Assessing risk is the heart of making smart lending decisions. Faced with an application for a loan, that creditor assesses the amount asked for, the collateral that will be pledged to secure the loan, and the likelihood that a borrower will default. The first two are fairly easy to predict. Loan amounts are calculated to the penny, and the value of collateral can be calculated as well. The difficulty for a lender is to assess the likelihood that a particular person will fail to pay back the loan. Credit scoring is meant to help lenders make that assessment. As more and better scoring models come along, they are updated to provide the service that lenders need: a predictor of default. That is why there have been recent changes to the credit scoring models.
The credit score models attempt to take a range of very different data and make a standardized assessment of it. As hard as it is to quantify, the scores attempt to compare a late payment with a high balance on a credit card, and figure out which one is a better predictor of default. Over time, experience helps to refine those models. But more importantly, consumer habits and behavior change over time also. It used to be that people had one or two credit cards, if any. Now, many more cards can be found in consumers’ wallets.
The credit scores consider a handful of basic information: how much credit do you currently have, how close are you to being at the maximum, and what is your history of repayment? They add in some other factors, such as the length that you have been in a relationship with some creditors and the number of times that other people have been looking at your credit score to see if you are credit worthy. The scores like people who: (a) do not use more than a small portion of their available credit, (b) make their payments on time, (c) do not have lots of accounts or many new accounts, and (d) have not been applying for loans recently.
Recently, the credit scoring companies realized that some of their models were not giving their customers (lenders) the level of detail that the customers wanted. Specifically, as credit tightened and the financial industry went through some turmoil, the customers wanted to be better able to separate people who have occasional infrequent problems with late payments from people who are habitually posing a credit risk to their lenders. The result has been described as adding a little bit of forgiveness into the middle of the credit pack. The stellar credit risks are unaffected, and the people who have a serious problem paying their debts on time are still in trouble. But the people who work hard at paying on time an occasionally stumble are not going to be lumped in with the serious deadbeats.
The scoring agencies also made some changes to recognize the new reality that many people carry multiple lines of credit. The scores are deemphasizing the importance of how many account you have, and putting more emphasis on whether you are maxing out card to the full balances. One lesson from recent years was that when financial bumps in the road come along, consumers with room on their credit lines to absorb some costs are less likely to default.
Finally, the scoring agencies are making it harder for people to piggyback on a strong score to get a bette score themselves. People who are listed as authorized users on other people’s accounts will have a longer and slower road in using that technique to build their own credit.
The changes in the credit scoring models reflect a continuing effort to provide more detail, and useful detail, to lenders who rely on the scores to evaluate risk of default. By refining the models, those lenders can make appropriate decisions about who to lend to, and what interest rate to charge.