Urban concentration and economic growth: checking for specific regional effects

By using a semiparametric specification, we examine the impact of urban concentration in economic growth on different groups of countries that we classify according to a geographical criterion or according to their level of development. Facing a significant proportion of missing data, we handle that problem with a multiple imputation algorithm as advised in the statistical litterature. Therefore using a Bayesian estimation we obtain parametric coefficients and non parametric curves. Then we may perform Yatchew's tests of equality of non parametric effects to check out if the model specification is the same in the different groups of countries.