Tax determinants revisited. An unbalanced data panel analysis

Abstract

Quantitative research on taxation is important to test which variables affect it and hence to identify strategies in order to boost tax revenue or change its composition. Albeit research on this topic is not recent, many authors have focused on developed countries, whereas others have avoided including variables that are available recently, for example many governance indicators. This paper contributes to the matter by expanding the dataset with a relevant sample of countries, variables, and time observations. Using a dataset that covers over 138 countries and a 40-year period (1976–2015) we estimate static and dynamic models, although the results generally keep unchanged. In addition, and according to the coefficients significance, some variables as the following influence taxation and/or its composition: Agriculture’s share in gross domestic product, financial intermediation, natural rents, education, population share above 65 years, quality of government, and democracy.