Instructor: Jeffrey M. Wooldridge, Professor in the Department of Economics, Michigan State University, USA.

Course Objective

The course focuses on methods for exploiting and modeling spatial correlation in econometric applications. It will cover both cross-sectional applications and panel data. We will start with linear models, allowing for exogenous and endogenous explanatory variables. But we will also cover methods for estimating nonlinear models with spatial features. Along the way we will use methods for cluster samples as simple alternatives that exploit spatial structures yet are computationally simple.


Li Yijie , ASSEE 2017

The whole program reaches an excellent balance!

I learnt a lot about duration analysis, including both theories and how to do empirical analysis with it. The lab sessions are very helpful in improving my understandings of the contents of the lectures. The course is great, and the dinners and excursions are also great.

Day 1: Course overview

The nature of spatial correlation. Models with spillover effects. Policy analysis with spatial structures. The suitability of models with spatial lags. Ordinary Least Squares estimation of spatial models and robust standard errors.

Day 2: Generalized least squares and quasi-GLS estimation of linear models

Models with endogenous explanatory variables and instrumental variables. Generalized Method of Moments Estimation.

Day 3: The nature of spatial correlation and serial correlation with panel data

Linear panel data models with unobserved heterogeneity. Fixed Effects, Random Effects, Correlated Random Effects Estimation with Spatial Data. Instrumental variables methods for spatial panel data.

Day 4: Nonlinear models with cross-sectional data and spatial data

Maximum Likelihood Estimation, partial MLE, Generalized Estimating Equations.

Day 5: Nonlinear models with panel data

Models with strictly exogenous explanatory variables. Unobserved heterogeneity. Binary response models. Fixed Effects Poisson estimation and its robustness properties. Incorporating lagged dependent variables with spatial correlation.

Reading List:

"Using generalized estimating equations to estimate nonlinear models with spatial data," arxiv, (2018) link

"Quasi-generalized least squares regression estimation with spatial data," Economics Letters, (2017) link

"Partial Maximum Likelihood Estimation of Spatial Probit Models," Journal of Econometrics, (2013) link

Econometric Analysis of Cross Section and Panel Data, Jeffrey M. Wooldridge (2010) link