Are Some Banks More Lenient in Implementation of Placement Classification Rules?* An Application of Dichotomous Rasch Model to Classification of Credit Risk in the Banking System Tomislav Ridzak, Financial Stability Department Croatian National Bank *The views expressed in this article are those of the author and do not necessarily represent the views of, and should not be attributed to the CNB
17
Embed
Are Some Banks More Lenient in Implementation of Placement Classification Rules?* An Application of Dichotomous Rasch Model to Classification of Credit.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Are Some Banks More Lenient in Implementation of Placement Classification Rules?*
An Application of Dichotomous Rasch Model to Classification of Credit Risk in the Banking
System
Tomislav Ridzak, Financial Stability DepartmentCroatian National Bank
*The views expressed in this article are those of the author and do not necessarily represent the views of, and should not be attributed to the CNB
Motivation
Evaluation of credit risk in the portfolio is a key issue in bank management: Loss on a loan translates in to profit and loss and
influences capitalization level through increased loan provisions
If bad loans are not accounted for in a truthful manner, in the limit the bank stability is at stake
The loan classification is therefore important for bank management, depositors, owners and naturally regulators
Introduction
Loan classification in most countries involves substantial subjective judgement (World bank study by Laurin and Majnoni, 2003)
Inspecting the loan classification used by banks is difficult and costly (in terms of time and data)
This research compares the differences in placement classification of a common portfolio and obtains estimates of strictness / leniency for each bank
Related literature
Carey (2001) presents one of the first attempts to tackle the issue of consistency of banks’ ratings comparing ratings by different lenders to the same borrower
Hornik et al. (2007) use information from all possible bilateral comparisons and then detect outlying banks
Jacobson et al. (2005) use the sample of common borrowers rated by two banks and show there are substantial differences in the implied riskiness between the banks
Rasch model
Rasch model is used in order to obtain stricness / leniency estimate
The model was developed in order to separate measures of person ability (B) and item difficulty (D) in education research
It can be shown that the odds of a correct response by a person to one question, conditional on answering at least one of them is equal to difference between question difficulties
)()1( inn DBfxP
Rasch scores explained
The more able you are, higher the probability of getting the answer right
Bank leniency and application of the Rasch model
The credit risk classification is far from being a well established program with minimal human interaction
The Rasch model enables ranking of the banks according to their strictness by treating the banks as examiners and the companies as examinees
As a result the strictness / leniency estimate for each bank is obtained
Sample credit risk classified by number of company – bank links