Managing Credit Risks In The Buy Now Pay Later Ecosystem – Part 1

Reading Time: 4 minutes

Credit risk occurs due to the default of a party, i.e. the risk that the party would not meet a financial obligation when due or fail to repay the amount payable. Since digital payments are monetary transactions that enable the purchase of goods or services, the credit risk is an ever-present risk that every digital payments player seeks to mitigate. The mitigation strategies vary- from the deployment of cutting-edge technology tools like Machine Learning, AI, and predictive Data science models that study past patterns, building a strong underwriting team to enforcing robust risk management practices. In Buy Now Pay Later payment models that function as a form of credit at the time of check out, the need for effective credit risk control is of paramount importance.

Understanding credit risks

Broadly, credit risks have two underlying components.

1. Risk Exposure

This pertains to the amount or value of the transactions, which is equivalent to the credit sanctioned to the customer and the amount reimbursed to the merchant.  The risk is the loss that would need to be sustained by the BNPL player in case the customer fails to repay the amount due. Other risks include the risk of the merchant or the customer becoming insolvent, which could snowball into a liquidity problem for the BNPL player.  


2. Likelihood of financial failure

Credit risk analysis makes use of multiple tools and models to determine the probability of financial failure of the merchant or the customer. These range from detailed financial analysis, past credit history,

and ascertaining an internal credit score based on user behaviour and purchase patterns. 


3. Fraud Risk

In any lending business, be it secured or unsecured credit, the risk of fraud is present. The key is to do thorough due diligence on the credit profile of the customer and partner with reputed merchants and well-known D2C brands. False orders and fictitious customers without proper KYC can pose a huge financial risk for BNPL players. 


Parameters that determine the credit standing

•  Intent to pay

Ultimately credit risk depends on the creditworthiness of the customer. The BNPL player needs to be assured of the integrity of the customers and their intent to repay the borrowed amount. Defaults or NPAs are a huge drain on the finances of BNPL players and proper screening is vital to mitigate this risk. Several factors namely the background, income group, buying behaviour on platforms, repayment track record, and recommendation by the merchant play a crucial role in the credit decision process. 


• Repayment capability

This refers to the income level or cash flow position of the customer, which would indicate the credit servicing capacity. Though BNPL credit amounts are generally limited with an upper cap, studying the payment history of past borrowings and the ability to repay loans would translate into deciding upon a specific borrowing limit for each customer.


• Credit checks

Often it might be that the past credit history is positive, but the customer still defaults due to her/his current financial condition. Simpl offers a convenient Pay in 3 option for customers to split their bills into 3 easily manageable amounts. A soft credit check would consider the present industry conditions and the economic scenario without enquiries into the CIBIL score, prior to enabling credit purchases. 


• Merchant recommendation

The recommendation by the merchant to provide BNPL to their loyal and valued customers would play a critical role in determining the borrowing amount for purchases at the time of checkout. While not in the role of a guarantor or collateral provider, merchants have access to deep insights about buyer purchasing patterns and transaction amounts. 


Technology as the panacea to managing credit risk

At Simpl, we accord the highest priority to technology as an enabler of prudent credit decisions. All our credit decisions are 100% handled by machine intelligence, thus eliminating any scope of human error or bias. Our underwriting process has been built by over 24 months of intense R&D and in-depth study of best practices in risk management, powered by Machine Intelligence. We process 1000+ features, including user behaviour on merchant platforms, the behaviour of similar users in the past, signals derived from app installations, and then run all the data through an ensemble model, a unique combination of a decision tree, gradient boosting, Bernoulli Naïve Bayes classifiers and logistic regression analysis. 


We have ensured a key focus on anti-fraud and security measures. Towards this end, we have conducted experiments with repurposing time series prediction models and included our observation to efficiently detect suspicious transactions, identify exceptions and outliers.  

Besides merchant integration data and proprietary data, we have taken learnings from fraud events to build better risk and underwriting models.


According to Vesta, the following digital footprint tracking mechanisms are some of the commonly used methods to mitigate risks:

• Detection of Anomalies: Sudden changes in buying patterns or irregular ordering habits like placing orders during uncommon timings or abnormally high order size of multiple small-value items etc indicate possible default risk or exposing the nature of the buyer with no real intention to repay.

• Device tracking: It is useful to keep tabs on the devices used by the buyer to log in. By deployment of machine learning algorithms, it is possible to find data points like browsing time, time zone, language settings, type of device, etc. Any deviation from the normal, say, for example, multiple attempts to log in or a user is suddenly found logging in from a remote location or a location that is miles away from the usual residence would help red flag unauthorized access. 

• IP profiling: A user account that fails to provide geo-location or masks the location details or if there is a variance between the billing address and the IP location serves as early warning signals.

• Profiling of Email: The email address is commonly used by customers to log into the platform and place an order or as a double security measure, the OTP is sent to the mobile and the email. This indirectly confirms the legitimacy of the user. Email addresses with spam names or no names or just initials or those constructed randomly with the intention to make detection of the identity difficult are obvious signals to fraudulent conduct.

• Monitoring the phone number: Multiple phone numbers resulting from a single device or user account is a prominent red flag. 


Concluding thoughts

The use of automation tools to control credit risk and prevent fraud activity brings a plethora of advantages like accelerated, accurate decision-making, access to a vast pool of consumer data, speed of transactions, and reduced costs while maximizing ROI.

In the next article, we shall explore some of the other tech tools in the playbook that BNPL players are leveraging to manage credit risks. 

Leave a Reply

Your email address will not be published. Required fields are marked *