2. Credit Data and Processing

Credit data refer to any data related to a credit product during its lifetime. Credit data vary with different products, for example, retail and wholesale loans have different types of credit data. Credit data also depends on availability by aggregation levels. Account-level data have more granular information for each account compared to aggregated cohort-level data. In the time horizon, credit data can be classified as origination data and transaction data. While the origination data describe all characteristics at the credit product origination, the transaction data record the periodic (e.g., daily, weekly, monthly) status changes of the credit product. So, the origination data are static, and the transaction data are dynamic. In the content horizon, origination data includes feathers describing borrower/obligor characteristics, product characteristics, and collateral characteristics if the product is secured by some underlying asset. For transaction data, it could consist of the full transaction history since origination, or transactions from a booking date, or just a snapshot of the booking at some specific dates. Figure 2.1 shows the structures of credit data by these different dimensions.