METHODS OF ROBUST FACTOR ANALYSIS: A COMPARATIVE STUDY WITH APPLICATION TO THE EGYPTIAN BANKING SECTOR

Factor analysis is a vital statistical technique for understanding the underlying structure of complex datasets by identifying latent variables that explain the correlations among multiple observed variables. However, traditional factor analysis methods, such as Principal Factor Analysis (PFA), are sensitive to outliers and noise, which can distort results and limit their applicability in real-world datasets. Robust factor analysis is essential for accurately identifying latent variables in datasets with significant contamination. This study provides a comprehensive evaluation of three robust factor analysis techniques Minimum Covariance Determinant (MCD), -norm Exploratory Factor Analysis -norm EFA), and Principal Component Pursuit (PCP) in comparison to the Principal Factor Analysis (PFA) method. Using a real-world dataset comprising monthly stock market data from the Egyptian banking sector, including stock prices, trading volumes, and 14 key economic indicators over 10 years (2014-2024), the study examines the effectiveness of these methods in identifying meaningful factor structures under challenging conditions of noise and outliers. The findings offer valuable insights into the performance of robust methods in handling data contamination, improving factor estimation, and providing more reliable insights for financial decision-making. This paper contributes to the growing need for robust analytical techniques in financial applications, guiding researchers and practitioners in selecting reliable and appropriate factor analysis methods for complex and noisy data environments.