Empirical Validation of Robust Estimators in Panel Data Models under Multicollinearity, Heteroscedasticity, and Autocorrelation

Aims: The study aims to simultaneously address multicollinearity, heteroscedasticity, and autocorrelation, which commonly undermine the reliability of conventional estimators such as Ordinary Least Squares (OLS), Feasible Generalized Least Squares (FGLS), First Difference (FD), and Between Estimators (BTW). Study Design: Quantitative research design Place and Duration of Study: The study used a real Cigarettes SW panel dataset from the AER package for 50 US states from 1970 to 2000 Methodology: We proposed new estimators which include the Robust Shrinkage GMM (RSGMM), Panel Adaptive Ridge GMM (PARGMM) and Heteroscedasticity-Autocorrelation-Robust Shrinkage GMM (HARSGMM). The metrics considered included bias, variance, mean squared errors, efficiency and robustness. Results: The results revealed that all estimators indicated the positive impact of tax and population on cigarette price and the negative impact of income levels on cigarette prices. The proposed estimators such as HARSGMM achieved 1.74 efficiency and lowest MSE 0.0922 and RSGMM achieved 1.52 efficiency and 0.1003 MSE, which consistently outperform traditional estimators. Similarly, the findings indicated that HARSGMM and RSGMM are more robust estimators given their low condition numbers, variances and high efficiency. Conclusion: This study recommends the broader application and integration of these robust techniques into econometric software and policy-oriented research. This will be beneficial for empirical researchers, students and the government organizations when dealing with complex panel datasets.