©2019 by Caspian. All Rights Reserved.

Fintech Lending – Challenges and Lessons

August 7, 2018

 

I have to admit that we (Caspian) have been a little late to the fintech party and haven’t really done a lot of investing/lending in the sector. This is in contrast to the past where, Caspian played an active and pioneering role in sectors like microfinance, affordable housing finance, and affordable school finance segments. We were the first or early institutional investors in pioneers like Ujjivan, Janalakshmi, Equitas, Micro Housing Finance Corporation, Aptus Value Housing Finance, India School Finance Company, etc. A significant reason behind our hesitance to enter the fintech party has been the lack of clarity around the real value proposition of fintech lenders - is the fintech lending business one of a.) real credit analytics based on dependable surrogate electronic data instead of offline and subjective data sources or b.) is it a case of building an efficient distribution channel for loans based on an assumption that most people are good borrowers and will repay the loan? (No points for guessing that we prefer option a. above, because lending business is about making loans and collecting repayments. It doesn’t stop at just making the loans.). The problem (lack of clarity) is aggravated by the fact that a lot of fintech lenders use the term “proprietary machine learning algorithm” to justify inability to share details of the logic behind credit calls. So, while we do “get” the high-level value proposition, we weren’t too clear about how it is actually being done/implemented.

 

Another critical aspect is that most fintech firms have adopted a model that is a high cost and high-volume, which means that they aim to invest significantly upfront and be profitable at a scale of around INR 10 Bn or more. This also means that they need to raise huge amounts of capital before they demonstrate profitability and proof of long term success. And, they have done that based on a few key assumptions. What are those assumptions?

 

It is understood that fintech lending is based on two major assumptions:

a.) there is enough electronic “behavioural” data and enough established evidence of direct relationship of that behavioural data with credit behaviour.

b.) by making extensive technology investments upfront, customer acquisition costs will be minimal, at scale i.e. all customers can be acquired online.

 

The key question is whether the assumptions are correct. Today, the short answer is, it is grey. Firstly, there is limited evidence that electronic footprint/behaviour can be a correct predictor of credit behaviour and hence cannot be the sole sources of credit decision making. Secondly, significant (if not majority) of sourcing done by fintech lenders today are through DSAs or offline modes. This results in significant sourcing costs being borne to pay DSAs or to have physical presence.

 

To fill the gaps, some of the fintech lenders have adopted closed loops to source customers and data. Eg: e-commerce platforms or employers. In some other cases, there is a mix of offline and online processes (this goes against the whole pitch of fintech lending but at least some credit process is adopted.). The cost and pricing implications (due to competition) of both these approaches are to be observed going forward.

 

Also to be noted is that there is a crucial difference between lending to MSMEs and lending to individuals in terms of availability of data. While it is a lot easier today to get data on individuals through the India Stack integrations, it is still quite difficult to get entity/MSME data online or on electronic mediums. Even though, for small ticket MSME loans, it is the individual promoter that gets evaluated, the question of data remains for the larger ticket lending and larger MSMEs. There is hope that Udyog Aadhaar will help things in future. Moreover, we are all hoping that with the GSTN network, we will get access to better quality data on the MSMEs.  Across both segments (consumer and enterprise), tools like bank statements analysis, fraud analytics, etc are being developed. So, it is all evolving in the positive direction.

 

One thing is for sure - Fintech lenders have demonstrated how customer experience; loan processing methodologies can be shaken up to be more efficient and data driven. There is a lot to learn from that. However, the jury is still out on whether pure fintech, with credit decisions being dependent only on electronic third-party data, is a reality in India, today. Tomorrow? Sure.

Please reload

Our Recent Posts

Please reload

Archive

Please reload