Fixed effects model and its estimators: benefits and drawbacks

This thesis addresses the escalating emphasis within theoretical and empirical research on panel data analysis. Despite the increasing prevalence of the fixed effects model in the empirical literature, the descriptions of relevant estimators often remain overly simplistic even in advanced econometrics literature. This study undertakes an extensive examination of the theoretical underpinnings of the fixed effects estimator and thoroughly delineates the evidence regarding its consistency and asymptotic normality across various panel size configurations. Notably, the fixed effects model entails robust assumptions that frequently do not align with real-world conditions. Particularly, the assumption of random sampling is frequently violated, especially in the analysis of macroeconomic data. Nevertheless, empirical research in the domain of macroeconomic data predominantly relies on the fixed effects model, often overlooking the assumptions inherent in this estimator. Moreover, the paper presents simulation studies that illustrate the behaviour of the estimator when selected assumptions of the fixed effects model are violated in the context of more complex panel data structures. The outcome of these simulations indicates significant bias in the parameter estimates at different settings and contributes to a deeper understanding of the strengths and limitations of fixed effects estimators in panel data analysis. They highlight the importance of carefully considering the underlying assumptions and characteristics of the data when using fixed effects estimators, particularly in cases involving non-random selection or dynamic processes.