RBI Discussion Paper on ECL based provisioning: A brave attempt to usher in provisioning norms based on Expected Credit Losses (ECL)
Introduction
After announcing deferral of implementation of Ind AS in March 2019, RBI has January 16th 2023 issued a Discussion paper (DP) on Expected Credit Loss (ECL) based provisioning. Even though, Indian NBFCs have been maintaining provisions in line with IndAS 109 (Indian version of IFRS-9), the banks follow IRAC norms for assessment of provisions. The implementation of IndAS for Indian Banks had been deferred earlier citing the need for legislative amendments in Banking Regulation Act. The DP, inter alia covers, ECL estimation approach (i.e., IFRS vs CECL) Scope of assets falling under ECL based provisioning, Classification (i.e., Staging) of assets, Recognition of interest income and Treatment and Disclosure of provisions under the proposed provisioning regime.
RBI has indicated its preference for ECL measurement approach articulated under IFRS-9. Under IFRS-9, besides Probability of default (PD), Loss given Default (LGD and Exposure at Default (EAD) associated with the facilities, the size of provisions required to be maintained is also a function of Staging of assets. The provisions equivalent to Life time ECL are required to be maintained only for Stage-3 (i.e., defaulted) and Stage 2 assets. For Stage-1 assets, the provision shall be based on 12- month ECL. Under IFRS-9, Stage-2 assets are those facilities which had witnessed in Significant increase in Credit Risk (SICR) as on reporting date, when compared with origination date. For e.g., facilities which crossed 60+DPD
DP proposes to lay down certain regulatory backstops with a view to a) ensure adequacy of provisions from financial system standpoint and b) reduce variability of provisions maintained by different banks for assets/ borrowers with homogenous risk profile. The banks are expected to submit their responses to DP by February 2023. Post evaluation of responses and comments, RBI is expected to come up with draft guidelines on ECL based provisioning norms, which inter alia shall cover,
Scope
This article is not an exhaustive summary of salient features\key aspects covered by DP. I would like to highlight certain critical interesting issues discussed in the DP, which merits further deep dive/broader discussion.
Assessment of SICR
Under IFRS-9, the SICR criteria is assessed at facility level; where as in DP it is proposed that SICR should be evaluated at a borrower level, which is more conservative than the treatment followed in IFRS-9. As per DP, increase in credit risk implies increase in risk of default and since default is assessed at a borrower level, SICR should also be evaluated at a borrower level. While this principle sounds very logical for cases where SICR is triggered by criteria such as borrower rating downgrades, it may not be realistic under all scenarios, especially in situations where default risk is assessed & PDs are estimated at pool level.
To illustrate, consider the case of Mr. ABC, who has been onboarded with a Housing loan, Credit Card and a Car loan. On the date of origination, each of these loans have been assigned different pool-based PDs which are a function of application-level attributes and banks’ internal PD models. Now as on reporting date, Credit Card receivable has crossed 30 plus DPD, while other accounts are not overdue. Is it logical to tag the entire carrying amount against Mr. ABC as Stage-2, citing that default risk has increased, ignoring the repayment behavior in servicing other obligations, even when risk of default for his exposure itself is estimated at Product level?
Hopefully, a clear rational guidance would emerge in this regard.
Treatment of Performance Bank Guarantee/Bid Bonds
Under DP it is proposed that, regardless of the Staging derived basis default definition & internally defined SICR criteria, ECL based provision for performance guarantees /bid bonds shall be equivalent to its lifetime ECL. The regulatory intent behind this proposition is not very evident. The only educated guess, which I could think of is that perhaps RBI has decided to consider these instruments as Insurance contracts in line with norms laid under IndAS 104. Since Insurance contracts typically carry insurance risk; not financial risk as in case of financial guarantee, the highest possible loss allowance is proposed for such instruments.
Recommended by LinkedIn
Expanded scope of Default Definition
It is proposed that for the exposures which are in overdue for a continuous period of 90 days should be treated on par with default, considering the time value loss suffered by banks under those cases. This proposal is at variance with definition of default under IRAC norms and global practices. Further it is at cross roads with basic canon of IFRS-9/IndAS 109, which states that ECL estimates should be derived based on reasonable supportable information, which is available without undue cost or effort. Identification of such instances, which fulfill this criterion entails undue cost & efforts. Further in view of penal interest and other charges levied on the borrower for late payments, the time value loss for such instances is minimal in most cases.
Prudential floors
This perhaps may be the most awaited piece in the draft guidelines. Under DP, it is noted that Prudential floors for Stage 2 & Stage 3 assets shall be based on the time spent by assets under these categories and the kind of collateral and level of collateralization against these assets. It is not clear how the collaterals would be considered in fixing of these floors. Further it is stated that these floors shall be based on comprehensive data analysis. While this measure may save banking system from under-provisioning stemming from model risk, on flip side, this measure might discourage the banks in refinement of risk models, typically undertaken to sharpen the risk sensitivity of loss estimates.
As per RBI guidelines on implementation of IndAS for NBFCs, the NBFCs are required to compute provisions under IndAS 109 as well as IRAC norms. In cases where provisions (referred as impairment allowance in RBI guideline Dt March 13,2020) under IndAS 109, falls short of provisions under IRAC norms, such shortfall shall be appropriated from profit and shall be kept as a separate impairment reserve. Funds lying under impairment reserve shall not available for distribution of dividends, even though it shall be reckoned as part of capital funds. Given the requirement of prudential floors, one hopes that RBI shall not replicate the requirement of maintaining impairment reserve for banks, since that measure appear to be an overkill.
Guidance on principles to be considered while design of Credit Risk Models
IFRS-9 is not a prescriptive standard, it essentially advocates principle-based approach for measurement of ECL. However, as a measure to reduce variability in provisions arising from usage of disparate models for credit loss estimation, it is proposed that models used in ECL estimation shall undergo a process check, which entails evaluating compliance of ECL models with RBI’s proposed guidance. Globally, it is common for the regulators to issue detailed guidelines on model risk management. Model Management Standards (essential mandatory principles) and Model Management Guidance issued by Central Bank of UAE (CBUAE) has been the latest addition to the list. It would be interesting to track the evolution of RBI guidelines in this regard.
Conclusion
Intuitively it makes sense to move towards an Expected loss- based approach towards provisioning compared to the current ‘Incurred loss’ approach. It is more proactive and is expected to reduce the procyclicality of provisions. However, is this true in reality? CECL norms have been adopted by certain banks in USA from 2020 onwards. Federal Reserve Note (Dt. Dec 3,2021) has compared the sensitivity of provisions of CECL adopters with non-CECL banks post pandemic and has found that sensitivity of provisions to change in economic outlook was much higher for CECL adopters as compared to non-CECL banks. So, it appears that, while ECL based approach passed the muster of being proactive and facilitate early recognition of losses, it failed on the count of reducing procyclicality of provisions. Further in times such as pandemic, when it is extremely difficult to obtain reasonable and supportable information on economic forecast, is it possible to rely upon the loss estimates under CECL/IFRS-9?
While it is not clear whether Govt has approved of legislative amendments required for implementation of ECL based approach for loan loss provisioning, Indian Banks are in good shape with NPA/ delinquencies at historically low levels and it is widely regarded that Indian banks are in a position to absorb incremental loan loss provisions, which is expected to arise out of implementation of the proposed provisioning framework. DP has rekindled the senior management attention towards IndAS and ECL based provisioning. Let us hope that we get sufficient time to transition from Incurred loss approach to ECL based approach. More importantly, let us hope that ECL based approach would result in a more resilient Indian Banking sector.
(Views are Personal)
CFA, Risk Management
1yThe reason why RBI has proposed ECL computation at the borrower level instead of the product level is to align with existing IRAC norms which considers all the facilities of the borrower as NPA even if one facility is declared NPA while others are performing. Also, the possibility of prescribing an impairment reserve for banks on account of ECL cannot be ruled out. Till recently there was an Investment Fluctuation Reserve that banks were mandated to keep where all gains from HFT and AFS were parked to be used only for meeting depreciation requirements in these categories. But I personally think that introduction of ECL could smoothen the provisioning graph for banks.