The objective of this study is to reexamine the existing models to estimate the credit risk of financial assets and to determine the prices of credit-risky assets. It assesses the applicability of those models in the Korean financial market and points out the areas in which the improvement can be made.

For this purpose, this study starts with discussions on the models to analyze and estimate credit risk. The models have been broken down into two broad categories such as Markov and non-Markov models. Markov models discussed here include the multivariate discriminant analysis, the probit and logit models, the neural network analysis, and the repetitive partitioning approach. The non-Markov models discussed here include the KMV model, the credit rating migration analysis, and the age analysis.

Based on the credit risk and default probability to be estimated using the models discussed above, we are now able to determine the prices of credit-risky financial assets. The pricing models are now categorized into two groups, no-arbitrage models and equilibrium models. No-artbitrage models can be further broken down to structural models and reduced-form models. These pricing models are discussed and summarized in this study.

For the empirical part, credit risk has been analyzed through the credit rating migration analysis. Along with the assessment of credit risk, the prices of hypothetical credit-risky assets has been analyzed in line with the reduced-form model provided by Jarrow and Turnbull (1995) and Jarrow, Lando, and Turnbull (1997) and also in line with the equilibrium model developed in this study. In addition, for the term structure of risk-free rates used in the above analyses, models such as Cox, Ingersoll, and Ross (1985) and its extended versions have been estimated and compared based on the Korean CD yields data.