Identification of the Factors Affecting Stored Corporate Cash in Tehran Stock Exchange: Robust Variable Selection Technique

Abstract

Objective: Based on the existing theories, many variables may affect stored corporate cash. And on the other hand, occurrence of outliers in financial variables is common. This affects the performance of statistical methods, especially variable selection based on the ordinary least squares (OLS). In this article using robust variable selection, the most important factors affecting stored cash by firms accepted in Tehran stock exchange (TSE) are identified. Methods: To identify the most important predictors of stored cash, we apply robust variable selection technique based on the SCAD penalty function. For this purpose, the data from 196 stock firms in the period of 2009-2011 were analyzed. Results: Findings showed that robust variable selection identifies variables affecting stored corporate cash which traditional regression methods such as OLS can't identify. Conclusion: Presence of outliers in financial variables may spoil the performance of the traditional ordinary least squares (OLS) method. This problem highlights the importance of using robust techniques in empirical finance.

Keywords

• cash holdings
• outliers
• penalty function
• robust regression
• variable selection