Enrollments in education have been expanding constantly. Despite huge international differences in the timing and speed of expansion. This trend is clear throughout all world regions, income groups, and all levels of education (Marginson 2016). We know a lot about differences between countries’ trajectories in the expansion of higher education (HE) (for example Schofer and Meyer 2005; van der Wende 2017). Yet, dissecting this global trend unveils large inequalities between countries, regions, and individuals. Many studies show the pervasiveness of inequality in access to HE (Atherton, Dumangane, and Whitty 2016; Breen et al. 2010; Ilie, Rose, and Vignoles 2021; Shavit and Arum 2007). This paper adds to the existing literature in three ways: First, it focuses on access to rather than participation in HE by developing continuation rates, which, arguably, better measure the openness of HE systems. Second, it widens the scope of analysis by introducing an original data set on these continuation rates for a set of up to 183 countries from 1900 to 2020. Third, this paper rigorously tests theories on the expansion of education and quantitatively analyzes drivers for the recent expansion of HE.

Due to HE being outside the compulsory system, the use of enrollment figures in many studies overlooks crucial differences and developments in the inclusiveness of HE. Past studies have often equated participation in HE with access to HE. Although those two are connected, they should be operationalized based on the specific funnel-like structure of educational careers: students need to pass through the mostly compulsory primary education (PE) to secondary education (SE) and then to HE. Taking this structure seriously means that not every person in an age cohort has the same access opportunities. Most often, one can only enter HE when one has successfully graduated from SE. In the present paper I introduce a metric taking this into account by measuring who makes it from SE to HE. Indeed, the comparison with ordinary enrollment rates shows remarkable difference over time, with enrollments steadily increasing and continuation stagnating for the most part. This means that, globally, on average, approximately one-quarter of those who attended SE also gained access to HE. And this remained surprisingly constant from 1900 to 1990. Only in modern times did the development of continuation rates adopt the same logistic form as enrollment rates and continue to rise.

In the next step I set out to explain differences of this measure of continuation rates across time and space. Based on theoretical considerations of a) (micro-levels) Theories on Inequality, b) Sociological Institutionalism and c) Partisan Political Explanations I implement estimations in a regression discontinuity design and mixed-model regressions.

The estimations reveal that much of the expansion of access to HE in OECD countries can be explained by a shattered glass ceiling: increasingly more women continue to HE. In non-OECD countries, however, the glass ceiling is still a reality and girls’ education careers end in secondary school. Furthermore, I can show that right-leaning political leadership drives expansion in non-OECD countries. In OECD countries, however, elite (parties) seem to aim at limiting widespread access as long as possible. Additionally, notions of a threshold of enrollment after which expansion quickly increases in speed are shown to hold for women’s access to HE in the global sample, but not for males and neither overall.

The research employs two main quantitative approaches to investigate the expansion of HE access and the factors that drive these changes: Regression Discontinuity Design (RDD) and Multilevel Regression Analysis. The first method, RDD, is applied to test hypotheses regarding the existence of critical thresholds in educational enrollments that lead to rapid expansion. RDD is a quasi-experimental technique designed to estimate causal effects in situations where random assignment is not possible (D. S. Lee and Lemieux 2010). In this study, the “treatment” is defined as a country-year surpassing a specific enrollment threshold. By comparing observations just above and just below these thresholds, the method mimics the conditions of a randomized controlled trial, under the assumption that these observations are otherwise similar. To avoid bias, only observations close to the threshold—within a bandwidth of 15—are included in the analysis. In this paper I test several thresholds (15%, 30%, 50%, and 70%), which correspond to different stages of educational expansion, such as Elite, Low, Mass, and Universal (Barakat and Shields 2019; Marginson 2016; Trow 1973). Three different datasets are used to ensure robustness: interpolated net enrollment data (J.-W. Lee and Lee 2016), UNESCO gross enrollment data, and a combined series covering the period from 1900 to 2020. Additionally, the rdrobust package is used to estimate local polynomial regressions, which allow for non-linear effects and provide robust, bias-corrected confidence intervals for the treatment effects. The second method, Multilevel Regression Analysis, is used to explain cross-country differences. This approach involves a mixed-model regression with restricted maximum likelihood, nesting observations within countries, years, and governments (Garritzmann and Seng 2020). The dependent variable is the continuation rate, while the independent variables include measures such as income inequality, ethnic fractionalization, gender ratios in higher and secondary education, and international organization memberships as proxies for international pressure. The analysis also incorporates a government coalition left-right index and interaction effects between enrollment rates and political orientation. Controls for economic capacity and productivity are included, as well as enrollment ratios and time variables. The sample covers 104 countries from 1970 to 2018. The analysis is conducted both for the global sample and separately for OECD and non-OECD countries, which reveals more plausible results when accounting for differences in economic development.

This study introduces continuation rates as a more precise measure of access to higher education (HE), focusing on the proportion of eligible students who actually enter HE. Unlike traditional enrollment rates, continuation rates reveal that global access to HE remained static for much of the 20th century, only expanding from the 1990s onward. Regression Discontinuity Design and multi-level regression analyses highlight significant differences between OECD and non-OECD countries. Economic development is important, but its effects vary. No universal threshold for expansion – as might be expected from theories such as MMI/EMI – was found. Notably, the increase in continuation rates is primarily driven by women entering HE, especially in OECD countries, suggesting that the “glass ceiling” has been shattered for women in these contexts. Political ideology also plays a decisive role. In OECD countries with low HE enrollments, right-leaning governments tend to protect elite status and are associated with lower continuation rates. As enrollments rise, this effect diminishes. In non-OECD countries, right-leaning governments may actually foster expansion. Despite global expansion, persistent inequalities remain, particularly regarding socioeconomic status, gender, and geography. Traditional metrics often obscure these disparities. To some extent continuation rates help clarify who benefits from educational expansion. Building on the results, further research should take a closer look at the difference in study programmes and institutions. Access to HE might not, in its abstract form, ameliorate inequalities.

Atherton, Graeme, Constantino Dumangane, and Geoff Whitty. 2016. “Charting Equity in Higher Education: Drawing the Global Access Map.” Pearson. Barakat, Bilal, and Robin Shields. 2019. “Just Another Level? Comparing Quantitative Patterns of Global Expansion of School and Higher Education Attainment.” Demography 56(3): 917–34. doi:10.1007/s13524-019-00775-5. Breen, R., R. Luijkx, W. Muller, and R. Pollak. 2010. “Long-Term Trends in Educational Inequality in Europe: Class Inequalities and Gender Differences.” European Sociological Review 26(1): 31–48. doi:10.1093/esr/jcp001. Garritzmann, Julian L., and Kilian Seng. 2020. “Party Effects on Total and Disaggregated Welfare Spending: A Mixed‐effects Approach.” European Journal of Political Research 59(3): 624–45. doi:10.1111/1475-6765.12371. Ilie, Sonia, Pauline Rose, and Anna Vignoles. 2021. “Understanding Higher Education Access: Inequalities and Early Learning in Low and Lower‐middle‐income Countries.” British Educational Research Journal 47(5): 1237–58. doi:10.1002/berj.3723. Lee, David S, and Thomas Lemieux. 2010. “Regression Discontinuity Designs in Economics.” Journal of Economic Literature 48(2): 281–355. doi:10.1257/jel.48.2.281. Lee, Jong-Wha, and Hanol Lee. 2016. “Human Capital in the Long Run.” Journal of Development Economics 122: 147–69. doi:10.1016/j.jdeveco.2016.05.006. Marginson, Simon. 2016. “High Participation Systems of Higher Education.” The Journal of Higher Education 87(2): 243–71. doi:10.1080/00221546.2016.11777401. Schofer, Evan, and John W. Meyer. 2005. “The Worldwide Expansion of Higher Education in the Twentieth Century.” American Sociological Review 70(6): 898–920. doi:10.1177/000312240507000602. Shavit, Yossi, and Richard Arum. 2007. Stratification in Higher Education. A Comparative Study. Stanford, Calif.: Stanford Univ. Press. Trow, Martin. 1973. Problems in the Transition from Elite to Mass Higher Education. Berkley, CA: Carnegie Commission on Higher Education. van der Wende, Marijk. 2017. “Opening up: Higher Education Systems in Global Perspective.” Centre for Global Higher Education working paper series Working paper no. 22.