Analyzing Systematic Risk in Global Airline Industry: A Machine Learning Approach

Abstract

Beta (β) is associated with the systematic risk of an asset, which is one of the important concepts used in many areas. Accurate estimation of β enables investors, portfolio managers, risk analysts, and financial decision-makers to make effective decisions. Accurate β estimation requires the use of accurate analysis and modeling techniques. In this study, the systematic risk of airline companies traded on the U.S., European, and Turkish stock exchanges between 2015 and 2023 is evaluated, and the impact of different methods on β estimates is analyzed. Furthermore, the impact of machine learning (ML) algorithms on β estimation compared with traditional regression methods is examined. It is concluded that Lasso Regression is the algorithm with the lowest error and the highest R2 value among the methods used in the analysis of the data set consisting of airline companies, and the index returns on which β estimates obtained with ML algorithms are different from traditional regression indicate that the methods and model specifications used in the calculation of β estimates may change. ML algorithms improve forecasting performance by providing a more flexible and sophisticated modeling approach.