Replication Data Readme file for Chapter 3: The Relationship between Contextual Characteristics and the Intergenerational Correlation of Education in Developing Countries

Chapter 3 looks at the institutional factors that contribute to explaining the relationship between parent’s education and children’s education. Data for 48 countries in total, from multiple harmonized surveys, are utilised. A total of 149 surveys are included. Using multivariate regressions, we first present the correlation coefficients of the relationship between parent’s education and children’s education. These coefficients then serve as the dependent variable in the regression analysis with the institutional factors at the second stage. To this end, secondary data are obtained from the household Demographic and Health Surveys (DHS), and from the U.S. Agency for International Development (USAID) and the World Bank data catalogue. The DHS are nationally representative cross-sectional surveys where data on impact evaluation indicators on the population, health, and nutrition in over 90 countries are represented. The primary respondents of the surveys are women of reproductive age, between 15-49 years, who respond to a household questionnaire and a woman’s questionnaire (DHS Program, 2020). The man’s questionnaire is responded to by men of reproductive age (typically 15 to 49, 54, or 59). In the household questionnaire, the respondent provides information on household membership, individual characteristics, household head, health, housing, consumer goods, and living conditions (DHS Program, 2020). The factors from the USAID and the World Bank data catalogue are part of the world development indicators (WDI) and the worldwide governance indicators (WGI). Corruption estimates, political stability estimates, and voice and accountability estimates are taken from the WGI while the others (GDP, prevalence of HIV, life expectancy at birth, female-male labour force participation, government expenditure on education, pupil-teacher ratio, primary school starting age, primary school duration, secondary school duration, compulsory years of education, fixed telephone subscriptions, and mobile cellular subscriptions) are from the WDI. The WDI is a compilation of high-quality, relevant, and internationally comparable statistics about global development and the fight against poverty (World Bank, 2020b). 1600 time series indicators are contained in the database for 217 countries. These indicators are organized according to six main thematic areas that are poverty and inequality, people, environment, economy, states and markets, global links (World Bank, 2020b). The WGI are nationally comparable indicators of government selection, monitoring, replacement, effectiveness, and the respect of citizens and the state. The worldwide governance indicators generally report on six broad governance dimensions for over 215 countries and territories. These dimensions are government effectiveness, control of corruption, rule of law, voice and accountability, regulatory quality, and political stability and absence of violence (World Bank, 2019). Specifically, we focus on GDP, the prevalence of HIV, life expectancy at birth, female-male labour force participation, government expenditure on education, pupil-teacher ratio, primary school starting age, primary school duration, secondary school duration, compulsory years of education, fixed telephone subscriptions, mobile cellular subscriptions, the extent of corruption, the extent of political stability, and the extent of voice and accountability. The factors used in this chapter are selected based on data availability. The process looks at the correlation between these factors and the intergenerational correlation of education. The results show that these institutional factors account for 39% of the explained cross-country variation in the intergenerational correlation of education. The pupil-teacher ratio, primary school duration, and compulsory years of education reduce intergenerational correlation of education by 0.03 years, 0.03 years, and 0.02 years respectively, following a one standard deviation change in the variables. Besides these variables, GDP, female-male labour force participation, and extent of voice and accountability reduce intergenerational correlation of education by 0.01 years, 0.03 years, and 0.03 years respectively, following a one standard deviation change in the variables. This confirms our second hypothesis on favourable institutional characteristics being able to reduce intergenerational correlation of education.