Economics Seminar - IV Estimation of Spatial Dynamic Panel Data Models with Interactive Effects

This paper develops a new Instrumental Variables approach, designed for spatial, dynamic panels with interactive effects under large N and T asymptotics. For this class of models, most approaches available in the literature are based on quasi-maximum likelihood estimation. The approach put forward here is appealing from both a theoretical and a practical point of view for a number of reasons. Firstly, it is linear in the parameters of interest and computationally inexpensive. Secondly, the IV estimator is free from asymptotic bias that typically arises due to the incidental parameters problem. Thirdly, the approach can accommodate endogenous regressors, so long as external instruments are available. For the model with homogeneous slopes, we put forward a pooled, two-stage IV (2SIV) estimator, which is consistent and asymptotically normal as N, T → ∞, such that N/T → c, where 0 < c < ∞. For the model with heterogeneous coefficients, we develop a sqrt(N)-consistent Mean-Group IV (MGIV) estimator, which involves averaging of individual-specific estimated slopes. Currently, there is no method in the literature that allows for such level of heterogeneity in dynamic spatial models with interactive effects. Monte Carlo evidence shows that the finite sample performance of all IV estimators and associated statistical tests is satisfactory even for small values of N and T. As an illustration, we study the determinants of risk attitude of banking institutions. We mainly explore two questions: to what extent was the credit risk-taking behaviour of U.S. banking institutions affected by the risk attitude of their peers during the period that culminated in the GFC? Has such a relationship changed after the Dodd-Frank Wall Street Reform and Consumer Protection Act of 2010?