Redesigning Credit Risk Modeling Chapter2

Chapter 2: Start from the Goal and Work Backward

I'm no Charles Dickens, but this is my attempt to serialize my book. Many chapters are just a few pages, so my summaries will be brief and give you the main point.

I think the primary design flaw in credit risk analytics is that we start from a tradition of rank-order credit scoring and try to build out. It's like laying bricks with no architectural design. Any senior partner at a top consulting firm will start a new client engagement with a polished version of "What is your problem?", not "What score can I build you?"

In lending, the "problem" is not to improve K-S or Gini in a fixed-horizon credit score. The problem to be solved is how to maximize yield and minimize volatility while staying compliant with regulatory guidelines and ethical standards. That is not as vague as one might think.

Ask Finance if they have an account-level cash flow model for the product. Some do. Some don't. Any worthwhile cash flow model will need monthly (or quarterly) inputs for key drivers such as conditional default probability and conditional prepayment probability, conditioned on the account surviving to the previous period. This is where a credit score needs to connect. It also makes clear that a create score needs to integrate with the product lifecycles and economic scenarios.

Next week, we'll study what a cash flow model needs to look like and why timing is crucial. If you want to read ahead, the full book is at Amazon.

And thank you to those who responded to last week's post by making my little book of ideas the #1 New Release in Banks & Banking on Amazon.

Joseph Breeden
Posted on LinkedIn