Bayesian Credit Rating Analysis based on Ordered Probit Model with Functional Predictor

T. Ando (Japan)


Bayesian predictive information criterion; Functional data analysis; Markov chain Monte Carlo.


This paper presents a Bayesian method for the credit rating prediction modeling in the functional data analysis frame work. The credit rating model is designed to include the ef fects of trends in a certain set of accounting variables from ļ¬nancial statements. To estimate the model parameters, the paper presents a Markov chain Monte Carlo sampling al gorithm. The Bayesian predictive information criterion is employed to assess the goodness of the estimated model. The proposed method is applied to Japanese credit rating data listed on the Tokyo Stock Exchange. The results show that the proposed method performs well.

Important Links:

Go Back