Welcome to the 93rd edition of The Pattern, a weekly newsletter where we discuss and dissect rumblings from finance, technology, and the economy. Let us get started.  

 While the last week was about making predictions about what is to come in the lending industry, this week we have more than just a faint signal.   

For RBI’s (Reserve Bank of India) top brass has been active on public platforms and advocating more governance across many areas. One of these is algorithm-based lending where the regulator wants more oversight. What does that mean for the industry? Let us find out.  

The rhythm of lending algorithms   

If one were to list the most exciting recent strides in the credit world, automated decisioning would rank high up. While digitalisation made the process of availing financial products easier for borrowers, the real magic has happened in the background as lenders adopted big data and machine learning-based decisioning algorithms that help underwrite a customer precisely in a matter of seconds.  

It is not just a feat of technology but an amalgamation of miraculous progress in data science, credit risk, data processing, as well as modeling and technology infrastructure.  The result?

 A borrower with 10,000 pages of financial data can be assessed thoroughly and approved for a loan in a minute or so. At the same time, another borrower with barely any formal credit history can also be assessed based on their transactions/behavioural data and approved for formal entry into the credit ecosystem through a small personal loan.  

 However, the regulator  isn’t fully convinced yet . In a latest address RBI Governor Shaktikanta Das said lending decisions based on algorithms could “lead to a potential crisis” emphasising that NBFCs (Non Banking Financial Companies) and banks using these algorithms must “appraise robustness of models used for lending.” 

 This comes only weeks after RBI raised risk-weights for personal and unsecured lending – thus  hoping to put brakes  on the sharp acceleration of credit products often fueled through algorithm-based decision–making engines.  

“It was very clear to us that this kind of growth would not be sustainable going forward if it is not slightly moderated,” the governor said. “We clearly anticipated some problems ahead of us down the road. Therefore, we acted preemptively,” Das said.  

Black box through the looking glass   

The regulator’s discomfort with AI–based decisioning is not about technology use – in fact, the RBI promotes innovative technologies to improve the efficiency and effectiveness of financial services. The discomfort is more cautionary in nature and stems from the potential of invisible algorithms making decisions at scale that might be irreversible in nature and cause real-world damage.  

 “These advances, however, carry potential risks arising out of embedded biases in AI (Artificial Intelligence)/ ML (Machine Learning) systems and the lack of transparency in their outcomes. Furthermore, the technology potentially embodies new sources and transmission channels of systemic risks capable of undermining financial stability. It is, therefore, imperative to strike a balance between benefits and risks by strengthening the capacity of REs and surveillance by oversight authorities, formulating/updating relevant legal and regulatory frameworks, proactively engaging stakeholders to identify risks, and expanding consumer education,” stated RBI’s recent trends and progress in banking report 2023.   

Hence, the regulator would ideally like to scrutinize the models and ensure their robustness in terms of fairness, equity, auditability while ensuring they are non-discriminatory and bias-free.  

This is what the RBI prescribed in the above-quoted report:   

“The Reserve Bank has emphasised that the algorithm used for underwriting should be based on extensive, accurate and diverse data to rule out any prejudices. Algorithm should be auditable to point out minimum underwriting standards and potential discrimination factors,” the report stated.   

It added that lenders should adopt ethical AI (Artificial Intelligence) which focuses on “protecting customer interest, promotes transparency, inclusion, impartiality, responsibility, reliability, security, and privacy.” 

  Prediction vs caution  

 In effect, what this means is lenders might be up for additional scrutiny from the regulator, but they should mostly emerge unscathed if their decision algorithms work towards financial inclusion and not against it.  

It is par for the course for any new technological advancement to be scrutinized by stakeholders, but we have seen the unfettered support for UPI (Unified Payments Interface), Account Aggregator, digital lending, as well as digital currency among many other advancements.   

India’s banking regulator is adept at balancing innovation and larger good, it is time that the rest of the ecosystem picks up this skill too.  

This is all for this week. As always, I'll leave some interesting data points and reading recommendations below.  

Between the digits    

Rs 3,529 crore:  Life Insurance Corporation of India has  received  tax demand notices of Rs 3,529 crore from the income tax authorities. The state-owned insurer will appeal against the notices, it said in a filing.  

$55 billion:  FDI inflows in India could rise to their pre-COVID  peak of $55 billion  in the next two years, said HSBC Research. The bet is on futuristic sectors such as EVs (Electric Vehicle), renewables, data centers etc. to drive this investor interest.  

Rs. 4,200 crores:  YES Bank is selling an NPA pile worth Rs 4,200 crore and inviting bidders willing to front cash to acquire these loans off the banks’ books. Among the  bidders  are JC Flowers ARC, Acre, Edelweiss and five others.  

Reading list   

1. Cobranding with Tata Neu, Swiggy widens HDFC Bank cards’ reach

    

2. Leading fintech firms face a profit puzzle

    

3. How FinBox BankConnect powers advanced income analytics

    

4. Take pre-emptive steps to mitigate risks: RBI to banks

    

5. The role of a business rules engine in India’s mortgage landscape

    

Thank you for reading. If you liked this edition, forward it to your friends, peers, and colleagues. You can also connect with me on Twitter  here  and follow  FinBox on LinkedIn  to always get all updates.  

Cheers, Mayank

All opinions expressed are my own and do not necessarily reflect the views of FinBox or its promoters