A0873
Title: Prediction-oriented modeling of three-dimensional panel data
Authors: Markus Fritsch - University of Passau (Germany) [presenting]
Harry Haupt - University of Passau (Germany)
Daniel Henderson - University of Alabama (United States)
Joachim Schnurbus - University of Passau (Germany)
Abstract: In the last decade, worldwide migration has been on the rise, and in 2020, an estimated 281 million people (3.6 percent of the world's population) lived in a country in which they were not born. Due to the large proportion of people on the move, modeling international migration flows, understanding their determinants, and predicting expected future migration is of great interest to political decision makers, researchers, and society in general. Starting from a general framework for modeling migration flows based on three-dimensional panel data when accounting for multi-dimensional fixed effects, dynamics, return flows (often referred to as 'reciprocity' in the literature), and asymmetric effects, we detail assumptions frequently encountered in empirical research that lead to a wide range of model specifications. It summarizes how to obtain consistent estimates of the model parameters by providing moment conditions for estimation with the generalized method of moments (GMM) and commenting on inference and the related weighting matrix. The approach is illustrated by modeling decadal and quinquennial migration flows into Europe from 1960 to 2019 based on potentially relevant economic, climate, and conflict variables while accounting for different types of fixed effects.
