Credit Score
- Alphonse Gaju Ndikubwayo
- Jean Luc Ugirashebuja
The Credit Score Model serves as a tool for assessing creditworthiness and facilitating data-driven decision-making. Developed to enhance risk management practices, the model leverages key features from historical data to generate credit scores that aid in categorizing customers based on their credit risk. The objective is to create a risk assessment scale(score range) that summarizes each customer’s risk profile.
This allows the user to see the credit score.
Any user who can view customer and account pages can view the credit score too.
Business Objectives
The primary objectives of the Credit Score Model are aligned with Bboxx’s broader strategic goals, emphasizing the enhancement of credit-related decision-making processes.
The model is designed to:
Improve Portfolio Management
Import Operational Efficiency
Encourage data-driven decision-making
Acceptance Criteria
Credit Score range is from 0 to 999, cannot be negative or above 999. Below are the color-coding bands (from Blue to Green) like gradient → with changing shades. 0 is for blue 999 is for green
0 → 500
500 → 700
700 → 800
800 → 999
New customers (less than 3 months) will not have a credit score.
We will have the message “Credit Score is not available”.
Advice messages for adverse features
current_utilisation_rate: Show long-term payment consistency
last_3_months_utilisation_rate: Show short-term payment consistency
std_payment_size: Keep payment amounts consistent
total_paid: Reduce your overall debt
days_since_last_missed_payment: Never miss payments
last_12_months_par30_times: Avoid 30 consecutive late days default mark
last_12_months_par60_times: Avoid 60 consecutive late days default mark
How to view the customer's account performance.
Display the credit score on:
On the customer page
· Same above but on the customer page
On the account page
· Under customer details, it's basically under the customer names and phone card (Should be visible without needing to scroll down)
The same credit score should be displayed on both the customer and account page because there is no difference (a credit score is an object of the customer)
The same credit score should be displayed on both the customer and account page because there is no difference (a credit score is an object of the customer)
Search for the customer you need in the Customer module :
o Click the 9 dots → Customer → search.
· The credit score is displayed on the customer page
Model Overview
The model considers 8 features during the credit scoring process. Those features are selected to ensure a comprehensive risk profiling:
Customer Age
Last 3 months Utilisation Rate
Total Utilization Rate
Total Payments
Std Payment size
Times Reached PAR30
Times Reached PAR60
Days since the last missed payment.
Customer’s Input Data
To obtain a fresh score for a customer, the above 8 features need to be collected in their raw format. Subsequently, each feature is assigned to its respective bucket. The bucket design ensures that any value of any feature automatically falls into one of them.
Scoring
Finally, the summary of the points associated with those buckets is the final credit score of the customer.
FAQ
Why the Credit Score of customer X is low while their Collection Rate/Utilisation Rate/Other is high?
To grasp this keep in mind that Credit Score is a risk profiling tool while Collection Rate/Utilisation Rate/Other are business metrics. As can be depicted by the scorecard, there is a specific type of customer that can cause a noticeable difference between those two metrics. (apart from total_ur and last_3months_ur) but at the same time, they can have a high Collection Rate/UR/Etc.
How often are the scores refreshed?
Monthly, but this can be configurable per OpCo.
What is a good score?
No such thing as a good score. For each business case, an analysis needs to be conducted to find such cutoffs.
BBOXX