Reading Time: 4 minutes
In the previous article, we looked at credit risk, its underlying components, factors that gauge the creditworthiness in the context of the Buy Now Pay Later industry, and how technology can play a significant role in controlling credit risk. As a continuation, we shall look at the risk mitigation strategies based on ML and data analytics, along with the advantages of the adoption of automation to manage risks. This comes as no surprise. The key strength of the fintech space, with Buy Now Pay Later as one of the major sub-sets is the enhanced deployment of technology as a key enabler. We can expect to witness increased technology dependence in the coming days- across sectors, to drive efficiency and reduce risks.
Credit risk assessment models should be developed as broad-based as possible. In other words, several criteria should be considered before ascertaining the credit score. Rather than a mere Yes or No, wide customer profiles, past order history, similar target audience behaviour, transaction amounts and a range of values should be included rather than a restricted rule-based decision system. Further, it is imperative to exclude human bias or human errors that can cloud accurate decision-making.
According to Vesta, the following are the risk mitigation techniques based on data analytics and Machine Learning:
• Closely interconnected data points like poor CC bin ranges, invalid phone numbers, dummy email addresses, and multiple card details across merchants can help detect irregularities. It is vital that these data are kept secure in accordance with privacy laws. Data analytics can provide valuable insights on past purchasing trends and buying behaviour to identify exceptions.
• Link establishment: Often, a transaction consists of a series of steps that link IP, devices, email, and phone numbers to trace out an individual. Thus multiple names with common phone numbers or email IDs would indicate hidden linkages. Alternatively, several orders being booked from a common delivery address or multiple account sign-ins from a single device are red flags that can be easily detected through Machine Learning.
• Velocity of transactions: Several high-volume transactions being completed in a series sequence could indicate a card fraud where cybercriminals are using stolen card credentials to quickly transact before being caught. Alternatively, fraudsters could be creating multiple accounts with fictitious user details and using stolen card data. However, this needs to be evaluated on a case-to-case basis. It may also be that a genuine transaction may include a high volume of items in a single purchase or by use of multiple payment modes.
• Checks on the delivery address: Machine Learning can effectively detect any frauds being conducted by matching the shipping address. Say the delivery address of the party is an overseas entity, but the billing is done to a domestic party, this is a red flag.
The pivotal role of Artificial Intelligence in risk mitigation
Buy Now Pay Later players are increasingly recognizing the game-changing benefits from AI toward reducing risks. Both fraud risk and default risk can be reduced by studying data patterns across risk metrics, business performance, cash flows, past payment track records derived from varied data sources. This can help update the risk profile and accordingly impact the credit decision. For example, a customer with a higher credit score may be offered higher credit amount that may be used in online purchase transactions. Similarly, heightened risk factors for a particular sector would also impact the credit risks.
• In-house expertise and 3rd party support
Rather than opting for 3rd party credit checks, many fintechs are scaling up their in-house capabilities and building expertise in data analytics, Machine learning, risk modeling towards automation of credit decisions. However, technology upgrade involves considerable costs. Hence certain start-ups may prefer 3rd party assistance and prefer to focus on the core business.
• Real-time risk management
The risk profile of an individual or entity is often found to be ever-changing. Thus automation techniques that adopt a real-time strategy are effective in helping to decide matters pertaining to approving or declining credit as well as portfolio monitoring. These features may not be possible in a manual underwriting system or those with limited tech functionalities.
Advantages of automation in credit risk analysis
Automation brings a plethora of benefits for Buy Now Pay Later players to manage risks better. Some of the advantages are:
a. Changing Customer expectations: Customers expect fintech players to offer speed, simplicity, and convenience -in a single bundle. All of these are made possible by the deployment of technology. Tech tools can easily assess the credit profile of customers in a remote manner, enable prompt approval, and enable Pay Later facility at the time of checkout with just one click.
b. Reduced hassles for customers: Technology is the ultimate solution that can provide comprehensive end-to-end solutions, which can be quite tedious in a physical environment. Multiple processes from emailing, texting the OTP, scanning KYC documents, and maintenance of customer records, etc are seamlessly possible through technology. Further chatbots can interact with customers and answer their queries, along with a data trail and digital capture of all data.
c. Manual work made redundant: Technology results in huge savings in time, costs, and resources due to the elimination of manual work. By easily connecting the customer’s details on the merchant platform with the integrated Pay Later API, customers can conveniently complete transactions without the need to use cards or remember card details.
d. Enhanced speed and accuracy: The digital-first economy has enabled prompt decision-making and higher productivity. Machine Learning can easily identify or detect genuine and fraudulent transactions and drive quick credit approval or rejection.
A McKinsey study indicates that fintechs are able to leverage upon a 40% lower cost-to-income ratio in comparison to banks and other traditional players. The fundamental goal of every Buy Now Pay Later player is to mitigate credit risks. Automation technology like Machine Learning, AI, etc. can analyse data records, report outliers or errors, and recommend decisions for multiple customers at scale within a few seconds. For fintechs, along with accuracy, time is of the essence when it comes to credit decisions. High speed and high efficiency are by-products of technology. Automation works towards reducing costs of Pay Later companies to deliver improved returns.