Sharing some of my notes on the colending piece from discussions with folks in the industry as it’s one of the lesser understood but critical pieces of the thesis considering the proportion that IIFL wants to take colending loans to its overall book.
-
Colending as a model while new to India, is a mature model in markets like Indonesia. Studying the evolution of the model in Indonesia might provide a reference point for how things might evolve in India too. In mature markets, it’s considered as a single business for both entities with a joint operating plan with metrics on how much loans to originate, what customer profiles to look at, acceptable levels of credit losses, pricing for the loans in terms of interest arbitrage between bank and NBFC. In a sense, colending fixes some of the issues from a NBFCs standpoint over direct assignment.
-
The core reason from a bank’s perspective to opt the colending model is it fits the bank when it might not have too many branches and have the reach and number of touch points of a large NBFC. Also, for many of the urban centric banks the model allows them to build a loan book in a segment of customer profiles and loan profiles that would be hard for them to build organically and to instead build it through colending partnerships. For example, for an urban centric bank to make the unit economics work to lend to tractor loans would be very difficult but relatively easy to co-originate these loans alongside a NBFC who has been doing that over multiple years and take 80% of these loans on their balance sheet.
-
The choice of NBFC to work with in the colending model comes from this point of choosing to work with NBFCs who has gone through multiple cycles lending to that particular loan segment in terms of customer profile and asset, has high number of touch points and infrastructure for collections and disbursals, has good quality credit underwriting standards and history of compliance.
-
From a customer standpoint - the flow is such that the originator of the loan becomes their touch point for the entire duration of the loan. The NBFC owns the customer. However, the bank has visibility on how the book is behaving under different circumstances which helps them to model things out in case they want to build their own book.
-
From the NBFC standpoint, important to track the number and quality of colending arrangements they have with multiple banks so that there is enough arbitrage to play with on the pricing front and also to mitigate the risk in terms of disruption in fund flows when you’re dependent only on few partnerships to grow your loan book in a particular asset class. If you have multiple lending partners, it becomes easier to pick and choose on how to distribute the loans you originate between the different partnerships and not be squeezed on pricing.
-
How a bank chooses which NBFC to partner with
- What is the customer segment that the NBFC can bring that the bank is not able to serve directly and avoid cannibalization
- What is the expertise, size and scale for collections, distribution and underwriting
- Pricing in which the bank can co-originate it’s 80%.
- Vintage in working with that asset class and profile.
- Risk premiums that can be modeled for the loan segment.
-
Colending partnerships also require a good degree of integration of systems between the NBFC and the bank. The underwriting process has to be fixed beforehand before the loan is originated. The new CLM norms allow for colending to be run sequentially instead of parallely. However, this doesn’t increase the disbursal TAT as both the 80% and 20% portions need to be disbursed same day. But the loans are disbursed only after both entities have assessed and underwritten the loan.
-
There are few ways in which the model and underwriting flow works
- API driven - very low disbursal TATs, real time underwriting decision across both entities. The loan goes through the underwriting engine of both the NBFC and the bank and when both the underwriting systems are positive, only then it goes through and is tagged as a colending loan.
- Through players like credavenue. This is a smaller chunk of the overall pie. Marketplace for colending.
- Individual file based loan creation. Both entities agree upon credit metrics and the loan is originated by NBFC for whichever profile fits that grid, the loan is disbursed. Towards the end of the day, for whichever loans that have been disbursed it gets sent to the bank that these are the possible colending loans that have gotten booked. These loans are run through the bank’s underwriting engine and there could be some drop offs due to various parameters and minus drop offs the balance gets tagged as colending loans on same day with 80% on bank’s books the same day and 20% on NBFCs books, as interest starts accruing the same day. And the drop offs remain on the NBFCs books.
-
The power in the dynamic between the bank and NBFC is very interesting. For the NBFC, they are getting lower cost of access to funds, they are growing their fee incomes, customer base on external funds and at higher capital productivity. For the bank, they are getting to grow their loan book with the access and reach of a large NBFC without having to replicate its distribution and collection infrastructure. If the NBFC already has deep pockets and is already mature in the ecosystem, the NBFC can pick and choose which banks it wants to work with without disruption to fund flows or getting squeezed on pricing. While from a bank’s perspective, the choice is to work with a set of large NBFCs where they can be reasonably sure that the NBFC can source customers and loans that will not be cannibalizing their own future loan book and also will not go belly up and will continue servicing the client and finish collections on the loan as the bank need not possibly have the collection infrastructure in place for that loan profile to finish the collections for the entire loan tenure, more so for high tenure loans like home loans. Opening up the balance sheet for customers that you don’t see is hard to do for a bank, which is where the NBFCs credit history, credit loss trends, size and scale of touch points, digital capabilities and relationship building comes into play.
-
Pricing for the colending loans is completely dependent on the forecasted credit losses expected on the book. Colending doesn’t start with a player who is new in the industry. Works both ways in terms of how pricing is structured for future co-origination based on how the previous co-originated book has behaved on credit losses based on credit quality indicators. The NBFCs get lower pricing on future co-originated loans if the previous book has had higher than modeled credit losses and vice versa.