Download Credit Scoring For Risk Managers: The Handbook For Lenders fb2

I started reading this book with a low expectation but I liked the book when I was done with it !. Though it doesn't explain the statistical methods in great detail, it still is a good reading for logistics regression based scorecard development. SAS is frequently referred in the text so it's easy for a SAS user.

The chapters I thought useful:

If the reader is not very familiar with US credit bureau data, Chapter 3 gives a good summary and explains the basic structure of the bureau file its importance to consumer finance.

Chapter 4- Scorecard development, Elizabeth Mays describes step by step how a scorecard is developed using logistics regression. More importantly, those who wondered how the final points (say a FICO of 580) arrived at, scaling and assignment of point weights to attributes are specifically discussed .

Chapter 5 - Variable analysis and Reduction details the information value calculations and commonly used variable reduction methods .

Chapter 13- Scorecard Monitoring Reports describes how to track them using front end and back end reporting. Though this chapter reads exactly like a FICO article on the same subject, its a good reading.

Overall, it does not have the disadvantage many people writing. Depth of some chapters is below expectation - for example Chapter 8 Score based loss forecasting.

I recommend this for anyone in consumer lending business. esp the beginning analysts.

This is a very practical book for anyone in the consumer risk management or non-developers of score models. Explains in plain language the practical aspects of interpreting KS scores (widely used in the credit card/mortgage/consumer finance/auto loan industries), dealing with the very pertinent issue of disparate impact (Reg B) and the even the basics of score variable analysis -including various types of biases that can come in to play.