Degrees of ‘eliteness’ in higher education systems: A comparison between Germany and France

3.1 Higher education institutions

France and Germany differ strongly in their degree of vertical and horizontal differentiation (Table 1). The most salient difference in vertical stratification is the existence of an elite sector in France (Brauns, Steinmann, Kieffer, & Marry, 1999; Duru-Bellat, Kieffer, & Reimer, 2008; Powell, Graf, Bernhard, Coutrot, & Kieffer, 2012; Windolf, 1992). Elite higher education institutions, called grandes écoles, are often public and prepare students for careers as higher-level civil servants, professors and researchers, engineers, and company managers. To become enrolled, upper-secondary-school graduates usually take part in a selective two-year preparatory course (classes préparatoires aux grandes écoles, or CPGE) that are part of France’s higher education system but are carried out within the secondary school system (Buisson-Fenet & Draelants, 2013). After having completed the intense and demanding CPGE, students take part in grandes écoles examinations. These examinations are competitions (concours) that allow only the most ‘brilliant’ students to enrol. Several of the political studies institutes (e.g., instituts d’études politiques, or IEP, generally known as Sciences Po) do allow students who obtained a ‘très bien’ on their upper-secondary diploma, an excellent final grade that is rarely awarded, to request admission based on qualification and without having to take the written examination. Among the approximately 200 grandes écoles, some of them, such as the École polytechnique, École normale supérieure, and École nationale d’administration, have earned reputations for being particularly selective, with the latter being the alma mater of many French prime ministers. Therefore, stratification also exists among the grandes écoles (Bourdieu, 1996). Achievement notwithstanding, enrolment in the most elitist and prestigious institutions is strongly dependent on social origin. In comparison to students with an upper-tier service-class background, students with a working-class background are 23 times less likely to gain access to these schools (Euriat & Thélot, 1995). However, social inequality among the ‘elite’ educational level has decreased slightly across cohorts (Falcon & Bataille, 2018).

Educational reform in both countries has led to a definite expansion in second-tier higher education and to the creation of new types of higher education institutions (HEIs) (Givord & Goux, 2007; Schindler & Reimer, 2011). In Germany, this differentiation is rather horizontal and can be described as binary segmentation (Kreckel, 2011; Teichler, 2008). It divides higher education into the traditional university sector and a more vocationally oriented sector for universities of applied sciences (Fachhochschule). Additionally, there are vocational academies (Berufsakademien) that combine vocational training and academic higher education to provide a university degree as well as a vocational degree at once. Differentiation in higher education is accompanied by a differentiation in types of upper-secondary schools and diplomas that control access to the different types of HEIs. For example, a subject-specific upper-secondary diploma (Fachhochschulreife) can be acquired in vocational upper-secondary schools (e.g., Fachoberschulen) that normally do not allow students to enter traditional universities. The Abitur (general upper-secondary diploma) obtained in traditional upper-secondary schools (Gymnasium) provides access to all types of HEIs. In contrast to Germany, French upper-secondary schools (lycées) are strongly stratified according to their individual reputation and the achievements of their students. While each type of upper-secondary school is connected to a certain type of diploma in Germany, three types of upper-secondary education and corresponding diplomas are integrated within the French lycée (Farges, Tenret, Brinbaum, Guégnard, & Murdoch, 2016): a general (bac général), a technical (bac technologique) and a vocational track (bac professionnel). Each type is differentiated into subject-specific series or specialisations (e.g., literature specialisation in the bac général, management and business sciences and technologies (sciences et technologies du management et de la gestion, or STMG) in the bac technologique). French higher education is threefold with a selective elite sector, a technical and vocational sector, and a non-selective university sector. Those who hold a bac techno or a bac pro generally pursue quite different trajectories in higher education compared to holders of a bac général. They mostly continue their education within the technical and vocational track. A portion of technical and vocational higher education is offered in special branches of lycées (section technicien supérieur, or STS) that lead to a technician certificate (brevet de technicien supérieur, or BTS). Besides the STS, school leavers can also choose to access the more selective institutes of technology (instituts universitaires de technologie, or IUTs) that are affiliated with universities (Deer, 2005). There are also specialised schools in French higher education (e.g., medical, engineering, trade or design schools). Universities provide non-selective ‘mass’ higher education in France and grant each upper-secondary graduate a basic path to pursue higher education (Goastellec, 2010). Similar to the German Abitur, the bac général offers the greatest number of possibilities to access the manifold types of higher education. Most students in a CPGE have previously obtained a bac général (MENSR-DEPP, 2016). As for bac techno and bac pro holders, their upper-secondary education does not normally provide adequate preparation for both universities and CPGE.

In both countries, the creation of second-tier HEIs has a geographical dimension, although configurations are to some extent different between them. France has built a large number of its newly founded IUTs and STSs in medium-sized or small cities. The same is true for the German universities of applied sciences. A particularity in France is the establishment of local campuses (antennes universitaires) that are branch universities of larger universities and serve the needs of students who live in rural areas. Generally speaking, local campuses have less favourable reputations and their students obtain lower grades in their upper-secondary diplomas (Deer, 2005; Filâtre, 2003; Levy & Jegou, 2013).

3.2 Fields of study

Horizontal differentiation is associated with the diversification of fields of study depending on the type of HEI in both countries. In France and Germany, the traditional academically oriented fields of study are offered in universities, that is to say, in their undergraduate and graduate degree programmes. In Germany, there is specialisation within the sector of traditional universities with technical universities providing degrees in engineering. Similarly, these degrees can be obtained in special engineering schools in France. Often, these engineering schools are grandes écoles and therefore deeply selective. German Fachhochschulen and French STSs and IUTs provide education for professions within technical occupations, business and management, social work or the service sector. In France, these are generally short-cycle study programmes. In Germany, some fields of study—for example, engineering or business studies—exist in both Fachhochschulen and traditional universities. In Fachhochschulen, however, these subjects are less research-oriented and do not necessarily allow access to PhD programmes.

In France, selection in public higher education is strongly dependent on the type of HEI, and French universities were not at all selective until 2018. In Germany, by contrast, universities can be selective according to fields of study. Traditionally, access to certain fields of study in universities has been nationally or locally organised through numerus clausus. Numerus clausus involves selection based on achievement by vertically ranking eligible candidates according to the average grade of their upper-secondary diploma. Entrance examinations, normal in many types of French HEIs, are not common in Germany and can mostly be found in specialised schools (e.g., schools for the fine arts). Besides selection, verticalisation of fields of study is created through national rankings. Since 1998, the German Centre for Higher Education (CHE) institute has ranked fields of study in all kinds of HEIs according to different indicators such as research and teaching reputation. Comparable rankings in France make use of publicly available data on student achievement and student success rates in labour-market integration. Some French rankings, such as the one issued by the consulting and rating agency SMBG, focus on a certain field of study such as business studies and therefore apply largely to business schools.

4.1 Measurements

The Gini index is commonly used for continuous variables (e.g., income) and is based on the Lorenz curve. When it is used with income, the cumulative proportion of the population is plotted against the cumulative proportion of the income in this population. Gini is calculated by dividing the area (A) that lies between the line of equality (the line at 45°) and the overall area under the line of equality and the Lorenz curve (A + B). The index is limited to values between 0 and 1, whereby 0 indicates a perfectly equal distribution of income in the population and 1 indicates that a single person possesses all the income volume of the population (perfect inequality). Unlike income, final grades do not have a volume: it is impossible to think of an individual person who accumulates all of the ‘very good’ final grades in a population. However, high achievers and their human capital can be accumulated by HEIs. The same is true for students’ cultural and economic capital. However, taking the absolute number of ‘elite’ students in the ith HEI and comparing their distribution between HEIs would replicate differences in the size of organisations. The same problem would occur by using the share of elite students within the ith HEI relative to the total number of all elite students in the student population. Instead, we suggest using the share of elite students relative to the total number of students in the ith HEI. A value of G = 0 indicates that all HEIs have the same share of elite students; G = 1 expresses that a single HEI holds all the elite students.

The dissimilarity index D is also based on the Lorenz curve and operates with only two groups (elite and non-elite students, denoted by E and N) by comparing them between units (universities, denoted by U). For example, N_u would indicate the number of non-elite students in unit u. The units (u) are ranked by increasing ratios of (N_u/E_u). Then, the cumulative values of the shares (H_u/H) on the x axis are plotted against the cumulative values (E_u/E) on the y axis. The curve obtained in this way is called the segregation curve. The dissimilarity index D is calculated by the equation

The value of D also ranges from 0 to 1 and can be interpreted as the fraction of (non-)elite students that would have to change their HEI in order to produce an even distribution of elite and non-elite students in all HEIs.

Power-law distributions are mostly used in economics to describe variables that are heavily skewed (e.g., wealth) and indicative of strong inequality. A popular distribution among the family of power-law functions is the Pareto distribution. Pareto observed that if the log income of persons with incomes above a certain threshold is plotted against the log cumulative proportion of the population with this income, the resulting graph is a straight line. The cumulative distribution function of the Pareto distribution is described as

where x_m is a scale that indicates the minimum value of x and α is a positive shape parameter. Variables follow a Pareto distribution if the coefficient α (Pareto index) is approximately 1.16. This would result in a distribution where 80 per cent of a population holds 20 per cent of the total income within the population. In economics, an ‘inverse coefficient’ β of the Pareto index α is widely used to describe the amount of inequality in distributions: β = α/α − 1. The higher the value of β, the greater the concentration of income. Empirically, the inverted Pareto coefficient typically ranges from close to 1.5 (low inequality) to 3.5 (very high inequality). An index of 1.5 indicates that the average income above a certain threshold is 1.5 times higher than the value of that same threshold (e.g., persons who own more than 1 million euros own 1.5 million euros on average) (Piketty, 2014).

4.2 Study design

Analyses that make use of segregation and inequality indices often rely on census data, administrative data or large-scale survey data that include all or a large number of entities in a population (e.g., schools or HEIs) as well as information about the social status of persons within those entities. For our empirical analysis, we used representative secondary cross-sectional survey data of students in Germany and France. Specifically, we used data from the national survey on the living conditions of students in France (Conditions de vie des étudiants en France, or CdV) by the Observatoire de la vie étudiante (OVE) and analysed their scientific use files (SUFs) from 1994 to 2006. Beginning in 2010, CdV altered its sampling methodology and discontinued its previous ID of HEI. This made it impossible to match this dataset to others via the HEI. We therefore only used the SUFs from 1994 to 2006. For Germany, we used the survey of new students (Studienanfängerbefragung, or SAB), which was conducted by the German Centre for Higher Education Research and Science Studies (Deutsches Zentrum für Hochschul- und Wissenschaftsforschung, or DZHW) from 1983 until it was discontinued in 2011. The French and German databases similarly collect information on students’ eventual choice of HEI and their socio-economic characteristics (i.e., previous school trajectories, school achievements, type of HEI and field of study, distance from home town, parental educational and occupational background).

In our initial step, we analysed distributions of students between HEIs. For this investigation, one has to bear in mind that the SAB and CdV sampling differs: In the SAB, the sample of first-year students was drawn from a sample of HEIs. In the CdV, a sample of students was drawn from every university, STS and CPGE. Since the sample of HEIs in the SAB changed in each of the 20 SUFs, there is a high possibility that almost every university and university of applied sciences was included at least once in the SAB. Since only first-year students were surveyed in the SAB, we only used data on students in their first semester in the CdV. In the French data, the number of students in private HEIs or in the CPGE of grandes écoles is very low. Therefore, we aggregated these HEIs into two categories (private HEI and CPGE). We excluded persons who did not state the precise name of their HEI. By pooling eighteen SUFs from SABs and five SUFs from CdV, we obtained 90 HEIs in the CdV and 78 HEIs in the SAB.

In our second step, we compared distributions of students between fields of study. In the German SAB data, information on a student’s chosen field of study is available as a grouped variable with eight main categories (humanities; economics and social sciences; mathematics and science; medicine; agricultural sciences, forestry, and nutrition; engineering; fine arts; and law) and as a more detailed variable with 52 single fields of study. We excluded one field of study (‘general mathematics and science’), which was studied by only five persons in our sample. In the French CdV, there are only grouped data for a given field of study: literature; languages; humanities and social sciences; law and political sciences; economics; administration; science; engineering; life sciences; sports; and health sciences. The variable has seven additional categories according to the course type in the STS (secondary/tertiary sector), IUT (secondary/tertiary sector) and CPGE (literature/commerce/science). These categories are not very indicative of the actual field of study, which is a limitation of our study. We therefore regrouped them into three categories (STS, IUT and CPGE) to capture the general effect of study programmes (especially selective and elite study programmes) that are related to the types of HEIs (see Tables A1 and A2 in the appendix for sensitivity analyses that were calculated with the original categorisation as well as without the fields of study in STS, IUT and CPGE. The results are robust to these configurations). In the French data, the final variable for the field of study has fourteen categories.

We used three dimensions for our indices: parental occupation, parental educational degree and a student’s final grade in upper-secondary education. In accordance with the French grading system in upper-secondary school, the French students were asked for the mention (distinction) in their baccalauréat (1 ‘very good’ to 4 ‘sufficient’) but not for the number of points out of twenty they received (in France, graduates receive a mention when their final grade is ≥12 points out of 20; très bien ≥16 points; bien ≥14 points; assez bien ≥12 points; no mention or sufficient = 10–11.9 points). The German students were required to indicate their final grade in their Abitur, which is determined to one decimal point (1.0 ‘very good’ to 4.0 ‘sufficient’). In both the French and German datasets, students were asked for their parents’ final degree (primary, lower-secondary, intermediate, upper-secondary, short-cycle or professional tertiary, and tertiary education). In the German data, parents’ occupational position was measured as self-employed persons (proprietors of small, medium or large businesses), freelancers, non-manual labourers (higher or lower tier), lower-tier professionals, civil servants (lower, intermediate, higher-tier service) and manual workers (skilled, semi-skilled, unskilled). In the French data, occupations were surveyed according to the Professions et catégories socioprofessionnelles (PCS) classification scheme: self-employed persons (as well as self-employed farmers; PCS 1 and 2), managers and higher-tier intellectual professions (cadre; PCS 3), intermediate professions (PCS 4), non-manual occupations (in administration or commerce; PCS 5), and manual workers (PCS 6). Educational achievement and a person’s social origin (i.e., parental occupational status and educational degree) were not surveyed using international standardisations (e.g., the International Standard Classification of Occupations, or ISCO, or the International Standard Classification of Education, or ISCED) in the two datasets but according to country-specific designs. Although this means that they are not perfectly comparable, we aimed for as much comparability as possible in our operationalisation of higher-status groups in France and Germany. For Duncan’s D, we constructed the following dichotomous variables. First, we compared high-achieving students with all of the remaining student body. In the French CdV, we compared the mentions très bien to all other grades. Among first-year students in France, less than 3 per cent obtained très bien as a final grade. In the German SAB, we took the distribution of the final grades and used the top 3 per cent (grade < 1.2). Secondly, we compared social origin by using the upper-tier service class (i.e., large-scale proprietors, higher-tier non-manual laborers, higher-tier civil servants) versus all other occupations in Germany. In the French data, we compared cadre to all other occupations. Occupations were only collected between 1997 and 2006 in the CdV, which restricts our sample to this period when analysing this variable. Lastly, we investigated educational background by comparing students from families with an academic educational background to those with non-academic backgrounds. For the Gini index G and inverse Pareto coefficient β, we took the number of high achievers, academic milieu and upper-tier service class within the ith HEI divided by the total number of students in the ith HEI. The same procedure was applied for the field of study. In our sample, some HEIs can have a share of zero high achievers. We calculated the Gini coefficient using ineqdec0 for Stata by Jenkins (1999), which allows one to calculate the Gini index when values of the variable of interest are negative or zero.

5.1 Inequality in HEIs

Overall, the Gini index of the shares of high-achieving students in HEIs is 45.9 per cent across all cohorts in Germany (Table 2). In comparison, the overall Gini index for France is slightly higher at 46.7 per cent. Compared to the distribution of high-achieving students, the Gini index is much lower in both countries when we observe students who originate from a family with an academic or higher-tier service class background. However, the Gini index has a higher value in France. These differences in the Gini index with regard to social origin reflect social selectivity in elite higher education in France that is not paralleled in Germany.

If we turn our attention to Duncan’s D, we can see that segregation across HEIs is highest for students’ achievement. It is much lower when we consider parental educational background or occupational class. Segregation between students who received a grade better than 1.2 compared to all other students is moderate at 35.1 per cent. In France, segregation of HEIs according to students’ achievements is a little lower than in Germany because France’s mass institutions (universities) are not selective compared to their German counterparts: between students who received the best mention (très bien) and all other students, the dissimilarity in HEIs is 29.7 per cent. With regard to social origin, the overall institutional segregation according to parental educational background is similar between both countries. As far as students’ social class is concerned, first-year students whose parents are in higher-tier occupational classes are more segregated from other occupational classes within French HEIs than in German HEIs.

As for the shares of high achievers, the somewhat similar inequality (Gini index) between German and French HEIs in France—despite its highly selective elite sector—is not in line with our expectations. On closer inspection, this is not surprising because the Gini index and Duncan’s D are based on the Lorenz curve and are potentially less sensitive to concentrations in the top segment. We will look at this issue in the next section.

5.2 Concentrations in HEIs

Differences between the two countries do become visible when looking at the percentiles of the distributions. The top 1 per cent of French HEIs have a share of high-achieving students greater than 12.1 per cent among their student body, whereas the top 1 per cent of German HEIs only have a comparable share of 7.9 per cent. This suggests that the degree of concentration of high-achieving students is much higher in the distribution of French HEIs than in the distribution of German HEIs. This is also indicated by the results of the percentile ratios: if we compare the upper and lower ends of the distribution, the share of high-achieving students in the bottom 10 per cent of HEIs relative to the top 1 per cent is around twenty-four times greater in France and twenty times greater in Germany. In comparing the bottom 10 per cent and top 5 per cent of HEIs, the ratio is twenty-one for France and only sixteen for Germany. As for differences between the upper end and the middle of the distribution, the ratios between the top 1, 5, and 10 per cent versus the median (50 per cent) are relatively similar between both countries. Since the Gini coefficient is more sensitive to changes in the middle area of distributions, it is not surprising that similar values were obtained for the French and German Gini index. The results reflect the strict process of selection by achievement for France’s elite HEIs, which has no equivalent in Germany.

The distribution of the share of students with an academic family background is slightly more concentrated in French HEIs, which is indicated by the higher percentile ratios. Also, concentration according to parents’ social class is greater in France: in the top 1 per cent of French HEIs, the social origin of 80 per cent of students is cadre; in Germany, only 40.9 per cent of the top 1 per cent have a higher-tier service-class background. It is, however, important to note that the share of higher-tier social classes and academic milieu is much higher in Germany when analysing the bottom percentiles. Among the bottom 10 per cent of HEIs, the share of students from an academic milieu is 15.4 per cent in France and almost twice as high (28 per cent) in Germany. Even if we consider only the bottom 50 per cent of HEIs, the difference in the share of students with an academic family background remains relatively high. This suggests that social origin plays an important role in Germany in the decision to continue one’s educational career in the higher-education system instead of the vocational education and training system. Therefore, the number of students with an academic family background is higher than in France, where only the elite sector is highly selective according to social origin.

5.3 Inequality in fields of study

Now let us look at the distribution of high-achieving students according to field of study (Table 3). For grouped fields of study, the Gini index is 39.4 per cent in Germany. For all fields of study, the Gini index is only slightly higher at 43.3 per cent. In the French data, only grouped fields of study are available; the Gini index is 49.5 per cent. This suggests that shares of students who graduated from upper-secondary education with a very good average grade are less evenly distributed across fields of study in France than in Germany. Results from cross tables (not displayed) show that high-achieving students in France are most frequently enrolled in a study programme of a CPGE (47.1 per cent), whereas their high-achieving peers in Germany most frequently choose medicine (9.2 per cent) among all the available fields of study.

In terms of social origin, the shares of students with an academic and upper-tier service-class background are more evenly distributed across fields of study in Germany. This is underlined by the relatively low Gini indices. This finding is robust regardless of whether the grouped variable or all fields of study are used in the estimations.

In France, 40.7 per cent of students who were graded très bien in their baccalauréat would need to change their field of study in order to obtain an even distribution of students in each field of study. In Germany, it is the other way around. This is because segregation among German higher-education institutions is more pronounced than it is for field of study. If we compare HEI and field of study in France, segregation is higher for field of study.

5.4 Concentrations in fields of study

In the top 5 per cent of fields of study, the share of students with a very good upper-secondary diploma is around 7 to 8 per cent in Germany. In France, that figure is almost twice as high (14 per cent). In Germany, the level of concentration of achievement across fields of study is quite similar to the level across HEIs. In France, by contrast, selection into a field of study by achievement is considerably greater than it is for the HEIs. This finding is related to the pattern of institutional differentiation among fields of study in France. Traditionally, they are more vertically differentiated than in Germany. The stratified domains in the baccalauréat in the French upper-secondary schools have a strong effect on channelling students into subsequent paths within higher education that relate to vertically ordered types of HEIs and their study programmes. For example, students who obtain the bac technologique STMG often enroll in the post-secondary trade and business study programmes within the STS of a lycée.

Concerning social origin, the percentile ratios do not indicate much inequality between the upper and lower deciles of fields of study in Germany. Percentile ratios in France, on the other hand, are three times higher on average. Again, shares of students from an academic milieu in fields of study are substantially higher in Germany than in France. The differences are particularly striking in the bottom percentiles: in the bottom 10 per cent of all fields of study, 38.8 per cent of students in Germany have an academic family background. The same is true for only 15.8 per cent in France. This emphasises that social selectivity in French higher education is smaller in mass higher education and stronger in the elite sector. In comparison, the academic milieu and upper-tier service class in Germany are less exclusively concentrated in certain fields of study because accessing the higher education system in general is socially selective.

5.5 Concentrations in HEIs and fields of study: Pareto coefficient

Lastly, we analysed the inverted Pareto coefficient (Table 4). We took the upper 20 per cent and upper 10 per cent tail of the distribution of HEIs and fields of study to capture concentration in the top segment. The coefficient is always higher in France than in Germany regardless of whether social origin or achievement is used. Concentration in France is highest when high-achieving students are considered: with a coefficient of 2.1, the accumulation of high performers is moderate within the top 20 per cent of HEIs. In terms of social origin, β is even lower. Compared to field of study, the values of β for HEIs are generally higher. We find the same pattern for Germany in both HEIs and fields of study.