Portfolio Efficiency Tests with Conditioning Information-Comparing GMM and GEL Estimators

Area of Science:

• Econometrics
• Financial Modeling
• Statistical Inference

Background:

• Conditional information is crucial for efficient portfolio management, enabling asset allocations to leverage all available market data.
• Generalized Empirical Likelihood (GEL) estimators offer statistical advantages, including robustness to model misspecification and improved finite sample performance.
• Traditional Generalized Method of Moments (GMM) estimators face challenges with increasing bias as more moment conditions are added, a common issue in conditional efficiency analysis.

Purpose of the Study:

• To evaluate the efficacy of Generalized Empirical Likelihood (GEL) estimators in portfolio efficiency tests within asset pricing models.
• To assess the performance of GEL estimators concerning size, power, and robustness, particularly when utilizing conditional information.
• To compare GEL estimators against traditional methods like GMM in the context of conditional efficiency analysis.

Main Methods:

• Utilizing Monte Carlo simulations to rigorously test the performance of GEL estimators under various conditions.
• Conducting extensive empirical analyses across different sample sizes and portfolio types for two prominent asset pricing models.
• Evaluating the impact of data contaminations, including heavy tails and outliers, on the performance of GEL estimators.

Main Results:

• GEL estimators demonstrate superior performance in portfolio efficiency tests, especially when dealing with data contaminated by heavy tails and outliers.
• The bias of GEL estimators does not escalate with the number of moment conditions, unlike GMM estimators, making them more suitable for conditional efficiency analysis.
• Monte Carlo experiments confirm that GEL estimators exhibit better size, power, and robustness properties in finite samples.

Conclusions:

• Generalized Empirical Likelihood (GEL) estimators provide a more robust and reliable approach for portfolio efficiency tests in asset pricing, particularly with conditional information.
• The findings suggest that GEL estimators are advantageous in practical portfolio management scenarios where data quality may be compromised.
• The study advocates for the adoption of GEL estimators in financial econometrics for enhanced accuracy and reliability in asset allocation and performance evaluation.