Advancing Robust Estimation in Simultaneous Equation Models: Application of New Techniques to the Klein Model I Dataset

This study aims to test the effectiveness and robustness of the five newly developed robust estimation techniques in SEM namely; ARIV, G2SAE, ENIV, HCGMM and 3SAEN by Okeke N.C, et al on real life data sets. The real-life application of the proposed estimators was conducted using the Klein Model I dataset, a well-known macroeconomic dataset comprising U.S. national income accounts from 1920 to 1941 (excluding war years). This dataset is commonly used to test simultaneous equation models due to the interdependence of macroeconomic variables. In our structural equation model (SEM), we modeled the relationship between consumption, investment, income, and government expenditure. The dataset can be accessed through the AER package in R or other econometric data repositories. The dataset is known to have both problems (heteroscedasticity and multicollinearity) we are addressing in this study. Empirical results show that HCGMM and 3SAEN frequently strike a favorable trade-off between bias reduction and estimation efficiency in the presence of multicollinearity and heteroscedasticity, yielding approximately (0.14%) lower RMSE and improved AIC/BIC among other newly developed methods.