The almost universal phenomenon of educational expansion since the emergence of state-provided mass schooling has been studied from educational (Boli et al., 1985), economic (Fuller & Rubinson, 1992), sociological (Archer, 1982; Jónasson, 2003) and system-theoretic perspectives (Green, 1980). However, comparative insights into the dynamics of school expansion have been hampered by the scarcity of consistent historical data for a large number of countries, remarkable efforts such as those by Meyer (1992) and Barro and Lee (2001) notwithstanding. This study investigates the growth pattern of school attainment at three levels (primary, secondary, tertiary) in 120 countries during the second half of the 20th century, based on a new dataset (Lutz et al., 2007). Partly due to data constraints, the study focuses exclusively on formal schooling and specifically attainment rather than enrollment. The objectives are twofold. At the descriptive level, it is instructive to compare expansion patterns during this period with analyses of very long-term trends available for a small number of industrialised countries (e.g. Rubinson & Ralph, 1984). In addition, contemporary debates about the feasibility of international education targets (Benavot et al., 2006) should be informed by an understanding of 'baseline momentum' relative to which improvements are sought. A further objective is to analytically investigate patterns that are consistent across countries and over time, i.e. to fit statistical models to the empirical expansion patterns. Both the parameter values and the qualitative shape of the curves provide potential insight. The parameter estimates provide insight into the 'typical' pace of expansion, the relative speed between genders and different school phases, and the identification of particularly slow/fast countries. Different theoretical perspective create diverging expectations about whether population education levels are driven by 'optimal' social and private investments in education, political steering decisions, or whether the social logic of education credentials eventually forces the universalization of attainment at each successive level. Accordingly different shapes of the growth curves are predicted by these competing explanatory frameworks. The support of the empirical results for these competing explanations of educational expansion and the implications for educational policy-making are discussed.

The study employs two distinct methodologies. Firstly, a systematic review of existing comparative time series of educational expansion and theories regarding typical growth patterns. Secondly, the fitting of aggregate and country-specific growth patterns to time series of educational expansion provided by the IIASA dataset is performed using Least Squares and Bayesian spline regression. The data source used here for the first time for this kind of analysis deserves further comment. It comprises of a new dataset (Lutz et al., 2007) of educational attainment (4 levels) by age and sex for 1970-2000 for 120 countries for the population aged 15 years and older (thus being informative of schooling roughly during the period 1960-1970), based on census data around the year 2000. Assuming that schooling is obtained early in life, the education profile of older cohorts is projected backwards, while correcting for mortality differences between education groups.

Despite the wide range of different country experiences with school expansion, and individual cases of extremely rapid or stagnant progress, a surprisingly consistent pattern of the diffusion of schooling can be observed across a large number of countries during the second half of the 20th century. The empirical evidence presented here confirms previous theoretical arguments that educational diffusion follows an s-shaped growth process converging to universal enrollment. This is found to apply to all phases of schooling and both genders. In addition, we find that female enrollment was already growing much more rapidly than that of males prior to the adoption of the MDGs and EFA goals, putting the achievements in this respect since 1990 into perspective.