Estimating regression model with a mixed structure - roots and theory.

Abstrakt

The historical development of models with fixed and random effects and their differences with respect to the model with dummy variables is explained. The robustified versions of the classical methods are proposed and in the second part of paper subject to the numerical study revealing their viability.

Klíčová slova

• Robust estimation
• regression model
• fixed and random effects
• Hausman test
• numerical study.