A0347
Title: Estimation of unbalanced matrix exponential spatial panel data model with interactive fixed effects
Authors: Ye Yang - Capital University of Economics and Business (China) [presenting]
Osman Dogan - Technical University Istanbul (Turkey)
Suleyman Taspinar - CUNY Queens College (United States)
Abstract: An unbalanced matrix exponential spatial panel data model is considered that allows for spatial dependence in the dependent variable, interactive fixed effects, and spatial dependence in the error term. It is shown that this model can arise as a best-response function from the maximization of a quadratic utility function, accounting for all levels of externalities generated by the heterogeneity of spatial units. An m-estimation approach is introduced based on the adjusted score functions of the quasi-log-likelihood function of the model under both homoskedastic and heteroskedastic error terms. For statistical inference, an analytical bias-correction approach is proposed that combines the sample counterpart of the expectation of the Hessian matrix with the plug-in estimator of the variance of the adjusted score functions. Through extensive Monte Carlo simulations, it is shown that the proposed M-estimator and inference method exhibit good finite-sample properties. In an empirical application, spatial dependence is examined in foreign direct investment (FDI) across administrative divisions in mainland China.
