2. Empirical and past evidence on the importance of parental education and socio-cultural characteristics in determining educational attainment

Over the 20th century, the massification of education has been a worldwide phenomenon, resulting in the rapid growth of tertiary education [Altbach et al. (Reference Altbach, Reisberg and Rumbley2009)]. Although there exists no scientific consensus on the link between countries’ broad characteristics and the expansion of higher education, Schofer and Meyer (Reference Schofer and Meyer2005) stress the positive role of democratization, human rights, scientization, and development planning. This evolution in educational attainment was made possible by cultural and institutional changes that took place after the Second World War, as expansion in higher education was increasingly seen as a source of progress that benefits both individuals and society rather than a source of inefficiency and anomie [Schofer and Meyer (Reference Schofer and Meyer2005)]. Since then, developed nations have seen, along with the emergence of the welfare-state and social security, a strong decline in the cost of education [Breen et al. (Reference Breen, Luijkx, Müller and Pollak2009)]. As more schools were built and travel conditions improved, living conditions also increased for working classes, resulting in universal access to primary and secondary education [Breen et al. (Reference Breen, Luijkx, Müller and Pollak2009), Barakat and Durham (Reference Barakat, Durham, Lutz, Butz and KC2014)]. Through a domino effect, this improvement in primary and secondary education also increased the postsecondary enrolment [Altbach et al. (Reference Altbach, Reisberg and Rumbley2009)].

Figure 1 shows trends in educational attainment in European countries for cohorts born between 1940 and 1979. As a general trend, we note that the proportion of low-educated population has continuously declined for most countries.Footnote ² The decline has occurred at a stronger pace for females when compared to males, and in countries lagging behind in terms of educational attainment, such as Greece. Overall, a convergence of all countries to a small proportion of low-educated population is clearly observed. Indeed, the arithmetic mean of low-educated population for EU28 countries decreased for females from 58.6% [standard deviation (SD) = 18.8%] for the cohorts of 1940–1944 to 14.5% (SD = 10.1%) for the cohorts of 1975–1979, and for males from 45.4% (SD = 19.2%) to 16.9% (SD = 12.8%). Despite this general decline in low education, significant gaps remain among EU28 countries. For instance, the range in the proportion of low-educated population varies from 3.3% (females born in Sweden) to 61.8% (males born in Portugal) for cohorts born between 1975 and 1979.

Source: Pooled data of ESS (2006 to 2014). See data section for details on variables and categories.

Figure 1. Evolution of educational attainment across cohorts (%) for European-born and immigrants arrived before age 25, by country (red line = arithmetic average).

Conversely, most countries have seen a general increase in the proportion of high education across cohorts. In general, the rate of change was greater for females than for males, so much so that females born between 1975 and 1979 were more likely to get a post-secondary degree than males of the same cohorts [Van Bavel et al. (Reference Van Bavel, Schwartz and Esteve2018)]. The opposite had been true for cohorts born 30 years earlier. Some countries, such as the Czech Republic and Romania, even saw their proportion of high-educated males stagnate at moderate or low levels. Overall, the arithmetic mean for the proportion of the high-educated population increased from 14.8% (SD = 9.3%) to 41.9% (SD = 12.7%) for females, and from 19.9% (SD = 8.1%) to 31.2% (SD = 9.7%) for males. Interestingly, and contrary to what was observed for low education, Figure 1 shows that there is no evidence of convergence between countries in post-secondary educational attainment.

It is known since many decades that the socioeconomic status of the family influences the educational attainment [Sewell and Shah (Reference Sewell and Shah1967), Sewell et al. (Reference Sewell, Haller and Portes1969), Lin (Reference Lin2001)]. Among socioeconomic characteristics, the education of parents proves to be an even better determinant of a child's educational attainment than the occupation [Shavit et al. (Reference Shavit, Yaish, Bar-haim, Stefani, Reinhard, Gunnar and Markus2007)]. Past research has consistently shown a strong correlation between a parent's and his/her children's educational attainment: individuals whose parents have a high level of education have a better chance of getting a high level of education themselves [Bowles and Gintis (Reference Bowles and Gintis1976), Hertz et al. (Reference Hertz, Jayasundera, Piraino, Selcuk, Smith and Verashchagina2008), Kogan et al. (Reference Kogan, Gebel and Noelke2012)]. Evidence shows that this type of intergenerational transfer occurs consistently in all developed nations and has remained stable since the Second World War [Erikson and Goldthorpe (Reference Erikson and Goldthorpe1992), Shavit and Blossfeld (Reference Shavit and Blossfeld1993), Pfeffer (Reference Pfeffer2008)]. Moreover, commenting a journal special issue on ethnic differences in educational attainment, Heath and Brimbaum (Reference Heath and Brinbaum2007) conclude that the socioeconomic status of parents, which is captured in part by their education level, has about the same effect for every ethnic group, since very few interaction variables were found to be significant.

Researchers have identified several mechanisms by which a child's education might be linked to the education of its parents: economic and cultural resources, the influence of other family members, track placement, and incentives to make more ambitious educational choices [Shavit et al. (Reference Shavit, Yaish, Bar-haim, Stefani, Reinhard, Gunnar and Markus2007)]. In short, the parents’ education is an important part of a child's social capital [Bourdieu (Reference Bourdieu1986)]. In addition, the educational attainment may also be linked to inherited abilities which are correlated between family members [Black et al. (Reference Black, Devereux and Salvanes2005)].

Along with parents’ education, other sociocultural variables may have an impact on educational attainment. Many studies in Europe and in the USA have found that some groups such as foreign-born children or racial minorities are at a disadvantage with respect to their educational trajectory [Hirschman (Reference Hirschman2001), Riphahn (Reference Riphahn2003), Heath and Brinbaum (Reference Heath and Brinbaum2007)] or on the contrary performs better than natives following the segmented assimilation hypothesis [Portes and Zhou (Reference Portes and Zhou1993), Alba and Foner (Reference Alba and Foner2016)].

Global expansion in higher education in the USA was shown to have been depressed by compositional effects, the expansion having been slower for Blacks and Hispanics than for Whites [Barakat and Durham (Reference Barakat, Durham, Lutz, Butz and KC2014)]. In Germany, Gang and Zimmermann (Reference Gang and Zimmermann2000) showed that children of immigrants meet a disadvantage in educational attainment that resists statistical controlling of several factors such as parents’ education. Moreover, the educational experience differs following the ethnic origin of children of immigrants, suggesting a persistence of cultural differences in a multicultural society. According to Heath and Brindaum's (Reference Heath and Brinbaum2007) review on ethnic inequalities, this persistent disadvantage affects mainly immigrants from low-developed countries. Among contextual factors explaining these differences, some researchers observed that minority groups are often concentrated in economically deprived neighborhoods, where the poorer quality of schools together with unequal access to resources and other contextual effects are likely to reduce their opportunities [Gronqvist (Reference Gronqvist2006), Heath and Brinbaum (Reference Heath and Brinbaum2007), Pong and Hao (Reference Pong and Hao2007), Zhou (Reference Zhou2009)].

3. Projecting the education

3.3 The CEPAM-Mic educational module

3.3.1 Data and variables

Because CEPAM-Mic aims at implementing sociocultural factors and the education of the parents as determinants of individual's educational attainment, it requires a microdata set that includes all or most of the theoretically relevant determinants of education for all countries, on which statistical models will be built to estimate the needed parameters.

Although the EU-LFS has a large sample covering all EU28 countries, it contains limited information on sociocultural characteristics. Moreover, education of the mother is only available for individuals living in the same household as their mother.Footnote ⁵ Despite its smaller sample size, the ESS was thus preferred to the EU-LFS for the analysis of educational attainment. Five cycles of the ESS (2006–2014) were pooled and reweighed in order to match the base population of the projection model (according to country/age/sex/region of birth/education).Footnote ⁶ Of the 28 EU countries, 13 participated in all five cycles, 13 were missing from at least one cycle, and two were completely missing (Luxemburg and Malta). These two latter countries are thus excluded from the analysis presented in this paper.

From this merged database, people born between 1940 and 1979Footnote ⁷ and immigrants arrived in their host country before the age of 25 were selected. Individuals were then classified according to their country of birth (if born in the EU) or country of residence (if born abroad). A description of the sample size for all countries can be found in Appendix Table A1.

Educational attainment is the dependent variable and is divided into three broad categories based on ISCED classification either:

1. (1) Low: lower secondary or less (ISCED 1 and 2);

2. (2) Medium: upper secondary completed (ISCED 3);

3. (3) High: postsecondary (ISCED 4+).

The independent variables used for the analysis are the following:

• • Education of the mother: categories are the same as for the dependent variable.

• • Country of birth (natives) or country of residence (immigrants): EU28 countries.

• • Region of birthFootnote ⁸: Native, North America or Oceania/Other Europe/North Africa/Latin America/East, South, and South-East Asia/Near and Middle East.

• • Religion: Christian/Muslim/Other religions/No religion.

• • Language spoken at home: Country's official language(s)/Other official languages in the EU28/Other languages. Language has to be official at the national or federal level.

3.3.2 The education module

In the CEPAM-Mic model, educational attainment is modeled in three steps:

Step 1. Determining educational attainment

This first step is at the core of the education module and requires parameters from ordered logit regressions (or cumulative logit with non-proportional odds) at the education level. When an individual is born, a variable indicates the highest level of education that will be reached in his/her lifetime. This is also done for immigrants who arrived before their twentieth birthday and for individuals aged less than 30 in the base population.

The ordered logit regression analysis has two purposes. The first is to estimate the net effect of relevant individual characteristics on educational attainment. The second purpose is to estimate country-specific cohort effects in order to make assumptions on the educational attainment of future cohorts. Because the sample size is insufficient to build stratified country-specific models, countries are grouped into two large regions, EU15/NMS13, corresponding approximately to former historical division of Europe during the second half of the 20th century, which still shapes the immigration patterns in terms of number, origin, and socioeconomic integration [Kahanec and Zaiceva (Reference Kahanec and Zaiceva2009)]. The country-specific effect is captured by an interaction variable between the cohort and the country. The model equation is thus formulated as follows:

(1)$$\ln \left( {\displaystyle{{E_{ij}} \over {1-E_{ij}}}} \right) = \beta _{0j} + \beta _{1j}Ct_i + \beta _{2j}Cr_i + \beta _{3j}\lpar {Ct_i \times Cr_i} \rpar + \beta _{4j}X_i + \beta _{5j}Z_i\; $$

Where

• • E _ij is the probability that an individual i reaches level of education j, where j equals high or medium;

• • Ct is the country;

• • Cr is a discrete variable for cohorts (1940–44 = 1; 1945–49 = 2, …, 1975–1979 = 8);

• • X is a set of sociocultural variables;

• • and Z is the education of the mother.

The ordered logit model provides distinct parameters for high and medium education, low education being the reference. Detailed parameters for all categories and variables are presented in Appendix Table A2. For the sake of simplicity, we focus our analysis on the odds of getting a post-secondary degree (high) compared to the odds of getting a lower degree (low and medium combined).

Note that the attribution of a highest educational attainment only concerns individuals with incomplete education paths: newborn, immigrants arrived before age 20 and members of the base population under 30 years old. For immigrants arrived in adulthood and older members of the base population, the highest degree is the one at the arrival in the host country or at the time of the survey. In the reference scenario, it is assumed to remain the same for the rest of the simulation, although other assumptions may be set in alternative scenarios.

Step 2. Graduation schedule

For those reaching at least the upper secondary level, the age at graduation is determined for all degrees using Eurostat distributions by ISCED levels for the latest graduated cohorts (2013–2014). For those scheduled to complete a post-secondary level, the education module first establishes age at graduation for the post-secondary degree, and then finds a coherent age at graduation for the upper secondary level.

For the three countries with missing data (France for high education; Croatia and UK for medium education), the average distribution of comparable countries was used as an approximation.

Unfortunately, no data exist on education schedules according to sociocultural characteristics or education of mothers and data quality sometimes appears questionable for certain countries. Nevertheless, we assume that variations due to these sources of heterogeneity occur within the age resolution of the model (5 years).

Step 3. Simulation of life course

The last step involves the actual simulation of individual educational events at the age at graduation that was predetermined. At birth, the education level is set to low for everyone. If the individual survives until graduation, the education state variable changes to reflect the appropriate educational attainment. As long as the highest level set at birth is not reached, the individual is tagged as being a student, along with his/her current educational attainment. Since the education variable is used for the modeling of other demographic events, a change in education immediately affects mortality and fertility rates as well as labor force participation.

3.3.3 Limitations

Due to both data and methodological limitations, a large part of the social determinants of educational attainment are discarded in the modeling. For instance, although the literature suggests that father's education is probably more important than the mother's in the prediction of the educational attainment of their children [Gang and Zimmermann (Reference Gang and Zimmermann2000)], CEPAM-Mic is a female-dominant model, that is fertility rates are applied to women, and thus, it is not possible to create a link between the father and the child within the current microsimulation model. It is technically possible to model union formations and dissolutions by pairing individuals to form households and thus access to the characteristics of a potential father, but there are no data covering all EU member countries that would allow consistent statistical estimates of the parameters of these events without generating several major inconsistencies in the projection. Moreover, such addition would necessitate computer power that is actually out of range of most institutions. However, educational homogamy is important as shown by the high correlation between the education of the mother and the education of the father (0.61 in this sample). For these reasons, using education of the mother appears as a good proxy in this context.

Additionally, other sociocultural variables would empirically be relevant, but are not included in the projection model and some heterogeneity remains even when controlling for religion, language, and region of birth. As an example, Muslim or those speaking a non-European language include people from different socio-cultural backgrounds. However, as the sample size is relatively small, it was necessary to create some broad categories, especially for minority groups, to reduce the variance of the estimated parameters. In new member states (NMS13) specifically, the small sample size along with the small number of immigrants do not allow for a distinction of immigrants by region of origin. For the same reason, it is not possible to obtain reliable and coherent parameters from an interaction between the region of birth and religion or with language spoken at home to capture patterns for specific ethno-cultural groups such as Roma.

Similarly, contextual and environmental factors could not be accounted in the modeling of education. Organizational properties of schools (classroom effectiveness, teaching quality, etc.) have a major impact on student achievement [Heck (Reference Heck2009)]. Including this dimension in country-level projection of education would be very hazardous, as it would require to build a standardized indicator for all EU countries that is internationally comparable, and to set assumptions on how this indicator would involve in the future. Summing up, we can nevertheless assume that some of those missing factors are implicitly taken into account in the country-specific parameter of equation (1).

4. Estimation of parameters for the education module

Table 1 shows Max-rescaled R ² and concordance levels for partial and full models. On average, adding mother's education (Z) and sociocultural variables (X) to cohort trends by country (Ct × Cr) increases the concordance by 5–10 points compared to models including cohort trends by country alone. The two performance indicators also show that mother's education is a better predictor of educational attainment than are sociocultural variables: models including Z alone perform better than those including X alone. Moreover, Max-rescaled R ² scores show that mother's education and cohort/country have a similar effect on the explained variance. Performance indicators also show that models perform slightly better for the EU15 region when compared to NMS13, as well as for females compared to males.

Table 1. Performance indicators for partial and full models

In order to assess the effect of the education of the mother and sociocultural variables, we compare their net and gross effect in Figure 2. Gross effects correspond to observed differences (translated into logit), which do not take account the effect of other variables. Net effects are obtained from the full model (country, sex, cohort, sociocultural variables, and mother's education).

Figure 2. Odds of getting high level of education over odds of getting a low or medium level of education.

The importance of mother's education stands out from all other variables as the main determinant of educational attainment. In both regions, the odds of getting a post-secondary degree compared to getting other lower educational levels fall below 0.2 for both males and females with low-educated mothers (reference is high-educated mother), meaning that individuals with a low-educated mother are approximately five times less likely to complete a post-secondary level than individuals with a high-educated mother. Results for individuals whose mother has a medium level of education are similar, although a little less pronounced (odds ratio: approximately 0.3).

Preliminary models also included interaction terms between the education of the mother and the country or cohort, but most of the resulting parameters turned out not to be significant. Although the absence of a significant interaction could be a consequence of a relatively small sample size, this suggests that the effect of mother's education is roughly the same in all countries and didn't change across cohorts (at least since 1940). This result supports many other empirical analyses showing that differentials in intergenerational mobility rates do not vary much over time and across countries [Piketty (Reference Piketty, Atkinson and Bourguignon2000)]. As stated earlier, including mother's education explicitly in the projection model should improve predictive capacity.

Parameters for the region of birth show that cultural background is an important driver of educational attainment, and its effect differs according to sex. Indeed, a strong heterogeneity is observed with respect to the region of birth of immigrants arrived in their host country during childhood, as differences between some immigrant groups are larger than between immigrants and natives. For males in EU15, being born in other European countries (non-member of EU28) significantly reduces the odds of getting a high education level, while the odds increase strongly for those born in African countries (excluding North Africa). Females born in Near and Middle East have a significant disadvantage compared to others. Interestingly, the net effect is even larger than the gross one. By contrast, females born in East, South, and South-East Asia are about twice more likely to get a post-secondary degree than native-born females. Note that due to small sample size and low number of immigrants arrived during childhood in NMS13, this variable could not be included in models for this region.

Another significant result can be observed for the educational attainment of individuals according to their religious affiliation. Compared to being Christian, being Muslim significantly reduces the odds of obtaining a post-secondary degree in both regions and for both sexes and the effect remains significant even when controlling for the other variables. Since the education of the father could be a better explanatory variable than the education of the mother for Muslims [Gang and Zimmermann (Reference Gang and Zimmermann2000)], it is possible that part of the Muslim effect is due to the use of this later variable rather than the former.

With the exception of females in the NMS13 region, a significant and positive effect of religion also remains for the category “Other religions,” which mainly comprises Jews. Having no religion has a small positive effect on education in the gross models, but when controlling for the other variables this effect completely disappears, except for females in NMS13. In general, we can also conclude that the observed differences between religious groups are in part explained by their different composition in terms of mother's education or other variables, as the net effect of religion is always smaller than the gross effect.

Concerning the language spoken at home, the effect of speaking a non-European language on the odds of completing a post-secondary degree is generally reduced after a statistical control, but still remains negative and significant. Social issues underlying these differentials are distinct between EU15 and NMS13. In Eastern Europe, the non-official languages group comprises mainly Romani, whose educational pathways are well documented [Forray (Reference Forray2002)]. In the EU15, this group mostly comprises first and second generations of international immigrants.

Our results have shown that the net effect of the education of the mother on educational attainment is particularly strong, but that other sociocultural variables such as religion, language spoken at home, and in the case of EU15, the region of birth are also playing a significant role. Cohort composition has changed significantly along these dimensions in the course of the 20th century, and so we must aim to disentangle changes that occurred from the evolution of cohort composition and changes that affected the whole population. The second part of the analysis thus concerns the net cohort effect, which is the trend over cohorts once changes in population composition in terms of sociocultural variables and mother's education are factored out.

Figure 3 summarizes the net and gross cohort trends for males and females. For a simplified overview of the analysis, the graphs show the arithmetic average of cohort trend parameters across EU15 and NMS13 countries, and only provide odds for high education compared to the two lower categories.

Figure 3. Comparison of gross and net cohort trends for the odds (logarithm) of getting a high level of education compared to medium or low levels.

When population composition in terms of sociocultural variables and mother's education is taken into account, cohort trends shift down significantly, in one case even changing the direction of the cohort trend from positive to negative. For males in the NMS13 region, gross odds ratios for high education followed a slightly increasing trend (Figure 3, NMS13, solid blue line). However, taking sociocultural variables and mother's education into account, the trend is reversed and becomes slightly negative (Figure 3, NMS13, dashed blue line). This result means that, ceteris paribus, a boy born in the 1970s from a mother with high education has less chance of obtaining a post-secondary level than a similar boy born in the 1940s. As a corollary, this shows that the observed improvement in the gross trends for NMS13 boys is more than completely explained by changes in population composition: there were more educated mothers in the 1970s than in the 1940s and consequently, children born in the 1970s are more likely to get a post-secondary degree. So the observed improvement in educational attainment of men in NMS13 among cohorts born between 1940 and 1979 is an echo of a past net cohort effect affecting previous cohorts of women. Because intergenerational transmission of education is high, a general increase in the level of education in a cohort reverberates in the following generations.

For females in both EU15 and NMS13 and for males in EU15, Figure 3 shows that population composition alone does not fully explain the observed improvement in educational attainment, since net cohort trends (dashed lines) still show improvements across cohorts. Nevertheless, the amplitude is reduced compared to gross trends (solid lines), meaning that a significant part of the improvement across cohorts is explained by sociocultural characteristics and by mother's education.

5. Implementing education of mothers and sociocultural variables in a microsimulation projection model of education

Given the results presented in Section 4, how does population composition in terms of mother's education and sociocultural characteristics affect the outcome of projections? Different forces will work in different directions.

On the one hand, international migration flows are likely to increase the proportion of people speaking a foreign language at home and of Muslims [Coleman (Reference Coleman2006)], which will likely have a negative impact on the average educational attainment. On the other hand, women are more educated than ever before, which is expected to positively affect their children's educational attainment. Moreover, the global increase in educational attainment, net of population composition effects, has been observed to level off or even decline in many countries.

To investigate how these dynamics could affect demographic projections of human capital, we designed five scenarios. First, we built two main scenarios to assess how taking into account sociocultural variables and the education of the mother impact projections of educational attainment:

1. (1) Gross cohort trend in education (GCTE)

In this scenario, educational attainment of future cohorts is extrapolated based on countries and cohort parameters for each sex (without controlling for sociocultural variables and mother's education). Because universal postsecondary attainment is unlikely to happen, the probability of getting a high degree of education is capped at 90% [Barakat and Durham (Reference Barakat, Durham, Lutz, Butz and KC2014)]. This type of scenario can be used in common cohort-components or multistate demographic projections, where future trends are a function of past trends by age and sex only [Lutz et al. (Reference Lutz, Butz and KC2014)].

1. (2) Multivariate determinants of education (MDE)

In this scenario, all parameters from equation (1) are used and cohort trends are extrapolated over the time span of the projection (postsecondary is capped at 90%, as in the first scenario). This second scenario allows isolating the effect of the different components of the model on the future evolution of educational attainment. As explained previously, taking many dimensions into account is best realized in a microsimulation model.

In short, scenario GCTE is closer to the reference scenario of the projection model used in Lutz et al. (Reference Lutz, Butz and KC2014), although without the specific convergence assumptions [Barakat and Durham (Reference Barakat, Durham, Lutz, Butz and KC2014)] and with different hypotheses in terms of immigration. Scenario MDE adds differentials according to sociocultural characteristics and education of the mother, so that the evolution of educational attainment can be decomposed into changes due to net cohort trends and changes due to the evolution of population composition.

In addition to these two main scenarios, we built three scenarios, taking the MDE scenario as a basis, but changing only a specific set of parameters. These scenarios allow analyzing how sensitive the modeling of education is to its different drivers of changes.

1. (3) Equality in education for Muslims (MDE-MuslimEq)

In this scenario, we set to 0 parameter for Muslims. In other words, this scenario assumes that there is no differential between Muslims and Christians in terms of educational attainment. Remember that the negative parameter associated with the mother's Muslim religion only describes a statistical relationship. It does not come from a causal analysis of the dynamics that could explain this observed relationship. The objective of this scenario is to test the sensitivity of the projection to this sociocultural variable, but additionally, it can also serve as an example of the potential impact of policies aimed at equal opportunities in education. Indeed, this statistical disadvantage to Muslim children may result from contextual factors associated with inequalities between neighborhoods and schools [Gronqvist (Reference Gronqvist2006), Pong and Hao (Reference Pong and Hao2007)], as well as unequal access to resources [Zhou (Reference Zhou2009)].

1. (4) Equality in education for children from low- and medium-educated mothers (MDE-EduM)

This scenario is set to 0 parameter for children from low- and medium-educated mothers and test how projection outputs are sensitive to this component of the equation. It thus assumes that these children have the same probability of getting the highest level of education than children from a high-educated mother. It may thus serve as an illustration of how policies improving the access to post-secondary education of children from less educated families may affect future education trends.

1. (5) Twice more Muslims among new immigrants (MDE-MuslimX2)

This scenario doubles the proportion of Muslims among new international immigrants (passing from about 30% to 60%). It tests how outcomes are sensitive to the migration composition in terms of religion.

In this paper, scenarios are built with the purpose of assessing how different models of education would affect projection results in the context of continued current demographic trends. Consequently, all scenarios assume continuation of recent trends for other demographic components of change, such as fertility, mortality, and domestic mobility.

Figure 4 shows the projected proportion of high education in the population aged 25–54 years old. First, concerning scenario GCTE and MDE, because of demographic inertia, the trends for high education are also very similar for the first decades of the projection. This occurs because educational attainment does not change for middle- and old-age adults: adults from the base population are only gradually replaced by new cohorts through a process of demographic metabolism [Ryder (Reference Ryder1965)]. At the end of the projection, however, results from the two scenarios differ by about five points, the proportion of post-secondary education being higher in scenario GCTE (52% vs. 48%). To a certain degree, in a scenario such as the GCTE in which no change in trends is explicitly modeled for major factors that are likely to change population composition, we can assume that gross cohort trends implicitly take population composition into account. The usefulness of microsimulation is that it makes it possible to explicitly model both the effects of a change in behavior and the effects of a change in population composition over time and to present the consequences of each on specific results pertaining to different population subgroups.

Figure 4. Projected proportion of high education, 25–54 years old, 2015–2060, EU26 (Luxemburg and Malta are excluded).

Different assumptions concerning parameters for the education of the mother may result into very different projection outcomes. The scenario MDE-EduM yields a much higher proportion of high-educated population in 2060 (62%). As shown by the scenario MDE-EduM, giving to children from low-educated mothers the same chance to get a post-secondary degree than those from a high-educated mother is likely to double the expected increase in the proportion of high-educated adults by 2060 compared to the MDE scenario. This outcome highlights the importance of the education of the mother as a driver of future educational trends and the potential gains in terms of future educational trends that can generate a policy aimed at increasing access to high education for children from less educated families. Scenarios MDE-MuslimEq and MDE-MuslimX2 yield about the same trend as MDE. The proportion of high educated is only slightly higher for MDE-MuslimEq and slightly lower for MDE-MuslimX2. Removing the parameter for the Muslim population is indeed unlikely to have a large effect on to whole European population because it only concerns a very small proportion of the population. For similar reasons, doubling the proportion of Muslims among immigrants cannot drastically change general educational trends among the whole population.

Integrating additional variables in the microsimulation model also allows for outputs that go beyond age, sex, and education, and that may thus provide valuable insights to European policy makers. Figure 5, for instance, contrasts the evolution of the proportion of Muslims in the total population and in the population with low education (age group 25–54).Footnote ⁹ It also illustrates the analytical possibilities provided by the microsimulation model which can generate outputs with much more dimensions.

Figure 5. Projected proportion of Muslims in the total population and in the population with low education, age group 25–54, 2015–2060, EU26 (Luxemburg and Malta are excluded).

Figure 5 shows that for all scenarios, the proportion of Muslims is higher in the starting population and grows faster among the low-educated population than in the total population. In the population with low education, the growth of the proportion of Muslims increases about 50% faster in scenario MDE (blue line, left graph) when compared to scenario GCTE (red line, left graph). The proportion of Muslims in the population with low education increases from 9% in 2015 to 34% in 2060 in scenario MDE, compared to 26% in scenario GCTE. In scenario GCTE, the proportion of Muslims in the population with low education grows faster than in the total population solely because of assumptions on the intensity and composition of future immigration flows. In scenario MDE, the proportion of Muslims in the population with low education is also driven up by the religion-specific regression coefficient used in the derivation of educational attainment as well as by parameters for characteristics correlated with Muslims that affect negatively educational attainment (mother's education, region of birth, and language). The difference between scenarios GCTE and MDE in this specific output illustrates the importance of taking sociocultural variables into account in order to measure the impact of immigration on future educational attainment or on social cohesion and inequalities. Given that low-educated women, Muslims, and speakers of non-European languages will likely continue to be overrepresented in future cohorts of international immigrants compared to the native population, the outcome from the model variant MDE appears more plausible than the outcome from GCTE.

We saw that scenarios MDE-MuslimEq and MDE-MuslimX2 only slightly affected the EU28 trend in education. However, when looking at the Muslim population specifically, the effect of these alternative scenarios is much more evident. In 2060, the proportion of Muslims among the low educated is about five points of percent lower in the scenario MDE-MuslimEq compared to the scenario MDE. In opposite, the scenario MDE-MuslimX2 strongly increases the proportion of Muslims not only in the total population (brown line, right graph) (about 22% in 2060 vs. 13%–14% for other scenarios), but particularly among the low educated population (brown line, left graph) (about 55% in 2060 vs. 26%–34% for other scenarios).

In addition to testing how the model reacts to changes in parameters, scenarios MDE-MuslimEq and MDE-MuslimX2 are also examples of the potential of microsimulation in the generation of alternative scenarios to help understanding the interaction among different variables and the potential impact of public policies on education trends. While the scenario MDE-MuslimX2 showed that a change of the composition in immigration might lead into more disparities in education among sociocultural groups, the scenario MDE-MuslimEq revealed that a better access to post-secondary education for Muslim children is likely to reduce significantly those disparities. Such results, moreover, highlight important social fragmentation issues that could emerge from increasing immigration flows to Europe and rising inequalities in education without implementing programs facilitating better integration of the second generation for some population groups at risk of experiencing lower upward social mobility.