FRONTIERS IN ECONOMETRICS
Topics in Advanced Econometrics
Lecturer: Professor Dr Peter Egger (ETH Zürich, Department of Management, Technology and Economics, KOF)
Date: 13 February to 17 February 2017
Venue: Halle Institute for Economic Research (IWH) – Member of the Leibniz Association, Kleine Maerkerstrasse 8, 06108 Halle (Saale), Germany, conference room (ground floor)
Registration: until January 15, 2017 via e-mail: annett.hartung@iwh-halle.de. The course is designed for at most 25 participants.
Announcement see pdf
Course outline
The aim of this course is to familiarize students with the use of some cross-sectional modeling techniques, primarily, to tackle endogeneity and interdependent data as well as non-i.i.d. disturbances. Most of the lecture will take these techniques to the case of panel data with a repeated observation of cross-sectional units over time.
The exam will be a term paper, written in groups of 2-3 people. The subject of the paper can be coordinated during the course.
Schedule
13 February 2017: 9:00-12:00
14 February 2017: 9:00-12:00 and 14:00-17:00
15 February 2017: 9:00-12:00 and 14:00-17:00
16 February 2017: 9:00-12:00 and 14:00-17:00
17 February 2017: 9:00-12:00
Additionally, we invite you to attend the presentation of Peter Egger in our research seminar on February 13, 2017, 14:15–15:45.
CROSS-SECTION MODELS
1. Review of OLS for single equations
a. with i.i.d. errors;
b. under heteroskedasticity;
c. with spatial autocorrelation in the disturbances;
d. seemingly unrelated regressions.
2. Endogeneity
a. with a single outcome equation (two-stage least squares);
b. with a system of equations (three-stage least squares);
c. with a spatial lag of the dependent variable.
3. Models for systematically missing data as a special case of endogeneity
a. sample selection;
b. treatment selection.
PANEL-DATA MODELS
1. One-way panel data models with exogenous regressors and homoskedastic error components:
a. fixed-effects models using least-squares dummy-variable estimates versus within estimates by least squares;
b. random-effects estimation with generalized least squares;
c. testing for fixed effects and random effects;
d. testing of fixed effects against random effects.
2. Two-way and multi-way panel data models with exogenous regressors and homoskedastic error components:
a. fixed-effects models using least-squares dummy-variable estimates versus within estimates by least squares;
b. random-effects estimation with generalized least squares;
c. mixed fixed and random effects models.
3. One-way panel data models with exogenous regressors and non-i.i.d. error components:
a. heteroskedasticity;
b. serial correlation in the disturbances;
c. spatial autocorrelation in the disturbances;
d. seemingly unrelated regressions.
4. Endogeneity
a. with a single outcome equation (two-stage least squares for fixed effects and two-stage generalized least squares for random effects);
b. with a system of equations (three-stage least squares);
c. with a spatial lag of the dependent variable;
d. with a time lag of the dependent variable (dynamic panels).
5. Models for missing panel data as a special case of endogeneity
a. Randomly missing data;
b. sample selection;
c. treatment selection.
References
Baltagi, Badi H. (2013): Econometric Analysis of Panel Data, Wiley.
Kapoor, M., Kelejian, H. H., Prucha, I. R. (2007): Panel Data Models with Spatially Correlated Error Components, Journal of Econometrics, 140, 2007, 97-130.
Kelejian, H. H., Prucha, I. R. (1999): A Generalized Moments Estimator for the Autoregressive Parameter in a Spatial Model, International Economic Review, 40, 509-533.
Kelejian, H. H., Prucha, I. R. (2010): Specification and Estimation of Spatial Autoregressive Models with Autoregressive and Heteroskedastic Disturbances, Journal of Econometrics, 157, 53-67.
Wooldridge, J. M. (2010): Econometric Analysis of Cross Section and Panel Data, MIT Press.