A scoring model is a formula that assigns points based on known information to predict an unknown future outcome.
The most well known example of a scoring model is the “credit score” or “FICO score” used by lenders to predict the probability of a customer defaulting on a loan. A credit score rank orders customers by the probability they will default, with a high score indicating a low probability of default and a low score indicating a high probability of default. This probability helps the lender to determine whether to accept or reject a customer’s application, as well as how to price a loan if granted.
There are many types of scoring models commonly used in the financial services industry. Some examples include:
Credit scoring models. There are different classes of credit scoring models. Generic models (such as the “FICO score” and other scores provided by the credit bureaus) make use of data reported by lenders to the credit bureaus. Based on a customer’s credit history, a score is calculated to predict the likelihood of a customer defaulting on a new account. Custom models typically make use of both credit bureau data and other application data (such as income, time at residence, etc.). Custom models are developed by a lender based on the performance of accounts in its own portfolio.
Behavior scoring models use credit and account performance data to determine whether to increase credit lines, re-price accounts, etc.
Collections scoring models utilize credit and account performance data to determine collections strategies. For example, how often to call a delinquent account and whether to sell an account or outsource to a collection agency.
Revenue scoring models are used to predict how long a customer will stay on the books, the amount of fee revenue an account will generate, etc.
Scoring models are also used in other industries and for other functions, such as insurance underwriting and marketing.