Corporate Information Factory (CIF) Resources by Bill Inmon, Inmon Data Systems

> home > view content

Credit Scoring: Getting Smart

By W. H. Inmon

There is no denying the leverage that can be obtained with the Internet. With the Internet and eBusiness you can reach people you would have never before been able to get to and you can do so in the comfort of their own home, while they are still in pajamas. But what are you going to do once you have their attention? What message are you going to send them?

At the end of the day the message that you send will be much more important than the way you send the message. Creating a smart message ultimately is much more important than sending a smiley face message that cheers up someone’s day but does not bring revenue into the fold.

Being smart is never so important as in the finance business where extending credit is the name of the game. There is an old saying (that is ever so true) to the effect that banks only make loans to those people who don’t need them. Banks don’t really make money loaning to the rich because the rich don’t take out many loans. And banks don’t make money by making loans to the poor who will never repay the loans. Instead banks get rich by making loans to people who are neither rich nor poor but who do have the capacity to repay the loans. This is the sweet spot for banks and this is where they make money.

But how does a bank know who a credit worthy individual is? What is it that banks look for and where do they look in order to assess the credit worthiness of an individual? They look, of course, in a data warehouse. The data warehouse just happens to have the information that banks need to determine the credit worthiness of an individual. Built properly a data warehouse will have in it:

Detailed data,
Integrated data, and
Historical data.

The level of detail will be such that any payment anomalies or disturbances will not be glossed over. The banker needs to get down to brass tacks when deciding how to assess the financial status of an individual. Having summary data about total quarterly profits may be of interest to the loan officer but not insofar as analyzing the loan application for an individual.

Integrated data is a blessing for the loan officer. In a properly constructed data warehouse the loan officer might expect to find:

Net worth, from all known accounts
Real estate holdings and liens on those holdings,
Outstanding loans and other obligations,
Assets, such as cars and businesses,
Income,
Outgo,
Checking profile – stop payments, bounced checks, etc.,
Stocks and bonds,
Other banking relationships,
Marital status,
Occupation,
Number of children,
Education,
Ethnic background, and so forth.

In short, a whole host of otherwise unrelated information is found in the data warehouse for an individual. Some records will be more complete than others, but in each there will be information about which to form an opinion of the credit worthiness of an individual.

The third thing the loan officer can expect to find in the data warehouse is history. The history includes as much financial historical information about the individual as is available. Historical data might include:

Credit card payment history,
Real estate payment history,
Checking account payment and balancing history,
Other loan payment history,
Alimony and child support payment history, and so forth.

In a word, the history of the financial status of the individual seeking a loan is factored into equation of the credit worthiness of an individual.

It is obvious that a data warehouse is at the heart of the analysis of the credit worthiness of an individual. Built properly, a data warehouse contains exactly what the loan officer needs.

Indeed, data warehouses have long been the backbone of what is termed “credit scoring”. Credit scoring is merely the name the loan officer applies to the activity of processing a loan application.

But with the Internet and the Web-based eBusiness environment, there comes the opportunity for online, instant credit scoring. How exactly does online instant credit scoring operate?

Suppose you go online. You seek a bank to whom you are known or to whom you have had a relationship. You go to the bank through the Internet and identify yourself. Once the bank recognizes who you are, you then ask for a loan for a car – a brand new Honda Accord. You tell the bank about the car and about how much the car costs. The bank then says – without hesitation – that you have a loan and that the money for the car will be shipped to the dealer today. You have a loan with no hassles, no qualification, and you have it instantly. You have done your business over the Internet.

How exactly did this transaction occur? What intelligence was occurring behind the scenes that allowed an online Internet loan to occur?

A typical scenario might look like this:

The bank over the years has constructed a data warehouse, for many reasons. The data warehouse contains integrated, historical, detailed data. But the data warehouse is used in a passive manner. One day the bank decides to be a bit more aggressive in its use of information contained inside the data warehouse. The bank decides to create an online file for qualification for loans.

The bank creates a special file where each known customer (or former customer) has a rating that has been established. The bank writes a program to go in and analyze each customer identified in the data warehouse. The bank’s program looks at the data just the way that a loan officer looks at the data and predetermines a loan limit for each type of loan for a customer. The results of the program passing through the data warehouse might look like:

Mary Adams	car loan	$20,000	home loan	$100,000	equity loan	$5,000
John Brown	car loan	$15,000	home loan	$125,000	equity loan	$25,000
Sally Clark	car loan	$7,500	home loan	______	equity loan	______
Paul Drake	car loan	$35,000	home loan	$500,000	equity loan	$200,000
Ernie Els	car loan	$95,000	home loan	$2,000	equity loan	open

……………………………………………………………………………………….

Now, when one of the customers calls in or enters through the Internet, the predetermined credit limits are established and the customer only has to sign a paper authorizing the transfer of funds and the loan is secured. The loan can be made with no fuss or bother and can all be done over the Internet.

There are two circumstances which warrant discussion. Suppose Mary Adams enters through the Internet and asks for a car loan of $25,000. This car has surpassed Mary’s preauthorized limit and has to be handled manually by a loan officer. The loan officer needs to manually assess the reasons for Mary’s purchase and whether she will get the loan.

The other circumstance is what happens when Noel Anderson enters the system and asks for a loan. Noel is not a person that is known to the system. Noel may well be a very credit worthy individual, but he does not appear in the data warehouse. Therefore the credit application of Noel must be handled by hand.

The place where the pre-approved loan amounts is stored is called an operational data store, or an ODS. The pre-approved list is not stored in a data warehouse because if it were stored in a data warehouse it would not be as quickly available. Data stored in an ODS is available in a matter of a second or two. Data stored in a data warehouse may not be available for an hour or two (or in some cases even longer). The Internet user cannot afford to wait an hour. Therefore the data needed on the Internet in an online mode must be stored in an ODS.

Periodically the bank’s credit analysis program must be rerun in order to make sure that up to date information is used in the analysis for pre-approved credit limits. Typically the rerun of the program is done every six months or so, as the basic data needed for credit authorization does not change very quickly.

Credit scoring then is one more example of how corporate data provides the foundation for smart eBusiness.