Doctor of Philosophy (PhD)



Document Type



This dissertation consists of three essays that contribute to the literature on the estimation of nonlinear spatial panel models. Chapter 2 investigates the performance of both the spatial autocorrelation probit model (SAR) and the spatial error correlation probit model (SEM) with the consideration of temporal effects under the panel data framework. The recursive importance sampling (RIS) method is adopted to compute the integral of the multivariate normal distributed error terms. Chapter 3 estimates a panel probit model with individual fixed effects. A spatial error process and conditional heteroskedasticity in an exponential form is considered. We derive the analytical solution for the average partial effect and its standard error. The estimators are proved to be consistent and asymptotically normal under the Geweke-Hajivassiliou-Keane's (GHK) simulated maximum likelihood framework. We also use a block bootstrap technique to generate estimators' standard errors. Chapter 4 estimates a spatial panel probit model with individual effects using the Gibbs sampler. We assume that the spatial correlation exists in both the dependent variable and the error term. In addition, we assume that individual effects follow the random effects assumption rather than the fixed effects assumption in order to avoid the incidental parameter bias. We illustrate the validity of our model using the simulated data.



Committee Chair

Hill, Carter



Available for download on Monday, March 31, 2025

Included in

Econometrics Commons