The application scope of behavioral scorecards: ①Retail pooling, ②Credit card limit adjustment, ③Post-loan risk monitoring.

In the credit risk control scenario, there are three main score cards: A card (application score card), B card (behavior score card) and C card (collection score card). The scorecard measures the probability of default, overdue, loss of contact and other risks in the future in the form of scores.

Among them, the behavioral score card is used in the loan process, that is, the user performance link after the loan is released. Based on the user's performance information on the day of observation and in the past, it predicts their future overdue or default status; behavioral score card The main purpose of the card is risk control, adjusting risk strategies based on customer performance to achieve the goal of maximizing profits or minimizing costs.

Behavioral scorecards are generally used for long-term loan products such as mortgages, car loans, credit cards, and cash loans. Such products have higher interest rates. For short-term loans, you can consider using the application scorecard directly. For new accounts that take a long time to open, the application scorecard model is suitable. Behavioral scorecards are often used when the loan cycle is long and the account utilization rate is high.

Default monitoring and quota management are based on the lender’s performance after lending, predicting the probability of default in this period of time. Different from the application scorecard, the behavioral scorecard represents post-loan management. The application scorecard is used to calculate the pre-loan review stage to see its probability of rejection. The application score is only evaluated once.

Modeling requirements for behavioral scorecards

The modeling process is the process of finding patterns in a large amount of data, and data is the key factor for the success of the model. The preparation of modeling data is often based on a full understanding of the business and data. First, before preparing data, two key points need to be clarified: the data time window and the criteria for good and bad customers.

Secondly, since the behavioral scorecard focuses on loan account management, changes in limits and terms during the period of use, user early repayment behavior, etc., therefore, when modeling, it is necessary to pay more attention to user characteristics. For indicators such as distribution, overdue, aging, rollover rate, etc., the data used are usually repayment behavior, consumption behavior, etc.