312507 Panel Data Analysis (V) (SoSe 2021)

This lecture covers the modeling of panel data. These are datasets where variables for a number of individuals (typically persons, regions, countries) are observed over a number of time points (often years, quarters or panel waves). Such data sets offer advantages in the modeling of real processes compared to cross-sectional data (where observations only occur at one time point) or time series data also called longitudinal data (where observations for only one individual occur).

We start with the most often used static linear models for continuous dependent variables. This is subsequently extended to nonlinear models for categorical dependent variables. We also discuss time series models in the panel context. If time permits this will also include non-stationary time series models.

The lecture assumes that students are familiar with the main concepts of statistical and econometric modeling such as least squares estimation and the maximum likelihood paradigm. Also the generalized method of moments is used in the lecture. All concepts are properly introduced.

The lecture mostly follows the classic book by Badi Baltagi: Econometric Analysis of Panel Data, John Wiley and Sons, 2005 (third edition).

The information is conveyed online via videos in Panopto supplemented with slides and R-scripts (where appropriate). Additionally bi-weekly Q&A sessions aid the understanding of the material.

The course is taught in the language of science, broken English.

Details on the examination depend on the study programs.

Bibliography

Badi Baltagi: Econometric Analysis of Panel Data, John Wiley and Sons, 2005 (third edition).