Panel data modeling: Identifying and handling outliers with the VSOM approach.

Area of Science:

• Econometrics
• Statistical Modeling

Background:

• Outlier identification is critical in panel data modeling, but existing methods are limited.
• Visual detection of outliers in panel data is often challenging.

Purpose of the Study:

• To introduce and evaluate the Variance Shift Outlier Model (VSOM) for outlier detection in panel data.
• To demonstrate the effectiveness of VSOM in both single and simultaneous equation models.

Main Methods:

• The Variance Shift Outlier Model (VSOM) approach is employed.
• Identification of outliers using the square of the standardized residual.
• Parametric bootstrapping to generate the distribution of squared standardized residuals.
• Downweighting outlier variance using a D matrix to reduce their impact.

Main Results:

• The VSOM approach effectively identifies and handles outliers in panel data models.
• Application to GDP and FDI data from ASEAN-China Free Trade Area (ACFTA) countries.
• The VSOM model achieved a lower sum of squared residuals (SSR) compared to the null model, indicating improved model fit.

Conclusions:

• The VSOM approach enhances the accuracy and reliability of panel data models.
• This method offers a robust solution for managing outliers in econometric analysis.
• Improved model fit suggests VSOM successfully captures underlying economic relationships.