How India’s NBFC Startups Are Using Deeptech To Change The SME Lending Game

The NBFC business model has been considered broken by many experts since the beginning. NBFCs rely on short-term funds which are lent out for longer terms to small and medium enterprises (SMEs) which brings in risk of asset-liability mismatch. This is said to be one of the major reasons behind the recent challenges in the state of NBFC liquidity after the IL&FS default crisis in October 2018. As their operations are catered towards unorganised and under-served segments, the risk of NPAs is also said to be much higher for NBFCs than banks and MFIs.

Yet, according to a May 2019 IBEF report, the public deposit of NBFCs increased from $293.78 Mn in FY09 to $4.95 Bn (INR 319.05 Bn) in FY18, registering a compound annual growth rate (CAGR) of 36.86%. According to RBI data, NBFCs loans and advances stood at INR 19,842 Bn at the quarter ending September 2018.

A large part of this growth has been contributed by the new-age SME lending startups. These startups are taking the traditional NBFC processes to an entirely new direction with the use of deeptech like artificial intelligence (AI), machine learning (ML) and data analytics. These are not just buzzwords when it comes to the fintech sector any more. The actual application of AI and ML have yielded new models for fintech companies. And this, in turn, has made borrowing much easier for the MSME sector.

Sanjay Sharma, cofounder and managing director, Aye Finance believes that most banks and financial companies had assumed the segment would be unprofitable and too risky, due to lack of formal business documents and small ticket sizes.

“However, SME lending startups are now solving these challenges of funding the micro-segment and enabling their inclusion into the mainstream of the economy with technology playing a crucial part,” he added.

Another startup NeoGrowth has a system in place where 85% of the loans are daily repayment products and 45% of loans have repayments from exclusive terminals which mitigates risk of default. “This model ensures a high collection efficiency of over 98%,” claimed Piyush Khaitan, Founder and managing director, NeoGrowth.

The players such as Aye Finance, NeoGrowth, LendingKart among others are not only reducing NPAs consistently but are also able to raise funds amid the ongoing NBFC crisis, indicating a greater investor trust.

NBFC Startups: Using Technology To Gain An Edge In This Tight Market

Pushing aside the inherited challenges from the business model perspective, NBFCs are today dealing with stiff competition from incumbents and the entry of fintech players. Also, dynamic regulations are continuously increasing the cost to comply and restricting the ability to freely impose pricing. Add to this the rapidly increasing expectations of customers and economic volatility, which are further hurdles in the growth for these startups.

To tackle these challenges, startups are utilising deeptech to create an edge for themselves in three key areas:

Credit Assessment And Underwriting Process

In order to ensure that the asset-liability cycle does not break, the NBFC startups need to bring in an efficient credit assessment algorithm right at the beginning. For instance, Aye Finance has designed its underwriting methodology based on the combination of industry-cluster approach, behavioral scorecards as well as alternate data underwriting methods like psychometric analysis.

“These innovative methods have allowed us to enable the financial inclusion of over 1,25,000 micro-enterprises,” added Sharma.

And that’s just one way of doing it. Mumbai-based ePayLater uses real-time microservices to gather data in real-time leveraging data science and algorithms, NeoGrowth uses digital payments data (notably card sales) while Lendingkart extracts up to 8,500 data points to assess intent and ability of the customer using machine learning and other deeptech models.

Further, the SME lending companies have at their disposal data sources such as bank statements, VAT/GST returns, Truecaller data, Justdial data, PAN/TIN/VAT/Address, ROC and MCA filings, Defaulters list; CA/CS membership; Facebook, LinkedIn, open searches for social profile verification, as well as private services such as Zauba, Probe42 and Experian Hunter, along with Aadhaar-based e-KYC among others to make a ‘leakproof’ credit assessment system at the core of their service.

Not only this, these deeptech algorithms further help in reducing the credit approval timelines for immediate needs. Akshat Saxena, cofounder, ePayLater highlights how its proprietary deeptech algorithm takes into account alternate data to provide approval within seconds.

“The customer can start using the credit line within hours of approval. The approval rate is also higher than traditional banks due to leveraging of alternate data,” Saxena said.

Automated Collection Process

Having the right collection process is as important as bringing a creditor onboard. While Aye finance uses cloud-based business process engine and variety of data models to maintain economies of scale of its small ticket size loans, ePayLater uses machine learning to streamline its collection process.

Another SME lending startup RupeeCircle uses its own RC score, a proprietary underwriting model to maintain its loss rate in check at below 2%. “We have also created a module that helps deploy investor money automatically based on the profile and risk criteria selected by the investor,” added Ajit Kumar, founder and CEO, RupeeCircle.

Interestingly, NeoGrowth has developed APIs into other permission-based rich data sources like the credit bureaus, GST data, merchant acquiring databases etc. Not only its underwriting or daily data insights model but even the repayment of NeoGrowth loans is based on a daily repayment model, where the amount repayable to them is remitted directly by the card acquiring bank to NeoGrowth.

“This is only possible because of our digital interfaces with the banks, and advance suite, our proprietary loan origination system (LOS) and loan management system (LMS) which is purpose-built for NeoCash, our unique loan product. Advance suite also directly interfaces with our data warehouse, which is a critical repository of all data sources and serves as a ’single source of truth’ for all management reporting and ‘deep analytics’,” added Khaitan.

Making Customer Comfortable With Technology

Despite having more than 627 Mn internet users, there is a large segment of Indian SMEs which is not comfortable with transacting online. However, Lendingkart founder Harshvardhan Lunia takes pride in the platform’s ability to acquire and service MSME customers at the remotest places in India using data. Startups are choosing other ways to bridge the human-digital gap. Aye Finance and Neogrowth are working with an “assisted fintech approach” thereby offering a suite of digital assets to support the Loan products.

As Sharma explains, Aye Finance’s 2000+ field staff are equipped with an Android app for loan workflow that works on the cloud infrastructure facilitating a low-cost operation that helps the inclusion of a larger number of grassroots businesses into the folds of organised lending.

On the contrary, RupeeCircle claims to be moving towards fully automated and digitised operations with minimal human intervention. The startup is also relying on certain third-party integrations to make its processes quick and user-friendly. Voicing the same strategy, Alok Mittal, cofounder and CEO of Indifi Technologies, a platform that helps lenders and NBFCs provide loans to SMEs said that the use of technology and data has allowed SME lending companies to automate a lot of steps through the customer journey from the time they land on their platform.

“For instance, at Indifi, things like auto bank statement analysis, auto decisioning, online data integrations etc have led to minimal manual intervention and quicker decision making, eventually leading to better customer experience,” he added.

SME Lending: AI/ ML Is The Core But Blockchain And NLP Is The Future

Almost all the SME lending startups Inc42 spoke to agree that to expand the access to credit to a larger population of underserved micro-enterprises, the NBFCs have to now start working on a variety of use cases for AI/ML which are currently used primarily for fraud detection, underwriting and backend operations. In future, the NBFCs may look at other deeptech use cases ranging from designing ML algorithms to automating credit approvals, image processing capabilities, etc.

“These deeptech models once deployed will bring improved efficiencies in our existing processes bringing a larger base of the underserved micro-enterprise sector into the folds of the formal economy,” said Aye Finance’s Sanjay.

As NeoGrowth’s Khaitan aptly said it’s time to leverage technology in smart ways for the future sustainability and growth of the business.

“These deeptech models become smarter as we feed them with more and more data. Thus, helping us create a bespoke offering in terms of the loan amount, tenure and pricing for various kinds of customers.”

With digital banking, the personal connection between banks and customers was lost to a certain degree. With AI and ML at the core, blockchain and natural language processing can be used to respark this connection in new and meaningful ways to become a game-changer.