The mass adoption of technology by hundreds of millions of Indians is changing the face of the country’s lending sector.
In the pre-internet, pre-Aadhaar(I.e. India’s SSN) age, getting a loan meant waiting for several weeks for a credit decision. More often than not, the bank or financial institution that you dealt with used manual systems and procedures. Obtaining a loan was a hit-and-miss affair. It usually involved the credit officer’s judgement as well as your ability to convince the lender that the money would be repaid on time.
But, this is no longer true. Dozens of Fintechs that rely on online data to evaluate a loan application from an individual or a small business have sprung up.
These firms take a lending decision by relying on the wealth of data that is already available on the internet. All that you have to do is to fill up a simple online form. The lender will then access the information that it needs directly from the web.
New ways to appraise loan applications
India’s growing digital infrastructure has given Fintech the ability to change the way that loans are provided in the country. This illustration from a report by Kalaari Capital, a technology-focussed venture capital firm, demonstrates how mobile phones, the internet, and the country’s large population present an enormous opportunity for new tech-savvy firms:
Increasing digital penetration along with public-private initiatives are expanding India’s digital infrastructure
To get an idea about the potential that the Indian market holds for individual finance and SME finance, consider the following facts:
- The consumer credit to GDP ratio in India stands at a paltry 13%. In the U.S., it is 80% and in China, 40%.
- 17% of India’s households account for 90% of the country’s consumer credit. This implies that the untapped consumer finance market is a vast 250 million households.
- Of the 50 million SMEs in the country, only five million have access to any form of credit at all.
The digital revolution is going to allow the recently established breed of lenders to open up a new market that consists of hundreds of millions of Indians at the bottom of the pyramid.
How exactly will Aadhaar cards, the proliferation of smartphones, and the internet help financial institutions provide loans to people who are currently out of reach of the formal banking system?
A critical benefit of the digital infrastructure that is already in place is that loan approvals will now take much less time. According to the Kalaari Capital report, it is currently possible to onboard a loan customer in 20% of the time that traditional methods took. But speed is not the primary benefit.
The most significant advantage that digitalisation will provide is the possibility of cost reduction in the customer acquisition and servicing of individual loans. Conventional methods require a bank to spend about INR 2,500 for these processes. It takes approximately INR 1,000 to acquire a new customer and another INR 1,500 to service the loan.
That makes small loan transactions almost impossible. An average Indian middle-class family that earns INR 15,000 per month will be eligible for a loan of only INR 35,000. But an INR 2,500 acquisition/servicing cost that translates into 7.2% of the loan amount makes the transaction unviable.
Aadhaar-based verification and payments over the Unified Payments Interface (UPI) can bring costs down drastically. Many financial firms have already launched Aadhaar-based e-KYC platforms for loans. The process of verifying a customer has been transformed from a tedious and expensive exercise into one that is low-cost as well as almost instantaneous.
For example, HDFC Bank uses social media analytics to offer credit cards to its customers. Lendingkart, a financing company that provides loans to entrepreneurs and SMEs has done away with lengthy approval procedures. If your small business meets their eligibility criteria, you can get an INR 1 crore loan in just three days. The company already provides its services across 950+ cities across the country.
The SME sector accounts for 95% of all industrial units in the country and about 40% of the industrial output. But it is severely underbanked.
A report titled, FinTech in India – Ready for breakout, says that the credit gap in the Micro and Small Enterprise segment is INR 1196 thousand crore. That’s a US$176 billion dollar loan market.
Credit Gap in the Micro and Small Enterprise segment
How will this demand be met? Companies like Capital Float, a venture launched in 2013 by Stanford University graduate Sashank Rishyasringa in partnership with his classmate Gaurav Hinduja, are attempting to fill this gap.
A traditional bank loan can take four to six weeks or even more for approval. Private banks and traditional NBFCs are faster, but they too have approval procedures that could stretch into a week or two.
Capital Float promises to disburse a loan to an SME in 72 hours. How do they do this? The application process takes just ten minutes and involves filling a simple online form. Borrowers are required to upload the necessary documents digitally. Approval can follow in minutes.
Loan Frame, a fintech based in Gurgaon, provides secured and unsecured loans of values ranging from INR 1 lac to INR 50 crores to SMEs. The firm works as a marketplace that connects banks and other financial institutions with potential borrowers.
Machine learning and artificial intelligence play an essential role in compressing the time taken for approval. Loan Frame’s founder and CEO Shailesh Jacob says, “Currently, for unsecured business loans, we take one or two working days from the time of application completion to loan sanction. We want to move towards a turnaround time of a few hours ….”
What does the future hold?
Soon, the loan approval and disbursal processes in the country are likely to see a significant change. Greater numbers of banks and financial institutions will have to start taking advantage of India Stack, a set of APIs (Application programming interface) that allow governments, businesses, and startups to utilize the country’s digital infrastructure. The lending industry will have to change as lower costs and digital efficiencies bring bottom-of-the-pyramid borrowers within the formal economy.
Lenders have been using traditional data such as the CIBIL score, bank accounts, credit-debit card transactions, existing relationships etc. They will have to expand and build credit evaluation models that can leverage alternative data sources to appraise loans. Alternative data which includes, telecom bills, home office rent payment history, utility payments history, auto loan payment history, social media profiles, online shopping behaviour, internet browsing data (I.e. clickstream) , text messages, survey data, mobile app data, any other EMIs etc. will allow financial institutions to form an idea about a loan applicant’s ability to repay.
Machine Learning in Credit Risk Modeling
Per FICO research, alternative data sources do add predictive value on margin to credit risk models based on top of traditional data. The amount of predictive value outlined in the table below is viewed as relative indicators, and not in absolute value terms. The real advantage of using this data lies in combining the existing model’s predicting power and the effective usage of alternative data’s predictive value.
Machine learning techniques such as neural networks, random forests and stochastic gradient boosting plays a key role in building prediction models on alternative data sources. With large, unstructured data sets, the smart use of these technologies can identify data patterns that relate to credit risk and make the model development process more manageable.
In additional to building an exhaustive credit evaluation model, lenders will have to digitize the process completely to bring approval time from a few weeks to a few minutes if they really hope to capture any portion of 250 Million household/45 Million SME market in India.
There are tremendous opportunities for the lenders who can keep pace with changing technologies and digitalise their systems. But traditional financial institutions that cannot change fast enough and remain bogged down with their established procedures could face challenging times.
I will be writing follow up article that talks more in detail about how to build credit risk models using AI, machine learning and much more. Stay tuned …