To argue that present debt collection techniques have an image problem is an understatement of their negative connotation. Threatening, harassing phone calls, coercion for money, or, in more serious circumstances, fraud, hounding, and social humiliation If a borrower fails on a loan, this is all part of the deal.
A short glance at the Reserve Bank of India’s (RBI) issues with debt collection companies between 2008 and 2021 reveals that little has changed – uncivilized, unlawful, and questionable behaviour and recovery techniques continue to dominate the roost. Except that things are only getting worse. Add a layer or 100 of extra information that lenders (and debt collection organisations) have access to, and we have a data privacy nightmare on our hands.
The RBI thinks that the regulated firm bears the ultimate responsibility for borrower safety, both physical and digital. In the recent past, there have been multiple stories of recovery agents harassing not only the borrowers but also their friends and relatives utilising borrowers’ phone contacts, images, or any other private material as a form of extortion.
But how did we end up here? In the absence of an on-the-ground debt collection crew, some lenders abuse signed agreements to gain access to borrowers’ mobile phone data and contacts in order to intimidate them into repayment.
Because of the lack of on-field collections teams and the total shift to digital collections, there is little segmentation – most collection models are based on restricted data points that do not reflect changing economic situations. These models cannot identify which loan accounts are likely to default or which of those that have defaulted can still be rescued.
The Justification for a Hybrid Collections Model
Proactively identifying the vulnerable is at the heart of an orderly, intelligent collection methodology. It’s worth noting that collection departments are now charged with dealing with borrowers who aren’t defaulting for the normal reasons – they may have been furloughed with a rapid, considerable drop in income, and credit bureaus have yet to discover these sudden changes in circumstances.
The household debt-to-GDP ratio in India increased to 37.3 % in 2020-21 from 32.5 % in 2019-20. This percentage is likely to climb further in 2021-’22 as households bear the burden of medical expenses incurred during the second wave of Covid-19.
Most lenders are proactive when it comes to pre-delinquency; nonetheless, there is a real need to go deep across the company to capture data and locate information that’s critical to distinguishing economic victims from steady state collections customers.
In its most recent assessment on digital lending, the RBI advises a hybrid collection model. Having an on-the-ground collections staff and an appropriately sized call centre would help the lender understand the difficulties that clients face in repaying. It also makes logic.
With intra-day, reactive changes in a COVID world, there are more people in arrears who were not expected to be there, and the data you had four weeks ago is now out of date – it’s critical to understand which customers are impacted by COVID and likely to bounce back, and which would be in financial trouble regardless of the crisis.
Although this is time-consuming, it is also one of the only ways to obtain data points such as:
- Unemployment, furlough/reduced income, medical, quarantined are all reasons for repayment failure.
- Payment holiday, lower instalment, repayment length adjusted – what relief did the customer receive and when did it end.
Making Use of Hybrid Data
The history of Artificial Intelligence (AI) and Machine Learning (ML) models has led us to believe that they can function without human intervention. Many people were taken aback when Google defended its use of human resources to improve its understanding of voice chats for Google Assistant.
AI and machine learning models efficiently filter data at breakneck speed, revealing insights about delinquency risk and how to handle at-risk accounts.
Collections go beyond simply reminding consumers to settle past-due installments; it also suggests a way out of an impending crisis, which is where AI comes in – workflow automation and borrower segmentation will create process efficiency and free up resources for more value-added services.
To reduce defaults, new era debt collection models incorporate early warning for delinquency, improved borrower categorization, and optimised customer engagement strategies.
System of Early Warning
Debt collection has historically been reactive; loss recovery after delinquency was the norm. This paradigm is altered by ML. Features can be developed where it assists in identifying borrowers who may default by mining previously unknown insights. It helps identify problematic borrowers by aggregating data points such as a low balance, multiple credit obligations, future loan dues, increased credit utilisation, and a poor financial inclusion score.
For lenders, this identification serves as an early warning mechanism. This system can also have features that also alerts hazardous debtors five days before their due dates, allowing lenders to proactively allocate collections resources and use targeted communication.
Lenders can create a detailed customer profile to determine which borrowers are likely to address delinquencies on their own and which borrowers require assistance (loan restructuring, modified repayment terms etc). Real-time cash flow data, such as account balance, credit appetite, over indebtedness, and so on, provides lenders with accurate insight into the borrower’s financial status following disbursement.
It develops priority buckets for each borrower, allowing for targeted communication as opposed to the one-size-fits-all strategy, which results in an unquantifiable decline in loyalty. It has been discovered that categorising borrowers based on risk helps lenders reduce collection costs by prioritising resources. A smooth collecting process also promotes client satisfaction. This means they are more likely to apply for credit from the lender again.
Emails, text messages, social media, mobile apps, and online chat bots are all ways to contact your borrower. But the issue isn’t the availability of platforms; it’s understanding which platform to use, when to reach out, and how to write an effective message.
Most lenders fail to recognise that these engagement aspects are all contextual and depend on various factors. Lenders, for example, can create a personalised outreach strategy based on their mobile app, identifying a preferred manner and time of contact that can be combined with demographic/financial information. Customer call audio can also be used to analyse how alternative scripts and offers affect customer reaction and collection.
The foundation of artificial intelligence has always been human intellect. Although it may be much better than other methods at predicting criminal behaviour, it still has a lot to learn from the fundamentally fallible perspectives that humans have. Doing the difficult work yourself and combining it with AI will ultimately result in a deeper understanding of your customers as well as the capacity to identify borrowers and engage in intelligent conversation with them.
Switching from a reactive to a more proactive approach to consumer engagement can help reduce delinquencies, borrower penalties, credit markdowns, and even insolvency when it comes to collections.
Ignosis provides Real-Time Risk Assessment and 360° Loan Monitoring powered by Machine Learning. It’s also your best bet to alternative data and open banking. You also get Plug-n-play modules for transaction analytics, categorization, FIU & FIP gateways to jump start your AA adoption. Do get in touch with us as soon as possible if you are a lender or a loan service provider interested in enhancing your existing loan origination system by incorporating an account aggregator.