all the countries examined, but it is more marked in Norway and less evident in Spain. The remaining columns in Table 3 show the association with the field of study; it is interesting to note that some subjects display the same pattern across the four countries, while others have some peculiarities. In all the countries, there is a negligible or small association between parental education and graduation in technical, scientific and economic disciplines, whereas students with better educated parents are more likely to study law. Graduates from lower educated families are more likely to enroll in humanities in all the countries except Germany, where they are more likely to study social sciences. There is heterogeneity in the relationship between parental education and graduation in health subjects:

M. Triventi / Research in Social Stratification and Mobility 32 (2013) 45–63

55

Table 3 Association between parental education and years of higher education attained (mean), graduation from a top institution (%), and field of study (%). Years of HE (mean)

Top institution (%)

Field of study (%) Social sciences

Humanities

Economics

Law

Science

Technical

Health

Germany One secondary or less Both secondary One tertiary Both tertiary

2.7 2.7 2.8 3.1

24.3 13.1 23.1 35.2

16.3 12.3 11.5 10.0

18.3 26.3 21.1 21.7

15.3 15.5 16.9 11.4

2.1 3.8 4.6 8.6

15.4 14.4 15.3 15.0

24.0 25.9 23.2 19.8

8.6 1.7 7.5 13.5

Italy One secondary or less Both secondary One tertiary Both tertiary

2.8 2.8 2.9 3.0

20.1 26.7 26.9 38.4

14.0 11.9 10.5 11.3

20.2 16.0 19.0 14.5

20.0 21.5 16.8 17.3

7.8 13.9 11.7 16.4

10.1 9.7 13.2 12.1

16.9 15.4 16.9 16.1

11.1 11.6 11.8 12.3

Norway One secondary or less Both secondary One tertiary Both tertiary

1.6 1.6 1.9 2.4

13.8 17.2 21.0 35.0

15.6 16.5 13.7 16.1

29.7 28.3 29.9 23.7

8.7 13.4 9.6 8.9

2.3 1.0 5.3 7.7

6.4 10.5 8.4 10.9

9.2 10.8 11.9 10.7

28.0 19.4 21.3 22.0

Spain One secondary or less Both secondary One tertiary Both tertiary

2.1 2.4 2.4 2.6

15.8 21.5 22.5 29.4

13.8 12.6 12.1 11.3

24.4 23.7 20.8 15.5

22.1 17.1 14.6 17.6

5.9 6.6 10.0 12.0

13.4 16.5 16.6 17.1

12.3 13.5 14.9 14.3

8.0 10.0 11.0 12.2

Parental education

in Germany, there is a weakly positive association, whereas it is null in Italy and negative in Norway. This is probably due to heterogeneous types of courses in this sector, which includes both the prestigious programs of medicine and other more ‘technical-oriented’ courses. We shall now look at the association between different types of degrees and occupational outcomes in order to understand whether graduates from different courses, institutions and disciplines are differently rewarded in the labor market five years after graduation, controlling for socio-demographic variables. Fig. 2 indicates that among dependent employees occupational rewards are strongly related to field of study and years of higher education, whereas institutional quality is less important. The second general finding is that the institutional stratification in higher education is more closely related to the likelihood of entering a top ISEI occupation rather than a top wage occupation. Therefore, educational stratification seems to matter more for the symbolic than for the economic side of occupational returns among dependent employees. Looking at more detailed results, years of higher education positively affect entrance in a top ISEI occupation in all countries except Germany. They also have a positive but lesser effect on the probability of entering a top wage occupation in all the countries, even if the estimates are not statistically significant for Italy. Graduation

from a top institution is weakly associated with both outcomes, with one exception: it positively and significantly affects the likelihood of entering a top status occupation in Norway (10 percentage points). In order to understand in which countries field of study is more important in the labor market, we developed a summary measure consisting in the simple mean of the absolute value of the average partial effects.15 This measure can be read as the average difference in the probability of being in a high status and highly paid occupation of graduates in humanities, economics, law, sciences, technical subjects and health compared to those in social sciences. The result of this exercise is presented in the last row of Fig. 2; standard errors around this measure have been computed using the ‘delta method’.16 Looking at entrance in the top quartile of the occupational status distribution, field of study is better rewarded in Germany, followed respectively by Italy and Norway, and finally Spain. But the average association between fields of study and entrance in a top wage occupation

15 The absolute value has been calculated because some of the average partial effects used in the computation are positive while others are negative, indicating that social science graduates are not always the most disadvantaged. 16 A detailed explanation of this method can be found at the following link: http://www.stata.com/support/faqs/stat/deltam.html.

56

M. Triventi / Research in Social Stratification and Mobility 32 (2013) 45–63

HE years Top ISEI

Top wage

DE

DE

ES

ES

IT

IT

NO

NO −.1

0

.1

.2

.3

−.1

0

.1

.2

.3

.2

.3

Top institution Top ISEI

Top wage

DE

DE

ES

ES

IT

IT

NO

NO −.1

0

.1

.2

.3

−.1

0

.1

Field of study Top ISEI

Top wage

DE

DE

ES

ES

IT

IT

NO

NO −.1

0

.1

.2

.3

−.1

0

.1

.2

.3

Fig. 2. Binomial logistic regression models predicting entrance into a top status occupation (left graphs) and in a top wage occupation (right graphs): average partial effects and 95% confidence intervals related to years of higher education attained (first row), graduation from a top institution (second row) and field of study (third row).

is smaller in Spain than in Norway, while Germany and Italy are in an intermediate position, with largely overlapping confidence intervals. A more detailed picture of the advantages and disadvantages associated with graduation in different

disciplines is presented in Fig. 3 where the average partial effects associated with each field of study are showed. Even if it is not simple to find a common pattern, we can see that graduates in health disciplines are generally those with better rewards both in terms of occupational

M. Triventi / Research in Social Stratification and Mobility 32 (2013) 45–63

57

Top ISEI DE

ES

IT

NO

Social sciences Hum−Educ Economics Law Science Technical Health −.20 .2.4.6.8 −.20 .2.4.6.8 −.20 .2.4.6.8 −.20 .2.4.6.8

Top wage DE

ES

IT

NO

Social sciences Hum−Educ Economics Law Science Technical Health −.1 .1 .3 .5 −.1 .1 .3 .5 −.1 .1 .3 .5 −.1 .1 .3 .5 Fig. 3. Binomial logistic regression models predicting entrance into a top status occupation (upper graphs) and in a top wage occupation (lower graphs): average partial effects and 95% confidence intervals related to fields of study.

status and wages. Also, science graduates obtain better occupational returns than social science graduates – both in terms of status and wage, even if some comparisons are not statistically significant. Graduates in economics and technical disciplines have better monetary returns than social sciences graduates, whereas graduates in law have better outcomes only in terms of status attainment. This could be due to the fact that in several European countries law is a discipline with postponed high earning returns. Their transition to the labor market is longer than that of other graduates, because they usually have to attend an additional period of ‘training on the job’ in a lawyer’s office with long working hours and a relatively low wage. In general, the association between graduation in such fields and labor market outcomes is greater in Norway and Germany than in the Mediterranean countries. As we have seen, data from the Reflex survey suggest that different dimensions of institutional stratification in higher education matter for labor market returns. Moreover, their introduction in the regression models substantially reduced the associations between parental education and occupational outcomes. This comparison, however, is not sufficient if we aim to measure to what extent the types of higher education qualifications account for differences in occupational returns between

graduates with different social origins. To tackle this issue, we applied the KHB method, which quantifies how much of the differences in occupational outcomes between graduates with different levels of parental education is (statistically) explained by years of higher education, type of institution attended and field of study. Fig. 4 reports the most relevant results of this decomposition, showing the proportion of the gap in occupational outcomes accounted for by institutional stratification in higher education, comparing individuals whose parents both have a tertiary degree and those whose parents have no more than one upper secondary diploma. We focus on this comparison because it has been found to be statistically significant in most countries for both outcomes. Since there is no association between parental education and labor market returns among recent tertiary graduates in Germany, we did not include it in this analysis. More detailed results are reported in Table A5 in Appendix A. The main results are as follows. First, the type of tertiary education qualification has a significant effect in mediating the association between parental education and both occupational outcomes. This suggests that institutional stratification has also mattered for social inequality reproduction among the highly educated in recent years. Second, as anticipated, there is crosscountry variation in the extent to which institutional

58

M. Triventi / Research in Social Stratification and Mobility 32 (2013) 45–63

Top ISEI

Top wage

0.20

0.20

0.15

0.15 67

0.10

0.10 105 45

0.05

0.05

60 46

0.00

11

0.00 ES

IT

NO

Explained

ES

IT

NO

Not explained

Fig. 4. Decomposition of the difference in the probability of being in a top ISEI occupation (left graph) and in a top wage occupation (right graph) between graduates whose parents are both graduated and those from the lowly educated families. The dark bar indicates the amount of the average partial effect accounted for by the type of qualification acquired in higher education (years of higher education, top institution, field of study).

stratification mediates this relation. Looking at both outcomes, we observe the same order of countries: the importance of institutional stratification is higher in Norway, followed by Spain and Italy last. Nevertheless, it seems that higher education differentiation is more important in the reproduction of social inequality as far as status attainment is concerned. In fact, it accounts for the entire gap between social categories in Norway, about 70% in Spain and 45% in Italy. But institutional stratification of higher education is able to explain a smaller portion of the gap in the probability of being in the highest quartile of the wage distribution: in Norway its contribution is 60%, in Spain it is around 45%, while in Italy it is remarkably lower (11%) and not statistically significant. 7. Discussion and conclusion The main aim of this article was to examine whether, among recent tertiary graduates, social origin was related to different labor market returns, both in terms of occupational status and hourly wage. We also aimed to understand whether graduation in different programs, fields of study and institutions was able to account for differences in occupational outcomes among those from different social backgrounds. Furthermore, adopting a comparative perspective, we tried to understand whether these phenomena varied across countries with heterogeneous institutional settings.

First of all, data from the Reflex survey indicate that, among recent cohorts, graduates from a different social background are not as likely to be in the top distribution of occupational status and wage. Our first hypothesis is thus confirmed, because those with tertiary educated parents are far more likely to enter a well-rewarded occupation than those with less educated parents. This result holds for Spain, Italy and Norway. Moreover, we showed that the strength of the social origin effect varied across countries with diverse institutional contexts, in terms of social selectivity in tertiary education attainment, higher education expansion and stratification, and the strength of the institutional mechanisms which connect tertiary education with labor market. The results partially corroborate our second hypothesis because the lowest association (nearly zero and not significant) between parental education and entrance in a top occupation is found in Germany, a country with a large social selection at lower educational stages, relatively high institutionalized connections between higher education and labor market, and small competition among graduates in the transition to the labor market. As expected, the largest effect of parental education is found in Spain, but the differences with the other two countries (Italy and Norway) are modest and not significant. The analysis of the association between social background and type of qualification, on the one hand, and the relationship between type of qualification and

M. Triventi / Research in Social Stratification and Mobility 32 (2013) 45–63

occupational outcomes, on the other, showed some interesting findings. On the one hand, parental education is related to some extent to all three dimensions of institutional stratification: students with better educated parents are more likely to graduate from a top institution and in a longer program and less likely to graduate in humanities and social science, which usually lead to lower occupational returns. On the other, we found that field of study and years of higher education are more closely associated with occupational returns than institutional quality. This is consistent with the fact that the degree of differentiation among institutions in European higher education systems is not as strong as in the United States or Japan. Indeed, it is only in recent years that universities have received more autonomy from the central governments and university rankings have been published at the national level (OECD, 2008). Norway is an exception here because it has a diverse higher education system and graduating from a top university is related to a greater probability of entering a top status occupation. Moreover, in all the countries, graduation from the best educational alternatives is more closely related to occupational status than wage, suggesting that being in the highest quartile of remuneration is only partially determined by educational attainment. This finding could be related to job horizontal mismatches: according to the human capital theory, working outside the appropriate occupational domain will lead to productivity losses, thereby reducing wages.17 It does, however, not per se result in a lower occupational status, in contrast to vertical job mismatches. Nonetheless, additional models (not reported) which control for horizontal mismatch indicate that this argument only partially explains the differences in the net association between type of qualification and the two occupational outcomes. Therefore, it may also be that wages, compared to occupational status, are more related to (unobserved) cognitive and non-cognitive skills which are not fully certified by education credentials (Bills, 2003). The results of the decomposition analysis corroborate our third hypothesis, since we found that institutional stratification played a significant role in mediating the association between social origin and occupational outcomes, especially if we look at the gap between graduates with the most and least educated parents. This suggests that those from better educated families take the best educational options in higher education in order to maintain their advantages in the labor market, as

predicted by the effectively maintained inequality thesis. Yet the importance of institutional stratification in the reproduction of social inequality varies across countries, showing a pattern which is consistent with our fourth hypothesis. Indeed, obtaining different types of qualification accounts for most of the social origin effect in Norway, where a high percentage of graduates and a stratified higher education system are combined with a high degree of institutionalization of graduates’ transition to the labor market. In particular, in Norway parental education has a remarkable effect on years of higher education and institutional quality, which, in turn, are associated with labor market returns, and more especially with socioeconomic status attained. The contribution of institutional differentiation is lower in Italy where the share of graduates in recent cohorts is modest, the higher education system is undifferentiated and there is a widespread reliance on social networks in job finding. Further research should analyze whether these findings are corroborated using other datasets, social class of origin as additional independent variable and long-term occupational returns. Furthermore, possible extensions of this work involve the inclusion of a larger number of countries and the direct measurement of the relationships between each macro-variable and the effect of social background on occupational outcomes. Funding This work was supported by a grant from the University of Milano-Bicocca, co-financed by Regione Lombardia through the “FSE-Dote Ricercatori” programme [Project: 25-AR]. Statements and opinions expressed in the article are those of the author and cannot be attributed to the financing institutions. Acknowledgments The author thanks the participants of the SIMLife Conference in Mannheim (7–9 July, 2011), the In3seminar at the University of Amsterdam (16 April, 2012) and the ISA-Rc28 Conference in Hong Kong (10–13 May, 2012) and, in particular, Thijs Bol, Valentina Di Stasio, Michael Gebel, Kristina John, Irena Kogan, Hyunjoon Park, Herman van de Werfhorst, Felix Weiss and the anonymous referees for useful comments and suggestions. Usual disclaimers apply. Appendix A.

17

We thank one of the referees for suggesting this explanation.

59

See Tables A1–A5.

60

M. Triventi / Research in Social Stratification and Mobility 32 (2013) 45–63

Table A1 Binomial logistic regression models predicting entrance into a top status occupation according to parental education: average partial effects and standard errors in parentheses. DE

One secondary or less Both secondary One tertiary Both tertiary N

ES

IT

NO

APE

SE

APE

SE

APE

SE

APE

SE

0.000 −0.073 −0.012 0.051 1043

(.) (0.052) (0.052) (0.054)

0.000 0.070** 0.094*** 0.186*** 2635

(.) (0.033) (0.024) (0.030)

0.000 0.022 0.058* 0.167*** 1703

(.) (0.026) (0.033) (0.042)

0.000 0.038 0.054** 0.127*** 1383

(.) (0.030) (0.024) (0.026)

Note: Control variables: gender, age, at least one parent born abroad, continued the same job started before graduation. * p < 0.10. ** p < 0.05. *** p < 0.01. Table A2 Binomial logistic regression models predicting entrance into a top status occupation according to parental education and type of qualification: average partial effects and standard errors in parentheses. DE

One secondary or less Both secondary One tertiary Both tertiary HE years Low institution Top institution Social sciences Humanities Economics Law Natural sciences Technical Health N

ES

IT

NO

APE

SE

APE

SE

APE

SE

APE

SE

0.000 −0.006 −0.005 −0.025 0.019 0.000 −0.030 0.000 −0.165*** −0.114** 0.522*** 0.004 −0.141*** 0.728*** 1043

(.) (0.040) (0.040) (0.039) (0.016) (.) (0.021) (.) (0.045) (0.047) (0.106) (0.057) (0.046) (0.064)

0.000 0.010 0.023 0.065*** 0.139*** 0.000 0.022 0.000 0.051* −0.088*** −0.051* 0.239*** 0.303*** 0.465*** 2635

(.) (0.028) (0.019) (0.024) (0.009) (.) (0.018) (.) (0.026) (0.021) (0.027) (0.028) (0.035) (0.029)

0.000 −0.005 −0.007 0.070** 0.114*** 0.000 0.028 0.000 −0.012 0.043 0.198*** 0.453*** 0.112*** 0.467*** 1703

(.) (0.023) (0.028) (0.034) (0.021) (.) (0.024) (.) (0.028) (0.029) (0.044) (0.040) (0.034) (0.044)

0.000 0.015 −0.010 −0.002 0.074*** 0.000 0.101*** 0.000 −0.059*** 0.210*** 0.537*** 0.176*** 0.042 0.062*** 1383

(.) (0.027) (0.020) (0.021) (0.007) (.) (0.026) (.) (0.021) (0.039) (0.198) (0.041) (0.036) (0.023)

Note: Control variables: see Table A1. * p < 0.10. ** p < 0.05. *** p < 0.01. Table A3 Binomial logistic regression models predicting entrance into a top wage occupation according to parental education: average partial effects and standard errors in parentheses. DE

One secondary or less Both secondary One tertiary Both tertiary N

ES

IT

NO

APE

SE

APE

SE

APE

SE

APE

SE

0.000 −0.063 −0.027 −0.023 958

(.) (0.058) (0.057) (0.058)

0.000 0.075* 0.062* 0.074* 2479

(.) (0.036) (0.025) (0.030)

0.000 0.011 0.045 0.113* 1328

(.) (0.029) (0.037) (0.045)

0.000 0.061 0.052 0.092** 1307

(.) (0.037) (0.029) (0.030)

Note: control variables: see Table A1. * p < 0.10. ** p < 0.05. *** p < 0.01.

M. Triventi / Research in Social Stratification and Mobility 32 (2013) 45–63

61

Table A4 Binomial logistic regression models predicting entrance into a top wage occupation according to parental education and type of qualification: average partial effects and standard errors in parentheses. DE

One secondary or less Both secondary One tertiary Both tertiary HE years Low institution Top institution Social sciences Humanities Economics Law Natural sciences Technical Health N

ES

IT

NO

APE

SE

APE

SE

APE

SE

APE

SE

0.000 −0.058 −0.029 −0.013 0.059* 0.000 0.038 0.000 0.069 0.234*** −0.070 0.125* 0.203*** −0.023 958

(.) (0.057) (0.057) (0.058) (0.024) (.) (0.035) (.) (0.047) (0.052) (0.047) (0.051) (0.047) (0.049)

0.000 0.050 0.039 0.039 0.054*** 0.000 0.047* 0.000 0.120*** −0.005 0.021 −0.006 0.162*** 0.117** 2479

(.) (0.034) (0.024) (0.028) (0.010) (.) (0.023) (.) (0.031) (0.028) (0.038) (0.028) (0.038) (0.037)

0.000 0.005 0.010 0.067 0.030 0.000 0.054 0.000 0.112* 0.090* −0.070 0.072 0.071 0.135* 1328

(.) (0.029) (0.036) (0.045) (0.023) (.) (0.030) (.) (0.044) (0.039) (0.042) (0.042) (0.041) (0.055)

0.000 0.055 0.024 0.041 0.057*** 0.000 0.049 0.000 0.002 0.416*** 0.050 0.102* 0.239*** 0.103** 1307

(.) (0.037) (0.028) (0.028) (0.009) (.) (0.031) (.) (0.032) (0.051) (0.040) (0.040) (0.049) (0.035)

Note: Control variables: see Table A1. * p < 0.10. ** p < 0.05. *** p < 0.01. Table A5 Average partial effects on the probability of entering a top status occupation and top wage occupation from basic and full models and their difference, estimated with the KHB method. Top ISEI

Top wage IT

NO

ES

IT

NO

Both secondary vs one secondary 0.584*** Basic model (0.212) Full model 0.057 (0.213) 0.526*** Difference (0.196)

0.131 (0.158) 0.021 (0.158) 0.110 (0.153)

0.472 (0.317) 0.250 (0.317) 0.222 (0.292)

0.431** (0.189) 0.288 (0.190) 0.143** (0.065)

0.044 (0.167) 0.076 (0.168) −0.032 (0.077)

0.440* (0.239) 0.338 (0.239) 0.102 (0.144)

One tertiary vs one secondary 0.721*** Basic model (0.145) Full model 0.174 (0.143) Difference 0.546*** (0.196)

0.359* (0.186) 0.104 (0.185) 0.255* (0.154)

0.374 (0.257) −0.105 (0.259) 0.479 (0.293)

0.362*** (0.136) 0.234* (0.138) 0.128** (0.064)

0.244 (0.199) 0.204 (0.201) 0.040 (0.082)

0.352* (0.190) 0.166 (0.191) 0.186 (0.145)

Both tertiary vs one secondary Basic model 1.322*** (0.171) 0.454*** Full model (0.167) 0.867*** Difference (0.202)

0.970*** (0.196) 0.548*** (0.194) 0.422*** (0.161)

1.022*** (0.255) −0.053 (0.254) 1.075*** (0.301)

0.431*** (0.157) 0.237 (0.160) 0.194*** (0.072)

0.583*** (0.222) 0.527** (0.225) 0.056 (0.093)

0.630*** (0.191) 0.252 (0.195) 0.378** (0.151)

N

1703

1383

2479

1328

1327

ES

2635 *

** ***

p < 0.10. p < 0.05. p < 0.01.

62

M. Triventi / Research in Social Stratification and Mobility 32 (2013) 45–63

References Allen, J., & van der Velden (Eds.). (2007). The flexible professional in the knowledge society: General results of the REFLEX project. Research Centre for Education and the Labour Market, Maastricht University. Allison, P. (1999). Comparing logit and probit coefficients across groups. Sociological Methods & Research, 28(2), 186–208. Alon, S. (2009). The evolution of class inequality in higher education: Competition exclusion and adaptation. American Sociological Review, 74, 731–755. Argentin, G., & Triventi, M. (2011). Social inequality in higher education and labour market in a period of institutional reforms, Italy, 1992–2007. Higher Education, 61, 309–323. Arnesen, C. A. (2000). Higher education and graduate employment in Norway. European Journal of Education, 35, 221–228. Ballarino, G., Bernardi, F., Requena, M., & Schadee, H. (2009). Persistent inequalities? Expansion of education and class inequality in Italy and Spain. European Sociological Review, 25, 123–138. Barone, C. (2011). Some things never change: Gender segregation in higher education across eight nations and three decades. Sociology of Education, 84, 157–176. Barone, C., & Ortiz, L. (2011). Overeducation among European university graduates: A comparative analysis of its incidence and the importance of higher education differentiation. Higher Education, 61, 325–337. Berggren, C. (2008). Horizontal and vertical differentiation within higher education – gender and class perspectives. Higher Education Quarterly, 62, 20–39. Bills, D. B. (2003). Credentials, signals, and screens. Explaining the relationship between schooling and job assignment. Review of Educational Research, 73, 441–469. Blau, P., & Duncan, O. D. (1967). The American occupational structure. NY: Free Press. Boudon, R. (1974). Education, opportunity, and social inequality. Changing prospects in Western society. New York: Wiley. Bourdieu, P. (1979). La distinction: Critique sociale du jugement. Paris: Edition de Minuit. Bourdieu, P. (1996). The state nobility elite schools in the field of power. Cambridge: Polity Press. Breen, R. (Ed.). (2004). Social mobility in Europe. Oxford: Oxford University Press. Breen, R., & Goldthorpe, J. (1997). Explaining educational differentials: Towards a formal rational action theory. Rationality and Society, 9, 275–305. Breen, R., & Jonsson, J. O. (2005). Inequality of opportunity in comparative perspective: Recent research on educational attainment and social mobility. Annual Review of Sociology, 31, 223–243. Breen, R., Karlson, K. B., & Holm, A. (2010). Total, direct, and indirect effects in logit models. Working paper. December 23, 2010. Available online at SSRN: http://papers.ssrn.com/ sol3/papers.cfm?abstractid=1730065 Brint, S., & Karabel, J. (1989). The diverted dream: Community colleges and the promise of educational opportunity in America, 1900–1985. New York: Oxford University Press. Brunello, G., & Cappellari, L. (2008). The labor market effects of Alma Mater: Evidence from Italy. Economics of Education Review, 27, 564–574.

Brunello, G., & Checchi, D. (2007). Does school tracking affect equality of opportunity? New international evidence. Economic Policy, 22, 781–861. Brunello, G., & Schlotter, M. (2011). Non cognitive skills and personality traits: Labour market relevance and their development in education & training systems, IZA Discussion Paper No. 5743. Charles, M., & Bradley, K. (2002). Equal but separate? A cross-national study of sex segregation in higher education. American Sociological Review, 67, 573–599. Duru-Bellat, M., Kieffer, A., & Reimer, D. (2008). Patterns of social inequalities in access to higher education in France and Germany. International Journal of Comparative Sociology, 49, 347–368. Erikson, R., & Jonsson, J. O. (1998). Social origin as an interestbearing asset: Family background and labor-market rewards among employees in Sweden. Acta Sociologica, 41, 19–36. Ganzeboom, H. B. G., De Graaf, P. M., & Treiman, D. J. (1992). A standard international socio-economic index of occupational status. Social Science Research, 21(1), 1–56. Gerber, T. P., & Cheung, S. Y. (2008). Horizontal stratification in postsecondary education: Forms, explanations and implications. Annual Review of Sociology, 34, 299–318. Granovetter, M. (1973). The strength of weak ties. Chicago: The University of Chicago Press. Hansen, M. N. (2001). Education and economic rewards. Variations by social class origin and income measures. European Sociological Review, 17, 209–231. Holmlund, L. (2009). The effect of college quality on earnings evidence from Sweden. Umeå Economic Studies 781, Umeå University: Department of Economics. Hout, M. (1988). More universalism less structural mobility: The American occupational structure in the 1980. American Journal of Sociology, 93, 1358–1400. Karlson, K. B., & Holm, A. (2011). Decomposing primary and secondary effects: A new decomposition method. Research in Social Stratification & Mobility, 29(2), 221–237. Kohler, U., Karlson, K. B., & Holm, A. (2011). Comparing coefficients of nested nonlinear probability models. Stata Journal, 11(3), 420–438. Kouck´y, J., Bartuˇsek, A., & Kovaˇrovic, J. (2009). Who is more equal? Access to tertiary education in Europe. Prague: Education Policy Centre, Faculty of Education, Charles University. Lin, N. (1999). Social networks and status attainment. Annual Review of Sociology, 25, 467–487. Long, J. S. (1997). Regression models for categorical and limited dependent variables. Thousand Oaks, CA: Sage. Lucas, S. R. (2001). Effectively maintained inequality: Education transitions, track mobility and social background effects. American Journal of Sociology, 106, 1642–1690. Mare, R. (1980). Social background and school continuation decisions. Journal of the American Statistical Association, 75, 295–305. Mastekaasa, A. (2011). Social origins and labor market success—stability and change over Norwegian birth cohorts 1950–1969. European Sociological Review, 27(1), 1–15. Mayer, K. U., Müller, W., & Pollak, R. (2007). Germany: Institutional change and inequalities of access in higher education. S. 240–265. In Y. Shavit, R. Arum, & A. Gamoran (Eds.), Stratification in higher education. Stanford: Stanford University Press. Mood, C. (2010). Logistic regression: Why we cannot do what we think we can do, and what we can do about it. European Sociological Review, 26(1), 67–82.

M. Triventi / Research in Social Stratification and Mobility 32 (2013) 45–63 Mora, J.-G., Garcia-Montalvo, J., & Garcia-Aracil, A. (2000). Higher education and graduate employment in Spain. European Journal of Education, 35, 229–237. OECD (2007). Education at a glance. Paris: Organization for Economic Cooperation and Development. OECD (2008). Tertiary education for the knowledge society. Paris: Organization for Economic Cooperation and Development. Paulsen, M. B. (2001). The economics of human capital and investment in higher education. In M. B. Paulsen, & J. C. Smart (Eds.), The finance of higher education: theory, research, policy and practice (pp. 55–94). New York: Agathon Press. Recchi, E. (2007). Italy: Expansion, reform, and social inequality in higher education. In Y. Shavit, R. Arum, & A. Gamoran (Eds.), Stratification in higher education. A comparative study (pp. 400–420). Stanford, CA: Stanford University Press. Reimer, D., Noelke, C., & Kucel, A. (2008). Labor market effects of field of study in comparative perspective an analysis of 22 European countries. International Journal of Comparative Sociology, 49, 233–256. Reimer, D., & Pollak, R. (2010). Educational expansion and its consequences for vertical and horizontal inequalities in access to higher education in West Germany. European Sociological Review, 26(4), 415–430. Reyneri, E. (2005). Sociologia del mercato del lavoro. Il mercato del lavoro tra famiglia e welfare. Il Mulino: Bologna. Shavit, Y., Arum, R., & Gamoran, A. (Eds.). (2007). Stratification in higher education. A comparative study. Stanford, CA: Stanford University Press. Shavit, Y., & Blossfeld, H.-P. (Eds.). (1993). Persistent inequality: Changing educational attainment in thirteen countries. Boulder, CO: Westview Press. Teichler, U. (1988). Changing patterns of the higher education system: The experience of three decades. London: Jessica Kingsley.

63

Teichler, U. (2002). Graduate employment and work in Europe: Diverse situations and common perceptions. Tertiary Education and Management, 8(3), 199–216. Torche, F. (2011). Is a college degree still the great equalizer? Intergenerational mobility across levels of schooling in the United States. American Journal of Sociology, 117(3), 763–807. Treiman, D. J. (1970). Industrialization and social stratification. In E. O. Laumann (Ed.), Social stratification: Research and theory for the 1970s (pp. 207–234). Indianapolis: Bobbs–Merrill Company. Triventi, M. (2011). Stratification in higher education and its relationship with social inequality. Evidence from a recent cohort of European graduates. European Sociological Review, http://dx.doi.org/10.1093/esr/jcr092. First published online: December 20 Triventi, M. (2012). Sistemi universitari comparati. Milano: Bruno Mondadori. Triventi, M., & Trivellato, P. (2009). Participation performance and inequality in Italian higher education in the 20th century. Higher Education, 57, 681–702. Van de Werfhorst, H. G. (2004). Systems of educational specialization and labour market outcomes in Norway Australia and the Netherlands. International Journal of Comparative Sociology, 45, 315–335. Van de Werfhorst, H. G., & Kraaykamp, G. (2001). Four field-related educational resources and their impact on labor consumption and sociopolitical orientation. Sociology of Education, 74, 296–317. Moris Triventi, Ph.D., is a Post-doc research fellow in the Department of Sociology and social research at the University of Milano-Bicocca. His research interests include education, social stratification, and labor market.