Determinants of Central Bank independence: a random forest approach

In this paper we implement an effcient non-parametric statistical method, Random survival forests, for the selection of the determinants of Central Bank Independence (CBI) among a large data base of political and economic variables for OECD countries.This statistical technique enables us to overcome omitted variables and overftting problems. It turns out that the economic variables are major determinants compared to the political ones and linear andnonlinear effects of chosen predictors on CBI are found.