We investigate economic and institutional determinants of ICT infrastructure for a broad cross section ofmore than 100 countries. The ICT variable is constructed from a principal components analysis. The explanatory variables are selected by variants of the Lasso estimator from the machine learning literature.In addition to least squares, we also apply robust and semiparametric regression estimators. The results show that the regressions are able to explain ICT infrastructure very well. Major determinants identified are real income per capita, the availability of electricity, the extent of urbanization and indicators for the quality of the institutional environment. We also find evidence of conditional convergence of the ICT infrastructure across countries.
determinants of ICT infrastructure global digital divide variable selection machine learning robust and semiparametric regression