Credit Scorecards
The assessment of default risk for the credit applicant is based on a decisional system which calculates the probability associated with a score that each applicant demonstrates to be a Good or Bad payer over a given time period (e.g. within a year from the score assignment).
A Scorecard is a statistical model that, by classifying the applicants into cells which are derived from the intersection of characteristics that have the greatest predictive power in identifying the good and bad payers, assigns a score to all the applicants classified in each cell. These characteristics can be selected from multiple data sources, which are available at the moment of assessment of the loan application. Information of the following types may be considered:
- Demographic
- Behavioural, in regards to pre-existing contracts and relations with the Credit Organisation
- Credit Bureaus (External data Bases)
- Guarantees
- Etc.
The use of this information for the development and implementation of these scorecards in automatic procedures for credit risk assessment entails that they be organised and updated in an appropriately structured Data Warehouse.
The development of the acceptance/refusal models involves carrying out statistical analyses on a sample of applications extracted from the Data Warehouse. The aim is to assign a score (weight) to each analysed characteristic. The total score for an applicant is given by the sum of the scores of each of the contributing characteristics of the Scorecard.
On the basis of the overall score obtained and the overall distribution of scores throughout the credit portfolio, the bank can decide to refuse all applications with a score lower than a pre-determined threshold or make economic conditions less attractive for these applications to protect themselves from the higher risk that they imply.
The use of this tool - the risk score - therefore allows us to minimise losses for a given acceptance rate.