Choosing Banks’ Profitability Predictors: A Comparison of Logistic Regression and Discriminant Analyses

  • Saadut Ullah Khan Government College of Management Sciences, Kohat, Pakistan
  • Zahid Iqbal
Keywords: Profitability, logistic regression, return on assets, discriminant analysis

Abstract

For researchers, banks' profitability has been a key factor. Various internal and external factors may have an impact on the banks’ profitability. Internal factors are unique to each bank and have an impact on management decisions that affect bank profitability. External factors are those outside of the banks' control that can affect how well they perform. This study examines significant financial variables identifying profitability perspectives and two predictive models of banks’ profitability using the banks' data from the period of 2007-2016. Logistic regression and discriminant analyses were used and evaluated their predictive accuracy. Although, logistic regression and discriminant analysis are unusual for such studies as the dependent variable is continuous and these techniques are applicable when the dependent variables are categorical. So, in order to convert the dependent variable i.e. return on assets to binary variable, ’1’ is assigned for a positive return on assets and ‘0’ for a negative return on assets. The empirical results described that there is a statistical difference between the prediction results of logistic and discriminant analyses. The discriminant models identified that non-interest income, interest rate, operating efficiency, and bank size have a significant effect on banks’ profitability. On the other hand logistic regression showed there are two significant factors i.e. interest ratio and bank size. In the comparison of the models and the accuracy rate of discriminant analysis is higher than that of logistic regression

Published
2023-06-29
How to Cite
[1]
S. Khan and Z. Iqbal, “Choosing Banks’ Profitability Predictors: A Comparison of Logistic Regression and Discriminant Analyses”, PakJET, vol. 6, no. 2, pp. 9-12, Jun. 2023.