Reconstructing Educational Attainment from the UN's Age and Sex Distribution (1950–2015)

We have updated the data on the historical reconstruction of population by sex, age, and education, as well as the mean years of schooling, for 147 countries covering the period from 1950 to 2015. Due to some changes in our methodology, the results differ across all countries. Our analysis relies on validated and harmonized historical data used in the 2018 version (Speringer et al., 2019). This back-projection is consistent with the 2023 update of the WIC population and human capital projections (K.C. et al., 2024). The back-projection utilizes a multidimensional cohort component model. The fundamental hypothesis for reconstructing past educational levels is that educational attainment is typically acquired in earlier stages of life and remains stable thereafter. Consequently, the educational composition of the population reflects historical trends in educational attainment by cohorts, adjusted for the effects of mortality and migration differentials based on education. We begin with the initial educational composition from the last available year. Furthermore, we extend the age structure of the population to include individuals aged 170 and above and incorporate historical education data. We back-project educational levels along cohorts for those aged 35 and older, assuming no further changes in educational attainment for this group. For these ages, the distribution of education primarily varies due to mortality differentials related to education. For the population aged 15-34, we back-project educational dynamics by calculating cohort-specific education attainment progression ratios (EAPR) and then reverse the progression ratio as we move backward in time. While projecting backwards, we apply mortality differentials (also used in the projections) to the population by splitting the overall survival ratio into six education-specific survival ratios. Finally, we apply the resulting education distribution by age and sex to the United Nations population estimates (United Nations, 2022) for the period from 1950 to 2020. This version number V01 used here corresponds to the version number V15 of the forward projection (updated soon). References K.C., S., Dhakad, M., Potančoková, M., Adhikari, S., Yildiz, D., Mamolo, M., Sobotka, T., Zeman, K., et al. Updating the Shared Socioeconomic Pathways (SSPs) Global Population and Human Capital Projections. IIASA Working Paper. Laxenburg, Austria: WP-24-003, February 2024. Speringer, M., Goujon, A., K.C., S., Potančoková, M., Reiter, C., Jurasszovich, S., Eder, J., Global Reconstruction of Educational Attainment, 1950 to 2015: Methodology and Assessment. VID Working Paper 02/2019.本次研究更新了1950年至2015年间147个国家按性别、年龄、受教育程度划分的人口历史重构数据，以及平均受教育年限数据。由于研究方法有所调整，所有国家的估算结果均存在差异。本分析依托2018版研究中经过验证与标准化处理的历史数据（Speringer等，2019）。本次回溯估算与2023年更新的WIC人口及人力资本预测模型（K.C.等，2024）保持一致。 本次回溯估算采用多维队列组分模型（multidimensional cohort component model）。重构历史受教育水平的核心假设为：个体受教育程度通常在生命早期阶段完成，此后将保持稳定。因此，人口的受教育结构能够反映各队列受教育程度的历史趋势，并已针对基于受教育程度的死亡率与迁移率差异影响进行调整。 我们以最新可获得年份的初始受教育结构作为起始点。此外，我们将人口年龄结构扩展至包含170岁及以上人群，并纳入历史受教育数据。对于35岁及以上人群，我们沿队列回溯其受教育水平，假设该群体的受教育程度不会再发生变化；此年龄段人群的受教育分布差异主要源于与受教育程度相关的死亡率差异。 针对15-34岁人群，我们通过计算队列专属受教育程度进展比（Education Attainment Progression Ratio，简称EAPR）来回溯其受教育动态，并在时间回溯过程中对该进展比进行反向运算。 在回溯过程中，我们将总存活比拆分为6个基于受教育程度的专属存活比，并将该死亡率差异（亦用于正向预测模型）应用于人口数据。最终，我们将基于年龄与性别的估算受教育分布，匹配至联合国1950年至2020年的人口估算数据（联合国，2022）。 本次使用的V01版本号，对应正向预测模型的V15版本（即将更新）。 参考文献 K.C.、S.、Dhakad M.、Potančoková M.、Adhikari S.、Yildiz D.、Mamolo M.、Sobotka T.、Zeman K.等. 《更新共享社会经济路径（Shared Socioeconomic Pathways，简称SSPs）全球人口与人力资本预测》. IIASA工作论文. 奥地利拉克斯堡：WP-24-003，2024年2月。 Speringer M.、Goujon A.、K.C. S.、Potančoková M.、Reiter C.、Jurasszovich S.、Eder J. 《1950-2015年全球受教育程度重构：方法与评估》. VID工作论文02/2019。