SELECTING A PORTFOLIO OF CREDIT RISKS BY MARKOV CHAINS.

Abstract : The paper assumes that a credit applicant, will, if accepted, pay debts in a probabilistic manner as described by a finite state Markov Chain. There are assumed to be a number of credit classes described by different Markov chains. Using the results of a previous paper with H. Davidson (Estimation of the Allowance for Doubtful Accounts by Markov Chains) AD-258 182 the authors compute the present value of accepting a customer in given class, and his net addition to overall expected value and variance. Customers are accepted beginning in the best credit category and proceding to successively less good categories, until the coefficient of variation exceeds 1 percent, at which time no further customers are accepted. This model thus permits dynamic evaluation of the portfolio of credit risks which take into account the riskiness of customers already accepted.