Dynamic Models and Reciprocity

This chapter discusses the specification, estimation and testing of dynamic models with multi-dimensional data. The difficulties in estimating dynamic models in standard two-dimensional panel data are well known and these challenges are exacerbated by the more complicated endogeneity problems associated with using multi-dimensional data. Furthermore, the availability of multi-dimensional data allows proper modelling of reciprocity. This chapter analyzes a general model containing both reciprocity and short-run dynamics. It is straightforward to show that endogeneity is an inherent feature of the general model and least squares type estimators will be inconsistent. A set of valid orthogonal conditions is proposed, which is then used in Generalized Method of Moments (GMM) estimation.