Spatial dynamic panel data models with interactive effects

We introduce a new instrumental-variables (IV) approach for spatial dynamic panel-data models with interactive effects under large N and T asymptotics. Alongside the methodology, we present the spxtivdfreg spxtivdfreg spxtivdfreg spxtivdfreg package, which implements the proposed approach in Stata. Most existing approaches in this literature rely on quasi–maximum likelihood estimation. Our IV approach is appealing from both theoretical and practical standpoints for several reasons. First, it is linear in the parameters of interest and computationally inexpensive. Second, the IV estimator avoids the asymptotic bias that typically arises from the incidental parameters problem. Third, the approach accommodates endogenous regressors, provided suitable external instruments are available. For the homogeneous-slope case, we develop a pooled two-stage IV (2SIV) estimator, which is consistent and asymptotically normal as N and T grow large. For the heterogeneous-slope case, we propose an N - consistent mean group IV (MGIV) estimator based on averaging individual- specific estimated slopes. To our knowledge, no existing method in the literature allows for this level of heterogeneity in dynamic spatial models with interactive effects. We also provide practical guidance on how best to run these methods in Stata.