An Analysis of Public Spending on Education

The substantive part of this thesis comprises three empirical chapters on national-level (total) public spending on education. It encompasses three related topics in Comparative Public (Education) Finance with Wagner’s law (after German Economist, Adolph Wagner, 1835-1917) of public sector growth providing the theoretical ‘theme’ linking all three empirical topics. All three empirical inquiries make use of the same (unbalanced) panel dataset, predominantly sourced from the World Bank, to operationalise a country-comparative study from 1989 to 2015. The national effort (spending as a share of GDP) and budget share (spending as a share of total government spending) are the two outcome measures of choice. For the most part, the size of government (total government spending as a share of GDP) is the key explanatory variable of interest, with a host of other controls ranging from economic, political, demographic, globalisation and social factors. The first empirical chapter looks at heterogeneity, and conjectures the existence of differences in the mean levels of education spending for economically and politically distinct groups of countries. The primary method applied is a generalised-form t-test using a regression framework. The key finding is summarised in the form of three inequality propositions about education spending and the size of government for richer versus poorer countries. These propositions could also be applied to other areas of public financing. The second empirical chapter can be described as one of comparative dynamics, and makes use of various dynamic specifications in the form of an error-correction model framework. It conjectures the existence of cyclical (short-run) effects being particularly well-evidenced for richer countries, which is shown to be the case. One explanation for this finding is that richer countries have a greater variety of fiscal components with which education ‘competes’ during periods of cyclical downturn (e.g., welfare). The third empirical chapter can be described as one of comparative statics, and applies a panel time-series method. It conjectures the level (long-run) relationship between the national effort (budget share) measure and size of government is positive (negative). The sign of the relationship in each case is broadly shown to hold true. For the national effort measure, common effects are particularly well-evidenced for richer countries, possibly suggestive of it being a benchmark comparative measure for these types of countries. For the budget share measure, a significant negative relationship for poorer countries, but insignificant relationship for richer countries, is possibly explained by the latter having larger, more mature and stable public sectors, for which the budget share measure is already very low and/or relatively stable around some mean level, to a greater or lesser extent, meaning no long-run relationship is necessarily expected in richer countries