Robust estimation of panels with cross-sectional dependence in the presence of outliers

Chua, Teck Thye ORCID: 0000-0001-7867-1670 (2024). Robust estimation of panels with cross-sectional dependence in the presence of outliers. University of Birmingham. Ph.D.

Preview

Chua2024PhD.pdf
Text - Accepted Version
Available under License All rights reserved.
Download (2MB) | Preview

Abstract

Outliers are common in empirical datasets. There is a reluctance to use robust techniques due to lack of expertise, despite the large availability of robust literature. Cross-section dependence in panel data is also the norm and not the exception. The Common Correlated Effects estimator (CCE) is robust to cross-section dependence, but not robust to outliers as it is based on least squares. We improved on the robustness of the current existing robust CCE estimator, the Trimmed CCE (TCCE) estimators. We applied our robust TCCE estimators to economic growth data, and we find overall positive and significant results under aggregated data. We also find that the global economy is more varied in structure than the typical mixed-distribution outlier model in the literature. We explored outlier robust versions of the CCE estimators in general, and discovered potential new robust estimators for further research.

Type of Work:

Thesis (Doctorates > Ph.D.)

Award Type:

Doctorates > Ph.D.

Supervisor(s):