Detection of changes in panel data

This thesis deals with the issue of parameter stability in time in a linear regression model for panel data, specifically in the context of Fama-French models. First, these models are introduced, then the principles of test statistics designed to identify changes in the model are elaborated, resulting in statistics adapted to datasets with a high number of units in the panel and a short time range. The theoretical findings are then applied in the empirical part, which analyzes data from European stock markets with a focus on the period of market shock caused by the COVID-19 pandemic.