Tag Archive for: digital lending

Vinod Kothari & Chirag Agarwal | finserv@vinodkothari.com

RBI had earlier directed NBFCs to compute expected credit loss (ECL) without considering the impact of any default loss guarantees (DLGs) obtained from its lending service provider (LSP). We had published a short note explaining why this position was debatable (See our article on the topic here) and had also made a formal representation to RBI on the issue. 

Back to the present, RBI has issued an amendment to the IRACP Directions, 2025 (dated February 13, 2026), permitting lenders to factor in DLG while determining provisions under the ECL framework across all stages.

Further, RBI has also specified that upon every event of invocation of DLG, the DLG cover reduces to the extent of invocation. Accordingly, REs shall recompute their ECL provisioning requirements across stages, after duly adjusting for the reduced DLG cover.

With these clarifications now in place, the next question that arises is: How should Regulated Entities (REs) appropriately factor DLG into their ECL computations? The article below discusses the above question at length.

How to factor in DLG in ECL computation?

Let us understand this in simple terms. Suppose a lender estimates that the expected loss on a loan pool is 3.8%. If the lender has received a guarantee of 5%, backed by fixed deposits that are lien-marked in its favour. The  guarantee is sufficient to cover the expected loss. In such a case, effectively, the lender does not expect to bear any loss. On the other hand, if the expected loss is 6.8% and the guarantee covers only 5%, then the lender’s net expected loss would be the balance 1.8%.

However, this adjustment assumes that the guarantee will actually be honoured when required. A guarantee does not, however, eliminate risk completely; it merely shifts the risk of default or loss from the borrower to the guarantor, up to the guaranteed amount.  

DLG & bankruptcy remoteness

The DLG guidelines specify the forms in which a DLG can be obtained. DLG can be accepted in any one of the following forms:

• Cash deposited with the RE; 
• Fixed Deposit maintained with a Scheduled Commercial Bank with a lien marked in favour of the RE; 
• Bank Guarantee in favour of the RE

Accordingly, DLG can only be obtained in fully funded forms, thus eliminating any question of incurring credit loss on such a guarantee. Does that mean that even in case of insolvency of the DLG provider the lender will have the right to invoke the guarantee? The answer to this is negative. Because unlike in the case of bankruptcy-remote SPV, the guarantor is an operating entity, and is prone to the risk of insolvency.

In case of initiation of insolvency proceedings, all the assets of an insolvent entity form part of the insolvency administration/liquidation estate and are beyond the reach of the creditors. The proceeds from the realisation of assets are paid to the creditors in accordance with the waterfall mechanism as specified under section 53 of the IBC, 2016 . 

Accordingly, it becomes important to determine how each permitted form of DLG would be treated in the event of insolvency of the DLG provider.

• Cash deposited with RE: The cash deposited with the lender is actually a liability held in the books till the same is invoked. As per Section 36 of IBC 2016, assets that may or may not be in possession of the corporate debtor including but not limited to encumbered assets form part of the liquidation estate. Accordingly, cash deposited by the DLG provider with RE would form part of the liquidation estate of the guarantor.
• Lien marked FD: Similar to cash deposited with RE, the lien marked FD will also form part of the liquidation estate.
• Bank Guarantee: In the case of a bank guarantee, the credit exposure effectively shifts from the original guarantor to the issuing bank. Given that scheduled commercial banks are subject to stringent regulation and supervision, the risk of insolvency in banks is generally remote. Accordingly, the probability of default in such a structure is unlikely to be impacted.

So, even if the DLG is structured as a funded guarantee, the actual invocation can become complicated if the DLG provider goes into insolvency before such invocation. In such a situation, the lender may not be able to simply invoke the guarantee and take the money. Instead, it may have to submit its claim and wait for distribution under the insolvency process, where payments are made in the statutory priority order. 

Under the waterfall mechanism, secured creditors rank alongside workmen’s dues. Now, in most DLG structures, the guarantor is a fintech entity or a co-lender. These entities typically do not have significant workmen-related liabilities. This may mean that the lender’s priority position is relatively stronger. 

Further, the actual invocation process of the DLG should also be considered. For instance, cash held with the lender can be easily invoked and adjusted as compared to a lien-marked FD or bank guarantee, where there could be procedural delays. 

Illustration: Consider a loan pool of ₹100 crore where the gross ECL rate is estimated at 6.8% (for the static pool covered by the guarantee), resulting in a gross ECL of ₹6.8 crore. The lender has a DLG cover of 5% of the pool (₹5 crore), structured as a lien-marked fixed deposit provided by a fintech sourcing partner. While the DLG is funded, there remains a risk that the guarantor may become insolvent. The first relevant question here is whether we will take a probability of default (PD) as per Stage 1 (12 months PD), or Stage 2/3 (lifelong PD). While the guarantor in question is not in default at all, however, given that the 6.8% ECL is a combination of Stage 1 as well as Stage 2/3 loans, in our view, the PD for the guarantor, to remain conservative, should be the lifelong PD over the tenure of the loans. Let us assume a 20% Probability of Default (PD) for the guarantor. Next question is assessment of Loss Given Default (LGD). As discussed above, the lender has the benefit of full security in form of lien on the fixed deposit, however, there may be depletion of the same on account of first priority in the waterfall, that is, costs of insolvency and bankruptcy process. On a conservative basis, we may, therefore, assume a 10% LGD. Thus, the expected loss on the DLG cover would therefore be 20% × 10% = 2%. 

As a result, the ECL computation may now be:

= 5%*2% + 1.8% = 1.9%

Based on the aforesaid discussion, in our view, while the guarantee is funded the lender may have to adjust the probability of default to factor in the risk of insolvency, particularly where the guarantee is funded in the form of a cash deposit or a lien marked FD. 

Which funded form of DLG is most suited?

As per the analysis, the various options of funded DLG can be ranked basis the maximum consideration allowable for ECL computation:

1. Bank guarantee:  Being bankruptcy remote and easiest to invoke
2. Cash deposit: May have to consider the risk of guarantor’s bankruptcy but the invocation would be easier
3. Lien marked Fixed Deposit: May have to consider the risk of guarantor’s bankruptcy and invocation may involve procedural delays

However, given that there will not be a sizable or material difference in the quantum of counter guarantee risk, the selection of the options for ECL computation may not be significant. 

Can we help this situation?

One of the ways to mitigate the risk of insolvency is by structuring the guarantee in such a way that the guarantee may be invoked upon the occurrence of an adverse material change in the financial condition of the guarantor. In other words, other than the occurrence of losses in the pool, if there are events of default such as adverse material change, insolvency of the guarantor etc., the lender may invoke the guarantee.

Early invocation upon identifiable stress on the part of the guarantor could help the lender realise the guarantee amount before the commencement of insolvency proceedings.

However, such clauses must be appropriately incorporated and drafted in the DLG agreement to ensure the following:

• A clear definition of “adverse material change”
• Identifiable trigger events
• Clarity on invocation mechanism

Impact of DLG invocation on ECL computation 

RBI has also provided a clarification that upon every event of invocation of DLG, the DLG cover reduces to the extent of invocation. Accordingly, REs would be required to recompute their ECL provisioning requirements across stages, after duly adjusting for the reduced DLG cover.

Pool-based guarantees presuppose that the pool is static. This is purely intuitive because if the pool is dynamic, new loans will continue to enter the pool, and therefore, the guarantor’s exposure will keep spreading over a continuing flow of new loans. 

Where the pool is static, the loans gradually get repaid (amortised) over time. As borrowers repay their instalments, the outstanding amount of the loan pool keeps reducing. Since the exposure is shrinking, the ECL on that pool will also typically reduce over time, assuming normal performance. Therefore, whether the utilisation of the DLG on account of pool defaults may cause the ECL computation to increase? This may be so for 2 reasons: one, usual terms of DLG invocation will be the full amount of the defaulted loan will be recovered (due to escalation of the entire principal outstanding). Thus, while the performing loans amortise over time, the non-performing loans are fully recovered once they reach “default”, causing the utilisation of the DLG to run faster than the amortization of the performing loans. Second reason is that once the pool actually starts defaulting, there may be a reason to provide higher estimates of probability of default as well.

Integral part of the contractual terms: Is DLG required to form part of the loan agreement? 

Para 36A of the IRAC Directions read with the principles under Ind AS 109 provides that credit enhancements may be considered while computing ECL only where such enhancements are “integral to the contractual terms.”  The expression “integral to the contractual terms” is taken from the definition of “credit loss” in Ind AS 109. Credit losses are measured after considering the expected cashflows from an asset. Those cashflows will factor in the recovery of any collateral, or credit enhancements, as long as the said credit enhancement is integral to the contractual terms.

What exactly is the meaning of “integral to the contractual terms”? Are we expecting the guarantee (DLG in the present case) to be a part of the terms of the loan contract? That would never be the case, as the so-called guarantee (which may legally be regarded as an indemnity contract) is a bilateral contract between the lender and the DLG provider. Neither is the borrower aware of the guarantee, nor is it desirable to have the borrower know of the guarantee, for obvious reasons. 

IFRS 9 uses the same language. US ASC has more elaborate discussion on this. Para 326-20-30-12 says:

The estimate of expected credit losses shall reflect how credit enhancements (other than those that are freestanding contracts) mitigate expected credit losses on financial assets, including consideration of the financial condition of the guarantor, the willingness of the guarantor to pay, and/or whether any subordinated interests are expected to be capable of absorbing credit losses on any underlying financial assets. However, when estimating expected credit losses, an entity shall not combine a financial asset with a separate freestanding contract that serves to mitigate credit loss. As a result, the estimate of expected credit losses on a financial asset (or group of financial assets) shall not be offset by a freestanding contract (for example, a purchased credit-default swap) that may mitigate expected credit losses on the financial asset (or group of financial assets)

There has been a significant discussion on whether the benefit of a guarantee or credit enhancement which is not a part of the contractual terms of the loan can be factored in ECL computation. From discussions before the IASB, as back as in 2018, two conditions for recognising the benefit of credit enhancements were discussed:

1. part of the contractual terms; and
2. not recognised separately by the entity.

The second condition is easy to understand. For example, if the risk of default is hedged by a credit default swap, the value of the same, amounting to a derivative, is separately recognised. Hence, the question of factoring the same while computing ECL does not arise. However, the first condition, relating to contractual terms of the asset, still remains vague.

One may try to get some clues in the US FASB discussions, where para 326-20-30-12 has been interpreted in technical interpretations. In addition, there is a definition of “freestanding contracts” under the Glossary of ASC 326:

A freestanding contract is entered into either:

a. Separate and apart from any of the entity’s other financial instruments or equity transactions

b. In conjunction with some other transaction and is legally detachable and separately exercisable.

The “forming integral part of the contractual terms” does not warrant the principal contract to provide for the guarantee or the credit enhancements. Insisting on the same will be counter-intuitive, except in case of trilateral contracts. However, the conditions indicate that the guarantee or credit enhancement integrates and becomes an inseparable part of the underlying loan or group of loans. For example, if the group of loans was to be transferred, is it such that the benefit of the guarantee may stay iwth the originator and loans may be transferred, or the guarantee travels along with the loans? If the latter is the case, there is no doubt that in reality, the guarantee has become an embedded part of the loan transaction.

Another factor may be the contractual association between the loan cashflows and the payout from the credit enhancements. Some relevant considerations:

• Is the guarantee specific to the contractual cashflows from the loans?
• Does the guarantor pay what the original loan asset would have paid, or pays independent of the contractual cashflows?
• If the lender subsequently recovers the cashflows from the asset, is the payout from the guarantee restored back to the guarantor?

The presence of these factors will suggest the integration or embedding of the guarantee into the contractual cashflows from the loans.

Conclusion

While the recent amendment by the RBI brings welcome clarity by allowing DLG to be factored into ECL computation, lenders must approach this carefully and realistically. A DLG can reduce the expected loss, but it does not make the risk disappear, as the DLG provider itself faces the risk of insolvency. The form of the guarantee, its enforceability, and the possibility of invocation- all of these matter in assessing the true level of protection. REs should not treat DLG as a mechanical deduction from ECL, but as a risk mitigant that requires thoughtful evaluation, continuous monitoring, and recalibration as the pool amortises and the cover reduces.

See our other resources:

1. Expected credit losses on loans: Guide for NBFCs;
2. Expected to bleed: ECL framework to cause ₹60,000 Cr. hole to Bank Profits;
3. Impact of restructuring on ECL computation.