Frequentist and Bayesian methods of estimating parameters in a non-performing loan model

摘要

In the literature, several macroeconomic economic factors such as GDP, inflation rate, unemployment, and exchange rate have been identified to influence the level of non-performing loan ratio (NPL) in the banking sector. Other macroeconomic variables such as industry production index, stock exchange index, and oil price are also well documented to have strong explanatory power on NPLs. In this study, we examine the effects of some macroeconomic variables (exchange rates (TL/$ and TL/&Euro), industrial production index (IPI), stock exchange index (BIST100), and oil price) on NPL ratio. Further more, we focus on estimating the parameters related to the above variables in a non-performing loan ratio model via Frequentist approach and Bayesian analysis. In the Bayesian method, we provide uninformative and informative priors and a likelihood function that determines the posterior distributions of the parameters. Using Markov Chain Monte Carlo (MCMC) algorithm, we sample the estimates of the parameters from their posterior distributions. The results of the analysis show that the above mentioned macroeconomic variables examined in this study have significant effects on non-performing loan ratio.