Digitizing the final frontier in lending- Debt collection and NPA Resolution
Background
The earliest form of lending system originated some 3,000 years back in ancient Greece and Rome. In ancient Greece, unsecured loans were given in maritime trade with creditors keeping 25% of the profit in cases of success.
In India, loans were first mentioned in texts dating 2000 B.C. where usury (lending money at unreasonably high-interest rates) was considered to be a sin. As lending slowly became legitimized, an instrument called ‘adesha’ (modern-day bill of exchange) became popular.
As Britishers arrived in India, banking became more uniform and structured. The first bank of India- Union Bank of Calcutta was founded in 1806 which unfortunately became a storied failure. In 1885, Allahabad Bank was established which remains functional to date. As customers became acquainted with banks, the practice of keeping deposits, earning interest, and taking loans was inculcated.
Evolution
Fast forward to 2021, digital lending has become a reality and a preferred route for the vast majority of Indians for its speed and convenience. It has a market potential of $820b with ~50% of all lending transactions expected to become digital by 2023. There is also a mammoth $600b credit gap in MSME financing (being addressed by players such as Lendingkart, Aye Finance, Kreditbee among others).
The opportunity in the next 5 years is enormous as only 10% of Indians have access to organized credit and digital lending will enable financial inclusion for the vast majority. The UPI moment for lending is already here with the AA ecosystem that went live on Sep 21 which will massively expand the net for new to formal credit Indians as well as new use cases like daily repayment lines for micro-enterprises.
At the core, any lending business comprises three main pillars- origination & distribution, underwriting & operations, and collection & recovery.
The 2010–20 period has seen various startups that have digitized the first two pillars and ushered various types of models in lending. We now have P2P lending, point-of-sale (BNPL) financing, invoice-based lending, supply chain financing, digital mortgages, and crowdfunding among others.
In terms of business models, there are primarily three types that have emerged:
i) Lead Generation model- Businesses that connect with prospects through various marketing channels, capture data and serve as a funnel for lending players. It helps reduce CAC for lenders and banks share some fee income. Eg. of these include BankBazaar, PaisaBazaar
ii) Credit data models- Companies use AI/ ML, transaction, and alternate data to generate credit profiles and help FI’s measure creditworthiness. The pitch to banks is that that we will build better underwriting models, do credit scoring and you pay for it. E.g. include Perfios, CreditVidya which aim to build alternate credit scoring.
iii) Digital lenders- These include off-balance-sheet players where banks/ NBFCs only offer capital. End-to-end loan lifecycle is managed by the tech players. There is a clear value prop for banks since they get guaranteed returns and a rich digitally savvy customer base to cross-sell their other products while providing only access to capital. The second category in this is full stack lending platforms which have their own NBFC’s and control the entire process. These players can also opt for co-lending with traditional banks/ NBFCs. Eg. include Krazybee, Stashfin in consumer loans, Lendingkart in MSME loans, and Capital Float, Zest Money in the BNPL segment.
The interesting thing to see now is that the final pillar on collections and recovery is getting digitized. This piece looks at the market opportunity for digital debt resolution/ recovery platforms, why the timing looks right, interesting startups, challenges and risks, and what the future holds in the space.
Market sizing and Trends
The debt management and NPA resolution platforms encourage banks to outsource the collection piece to them. The pitch is to get banks to give their delinquent buckets to these players who aim to reduce NPAs through the use of tech or reduce chances of NPA through predictive models and streamline the collection and recovery process.
Traditionally this part has remained abysmal in tech adoption within banks/ NBFCs. Further, they have typically used collection agencies (tele-callers and field agents) to recover dues which have led to poor customer experience and often harassment from agents for the borrowers.
Globally the debt collection software market is expected to grow at a 10% CAGR from $3b in 2019 to $4.7b by 2024. Higher need to provide customer-centric debt collection solutions increased consumer preference for self-service models on the collection process, and the rise of specialized debt collection agencies will be the key growth drivers. In India, the opportunity is also massive as the BFSI sector spends $3B+ on just collections.
Why is the timing right?
Moratorium induced recovery rethink: In Mar 2020, RBI announced a 3-month moratorium on payment of installments of all term loans outstanding which was extended by another 3 months to 31st Aug 20. The pandemic had led to severe financial crisis and loss of livelihoods among a vast majority, especially in the informal sector. The step to tide over the Covid-19 crisis was also to ensure no risk-averse action from the banks that adversely impacted one’s credit score.
But the second-order impact of this was a dip in collection efficiency even for loans that did not qualify under this extension. In Sep-20 Ind-Ra also revised the banking sector outlook to ‘negative’ from ‘stable’ for H2FY21 due to expected spike in stressed assets and higher credit costs.
Use of AI/ML to collect better, faster humanely: There are three types of borrowers- first who unknowingly miss the payment dates, second who are unable to pay due to financial hardships, and last but the most important being who are not willing to pay. AI/ML is now being used to predict recovery chances from a delinquent borrower based on several data points (location, EMI payment history, similar borrower profile, etc.). Technology is also helping lenders become more empathetic to recover unpaid dues. Many digital lending startups found it harder to recover their monies from their target segment which primarily consisted of MSMEs, gig workers post the moratorium. Many were also publicly shamed by sending messages to their contact list and bullying from agents.
While all this has meant a rethink in terms of borrowers to lend to and underwriting mechanisms adopted by these players, a discussion has also started on how to make the collection and recovery process more borrower-friendly.
Stress in repayments: From a Captable report, while the credit demand is recovering, repayments remain a worry. ‘Straight flow’ which means non-repayment of EMIs in the first 3–4 months of a loan being issued worsened to 1.25% in early 2021 (vs. 0.9% in Oct-20 and 0.5% in Jan-19). For loans under INR 1 lacs, DPD 90 shot up to 1.3% in Mar-21 from 0.9% a year ago. In personal loan portfolio, this was worse at 4.8% in Mar-21 compared to 2.8% a year back.
The continued stress build-up on collection means that both banks and fintech will actively seek solutions to help them reduce their collection and recovery costs.
Bad bank gives me business: In Sep-21, the Govt. set up a ‘bad bank’ to pave the way for major clean-up of the banking system. With unpaid corporate loans at record highs, this would allow banks to remove massive NPA piles from their books, reduce provisioning pressure and clean up books for a healthier balance sheet. The detox is well-timed as the improved efficiency would allow banks to reduce their interest rates which can lead to increased credit demand creating a perfect recovery cycle right after the pandemic.
The massive pile of NPA’s can be more intelligently recovered by partnering with digital debt collection players. The ‘fresh start’ will provide a ripe window of opportunity for these new-age players to aggressively pitch to integrate into the LMS system of banks to efficiently track and take remedial measures against non-recovery of dues.
Increased digital collection behavior during lockdown to stick and persist: As the lockdown made it tougher for banks and NBFCs to reach their customers digitally, many of them increased their collection through digital channels. The use of AI-based bots to guide consumers for payments, single-click payment options helped Kotak Mahindra Bank to increase digital payment for the credit cards segment to 98%!. Muthoot Finance saw customers repaying online increase from 18% to 40% during the lockdown. As banks realize the massive efficiencies through this and bottom-line impact through shrinkage of recovery and collection staff, they will continue to incentivize borrowers to repay online. Digital debt recovery players can again reap the advantage of this behavioral shift among FI’s and increasingly enable borrowers to repay online easily and in a time-bound manner. According to a survey from Spocto, 75% of customers are ready to pay their dues digitally.
Clear value prop for banks: According to a Crisil report, the banking system NPAs will rise to 8–9% by FY22 end (0.5–1.5% higher than FY21 levels).
Debt collection software will make the banks nimbler. It will considerably reduce the time spent on manual and repetitive admin tasks and save on collection team costs. Intelligent use of data can also help banks uncover hidden insights, offer fair loans, cross-sell products based on borrower history.
Interesting models
Global companies
Overall, the pitch for the new age companies is to use behavioral analytics and adopt an empathetic lens to support consumer repayments that helps its clients increase their brand reputation and increase trust.
Prodigal Tech with its AI-powered Score model helps to maximize collections by identifying accounts most likely to pay. This helps in faster liquidation, an increase in net collection per agent, and higher account closures per month.
Receeve set up in 2018 is a fully customizable all-in-one collection and recovery platform designed to meet the need of enterprise clients. Its client have seen on an average 30% more funds collected over the first two weeks of starting a digital collections campaign.
Collect.AI provides a cloud-based AR solution that integrates with the accounting software of businesses. It offers interactive invoices with an integrated payment link, intelligent dunning (payment reminders for customers who have forgotten to pay their bills), and digitizes the entire customer payment journey.
Other companies include Indebted which raised a $25M Series B in Jul-21, Collectly which works in the patient revenue cycle space, and Symend which raised a $43M Series B in Feb-21. It generates deep customer insights using a trifecta of behavioral science, data science, and advanced analytics to empower consumers to resolve past due bills before they reach collections.
India
While the space is still nascent in India, it is rapidly heating up with interesting companies emerging garnering both client and investor love.
Companies like CredGenics aim to convert bad debt into good assets. It is devising various ways to help in better recovery through a mix of AI and data science. Early warning signals help lending institutions identify stressed accounts likely to default. ML speech-to-text models help to check the misbehavior of agents and allow FI’s to meet compliance requirements. Intent identification during calls helps to identify users unwilling to pay and take prompt action. Recovery chance predictor help banks target collection efforts and hence prioritize/ de-prioritize accounts for loan recovery. Through all this, it helps clients increase overall recovery rates, reduce legal workflows, time to resolution, and collection costs. The company charges a resolution-based success fee on the amount collected and a subscription charge for using the platform.
Another player Creditas set up in 2015 also automates the entire workflow to provide a one-stop solution for debt collection through its Ethera product. It cleanses and validates data from different skip tracing sources to ease and improve customer contact ability. ML algorithms engage customers through hyper-personalized content and nudge them towards payment. It works across varied products such as loans, cards, delinquencies, NPA, and write-offs.
CreditMate which pivoted from a vehicle financier to a collection SaaS platform in 2019 is another interesting player. Its Sherlock product (nice name!) scores debt defaulters, manages debt resolution process, and optimizes on results and costs. It also uses AI/ML to provide an ultra-smart and evolving intelligence layer for field collection agents. It claims to have improved collection rates by 15–20% for lenders.
Other players in the market include Spocto which offers a bunch of products like Smart Collect, Spocto Trace, Kisan Pay, restructuring and closure settlement bot, Moneytor, and CreditMantri.
Challenges and risks
Changing banks/ NBFCs mindset who have typically relied on outsourcing their collection part and building a provision buffer for projected NPAs could be tough and lead to higher CAC initially for the digital recovery players. Although pandemic-induced stress in recovery coupled with rapid digitization across different workflows has made banks/ NBFCs much more amenable to try out these solutions.
The second challenge is to work and recover dues from the bad apples who truly have an intent to default and the non-traceable ones. Success in these cases will require an innovative approach along with prioritizing the right accounts based on DPD buckets to improve RoI on collections for the clients.
The third challenge is resistance that could come from the client’s team for fears of being replaced/ downsized once collection software is used which greatly improves efficiency and reduces the need to have an in-house collection team.
Other challenges include limited tech budgets and resources for small, mid-tier banks; higher integration time and need to redesign operational practices, workflows, and controls in the collection, building a consistent and frictionless experience, and inertia on the compliance and regulatory side which slows the overall process.
However overall, the clear value proposition for clients in terms of increase in NPA resolution and late recoveries, reduced collection costs, higher contact ability for skip customers vastly outweigh any major challenges.
What is the future here?
The true game in lending is not of how quickly you disburse but how efficiently and promptly you collect. Debt collection has been a significant pain point for banks and NBFCs for long. Overall, the space is at the right confluence for rapid disruption and high growth. Setting up of the bad bank, with an increased focus to contain stress levels in banking books will provide strong tailwinds.
Financial institutions be it traditional banks/ NBFCs, digital lenders are recognizing the possibility to reduce collection costs and bad debts through the use of debt collection software that leverages AI/ML and data modeling. As the new-age companies become integrated into the software systems of financial institutions, they can vastly improve the entire lending process for all stakeholders. Digitizing the final pillar in the entire loan lifecycle will be a true win-win for the lenders and borrowers.
Further Reads: Increase in NPA in banks- comparative analysis
NPA and its effect on bank profitability
Banking system NPAs to rise in FY22
Image credits: Hindustan Times, Leadsquared, Market Business News, Wikipedia, Business Standard, iStock
The views and opinions expressed in this article are those of the author and do not necessarily reflect those of any institute or organization he is or was previously associated with.
