Title: Determinants of economic development: SEM approach
Authors: Vladimir Potashnikov - RANEPA (Russia) [presenting]
Andrey Zubarev - Russian Presidential Academy of National Economy and Public Administration (Russia)
Oleg Lugovoy - RANEPA (Russia)
Abstract: The study is devoted to finding main determinants of economic development. The key idea of the study is to select potential variables not by the views of a particular researcher, but by some automatic procedure. Criteria for variable selection should be unified. We use SEM (structural equation modeling) technique for this purpose. Contrary to many other studies we model development as a latent variable. We use preliminary data transformation procedure and constructed ``correlation baskets'' (groups of variables, which explain the variation of a particular variable) to find clusters of variables that potentially indicate and explain economic development. Our database consists of more than 1400 variables from WGI, WDI, Frazer Institute database, and PWT. We chose GDP per capita (PPP) and different factors that describe healthcare and education system (e.g. childe and neonatal mortality, several types of school enrolment and etc.) as indicators of economic development. Amongst many potential clusters of explanatory variables that was characteristics of institution quality showed strong persistence in being important factors for economic development. Using some robustness checks we found the two most important institutional variables are rule of law and political stability. We also found that involvement into international trade enhances development level which is in line with Rodriks papers.