Credit scoring explained

Why assess creditworthiness in the first place? A wrong decision from a credit grantor is not good for anyone. Not certainly for the bank which will be hit by a loss along the road. Not for regular good payers who can experience higher interests or commissions in an attempt from the bank to make up for the loss. And it's not good for the actual defaulter who can face mild to severe inconveniences ranging from getting a bad credit report to actually losing his house.

The introduction of credit scoring has to do with automation of credit decisions, much in the way automation has invaded all kind of industrial activities. Back in the late 1950s Bill Fair and Earl Isaac devised a way of predicting payment behaviour of creditors based on information known by lenders at application time.

The idea was somehow ahead of its time for two reasons. As with any other revolutionary idea, banks have to be convinced to switch from fully manual credit underwriting procedures to automated ones. And the computing resources were worse than primitive those days. Despite these obstacles, they managed to start a successful company (Fair Isaac) and a whole industry in itself.

I do not know how many of you have faced the task of grading exams as a teacher, but most of us have been victims of it as students. Think of a test with several questions which you answer and the teacher reviews later. Depending on your answers you will be granted points on the individual questions. Adding them all up you will get a final mark.

This describes the mechanics of computing a score quite well. You are the student, the teacher is the bank (apologies for the analogy) and the questions are several pieces of information that can be found out at application time. Needless to say, the bank considers these questions relevant to the credit risk associated with the credit facility you are applying to. The points are called weights assigned to the possible answers to the questions. And the final mark is... yes, it's the credit score associated with this particular credit application.

Wait a minute! Did I say that a score is linked to a credit application? Isn't it common knowledge that every consumer has a credit score independently of the bank or the product applied for? The meaning carried by the words "credit scoring" is twofold. While for the general consumer it means a score calculated by a credit bureau, within the banking industry it means a score computed by the bank and using the variables and weights specific for that bank.

The basic principle behind scorecard building is as simple as this: future loans will behave similarly to past loans. Therefore, if homeowners have shown lower delinquency rates than renters in the past, it is likely that this behaviour will hold also for future loans.

For this to be acceptable, a certain mass of past loans is needed to support this evidence statistically. Actually Statistics is at the foundations of credit scoring development. A scorecard builder will scan all data available at application time and assess its eligibility as predictors of credit defaults before assigning weights to a subset of those data using specialised statistical techniques.

If we remember that credit scoring is

• A tool to automate credit decisions


• Applied systematically by computer programs


• Derived statistically from past experiences

then we know a lot about the subject. And all this by reading this short lens!