Development of CRISIL Ratings Prediction Model

Student Name: A. V. Krishna Kishore

Multiple Logistic Regression and Multiple Discriminant analysis models for CRISIL bond ratings were developed using Liquidity, Financial Leverage, Profitability and Asset Management Quality of Indian manufacturing firms for the calendar years 1999-2001. In this project, instead of standard financial ratios (as was done in the M.Sc.(Engineering) thesis on the same topic), above financial dimensions were measured using certain proprietary ratios used by CRISIL, which are not publicly available. These models were then used to develop a decision support system to be used by clients of CRISIL, which are mainly financial institutions, for predicting ratings of firms, which did not have a CRISIL rating but made the data on their financial fundamentals available; so that the financial institutions had an approximate idea about the kind of CRISIL rating a credit seeking firm would have got, if the firms had indeed gone through the rigors and expenses of obtaining a CRISIL rating for themselves.