Determining credit card eligibility is a complex and sensitive decision that can directly affect the organization's economic situation. The challenge for the institution is if it is a manual decision, it is time-consuming, error-prone, and leads to inconsistencies. If it is driven by Machine Learning and AI technology, it can be biased; explainability is hard and insufficient with ML and not acceptable to regulators with it comes to compliance.

The implication for organizations to get it wrong is massive and has several negative impacts, including and not limited to financial losses, reputation damage, missed opportunities, customer churn, and compliance fines.

In this blog post, we will suggest an automated credit card eligibility model and how it can help financial institutions to overcome the challenges they face in the industry.
By implementing automating credit card eligibility decisions, organizations can offer the following benefits to financial organizations:

  • Evaluate the creditworthiness of applicants more accurately.
  • Manage their reputational and compliance risks more effectively.
  • Increase consistency and visibility to the clients and the other stockholders.
  • Increasing the operational capacity by shortening the required time to make decisions.
  • Quicker response to changes such as adjusting product terms, rating, and risk categorization. Or introducing new products by responding to market dynamics based on customer demands.

The scenario of the credit card eligibility decisions

In general, the eligibility for the credit card decision requires multiple inputs. For instance, creating a decision flow that checks the eligibility requires applicants’ details like account type, residential status, annual income and etc. Based on these data, at the high-level there are two main critical decisions:

  1. Determine the Applicant's Demographic Suitability
  2. Determine the Applicant’s Credit Card Eligibility

Determine Applicant Demographic Suitability

The applicant's demographic suitability is evaluated upon submitting their application to determine credit card eligibility. This evaluation determines whether their Credit File from an external Credit Reference Agency is worth handling.
Automated Credit Card Eligibility
In this Decision Graph, two primary outcomes are defined by the Determine Demographic Suitability Decision Node: ‘ Suitable’ and ‘Unsuitable.’ To determine either of those outcomes, this Decision Graph uses two other decision nodes: Decision Tables (as shown) and Applicant input data.
We have used a boxed expression to derive the Applicant’s years of age from the date of birth in the applicant’s input data.

Book a Custom Demo

First or last name is too short





Determine the Applicant’s Credit Card Eligibility

Once the decision determines that the applicant is suitable, their Credit File is processed and assessed based on specific criteria for the credit card they have applied for.

Automated Credit Card Eligibility2

This Decision Graph uses the Applicant’s Credit File, Application History, and Applicant data as input data to determine overall credit card eligibility. The Applicant’s Credit File is obtained from the external credit agency. Those data are used in the Credit Score Decision Table aligned with the knowledge source, Lender Credit Policy.
As we have an expression to derive the applicant’s age in the previous Decision Graph, we have another Boxed expression associated with the decision nodes to calculate the applicant’s address history and to check whether the applicant has previous credit card applications in the last six months.
Based on the Credit Score and Balance transfer eligibility, the final credit card eligibility is determined as ‘Eligible’ or ‘Ineligible.’

Determining credit card eligibility model

The following flow integrates the two critical decisions, which are demographic suitability and credit card eligibility, to determine the overall credit card eligibility of the applicant. Based on the decision outcomes, the model creates a notification as an output stating that the application is approved or rejected.

In this scenario, we use rule-based decisioning, which increases the level of transparency and expandability, particularly regarding audit and compliance. Additionally, rule-based decisioning in this case, enables ease of adjusting criteria based on different markets and products' requirements even without having data to train a new ML model.

Conclusion

Though there are many different approaches to automating a credit card eligibility decision using ML/AI, code-driven systems, and other out-of-the-box solutions, however, by using the advanced decision management and automation platform and applying rule-based decisioning technique, you can empower organization for greater flexibility and agility and sets them up for a better audit, compliance and other strict regulatory requirements.

As you see in this example, the decisions are relatively independent and encapsulate the complexity of the rules; however, they are connected in a decision flow to implement the entire scenario. The decision graph enables managing the interconnection and dependencies of complex decisions by allowing you to decompose a complex decision into smaller decision units. The decision flow, on the other hand, allows you to connect the decision graphs to build your customized business scenario.

Last updated April 20th, 2023 at 04:25 pm, Published April 19th, 2023 at 04:25 pm