|dc.description.abstract||This thesis links the theoretical and the applied literature on interdependence between countries in growth models and their impact on convergence. Economic theory agrees on the existence of interactions between countries, but the empirical literature neglects these interactions. Econometric theory deﬁnes two types of dependence between units, which both needs to be taken care of when estimated.
The thesis consists of three chapters. The ﬁrst chapter presents a growth model, which motivates the weaker type of dependence, spatial dependence. In this model, migration, trade and foreign direct investments act as channels for the interaction of countries. The model predicts positive eﬀects of the interactions, especially of migration. It is common to model the second type of cross-sectional dependence in form of a multifactor error structure model in a heterogeneous slope panel. The model is estimated by the Dynamic Common Correlated Eﬀects estimator, which approximates the dependence by time speciﬁc averages. The second chapter introduces a Stata package to compute this estimator. It discusses practical challenges in its empirical application, presents examples for the estimation and highlights the requirements for the time and cross-sectional dimensions using a Monte Carlo simulation. The ﬁnal chapter combines the contributions of the ﬁrst two chapters. A spatial time lag controls for spatial dependence. The growth model in the ﬁrst chapter is used to motivate the choice of the weights. Strong cross-sectional dependence is taken care of by the methods explained in the preceding chapter. In addition, the chapter uses a general Lotka-Volterra model to determine the type of convergence in the presence of spatial interactions. Lastly, evidence for conditional convergence is presented for a panel of 93 countries.||en