The course could be conceived in several ways:
- A more intuitive voyage touching some important issues in panel data modelling and estimation, focussing on the underlying rationale and pointing to implementation possibilities using open source (and possibly free-of-charge) software such as R. The basic idea behind this apprroach is to show that probably most of the problems facing applied researchers can be solved by not relying on standard (commercial) software.
- An overview of some panel data models, with examples using a popular commercial econometric software package, such as Stata.]
- A more theoretical exploration covering rather recent panel data material.
The course level is appropriate for participants with background in economics, statistics, mathematics, and/or public policy. A strong background in quantitative analysis is required. Basic knowledge of the STATA statistical software is required.
A general fluency in the statistical/econometric lingo at the (post-) doctoral level (hopefully in a non-statistic/econometric discipline) is required. More specifically, the Law of Large Numbers and the Central Limit Theorem should be understood.
Every participant is allowed to submit one application, no later than 4 weeks before the course. A selection wil be made from the submitted applications. Some examples will be provided illustrating a possible approach to the proposed problem at hand.
1. Intro: Advantages of Panel (Longitudinal) Data
- Some Common Estimators: OLS - IV - GLS - GMM - ML - [ME]
- Some Common Tests: Likelihood ratio test - Lagrange multiplier (score) test - Wald test - Hausman Test - Sargan-Hansen J -Test
- Models, parameters of interest and the incidental parameters problem
3. Linear Models
- Static: Uncorrelated individual effects - Correlated individual effects – Hausman test
- Dynamic: No exogenous regressors - Exogenous regressors - Serially corelated errors, detection and treatment
- Random Coefficient & Correlated Random Coefficient Models
4. Non-Linear Models
- Static models: Random effects - Fixed effects - Bias reduction - [Orthogonal Parameters]
- Dynamic models: The initial conditions problem - Dynamic discrete choice and duration - Random Effects Tobit
5. Additional Issues
- Cross-sectional dependency and its treatment
- Variance estimation for general one- and two-way dependence; use of standard (cross-sectional) estimators with adjused standard errors
- Attrition - Sample selection
- Multilevel models
ABOUT The Instructor
Rembert De Blander, KU Leuven and Université catholique de Louvain: Rembert De Blander is a researcher at the Public Economics research unit at KU Leuven, Belgium, where he coordinates the Flemish Models of Simulation (FLEMOSI) project. In addition, he is an assocaite researcher at the Earth and Life Institue of the Université catholique de Louvain (UCL), Belgium. HIs main research interests are in the field of applied (panel data) micro-econometrics.
Click here to apply for this course. Applications are welcome until August 17, 2012.
Click here to review STATA. These materials are available for only course participants.