One up on stressed assets

The provisioning for non-performing assets, or NPAs, is a highly sensitive and debatable issue, particularly in the case of financial institutions. The current IFRS model (International Financial Reporting Standards) for NPAs is based on the “incurred loss model”. Here the recognition of credit losses is significantly delayed until a credit loss occurs. This model was severely criticised, as the delayed recognition led to the unexpected collapse of several financial institutions during the recent financial crisis. The International Accounting Standards Board was asked to explore alternatives that use forward-looking information.

The IASB earlier issued two exposure drafts (ED) proposing models to account for expected credit losses — at the time of initial recognition of the loan and based on forward-looking information. However, both models were aborted over concerns of operational challenges. The IASB issued a new ED on March 7, 2013 — Financial Instruments: Expected Credit Losses.

Here the recognition of credit losses is based on whether or not the credit quality of a loan has deteriorated significantly at the reporting date. Loans that have not deteriorated significantly in credit quality since initial recognition or which have low credit-risk at the reporting date — only 12-month expected credit losses are recognised. Loans that have deteriorated significantly in credit quality (unless they are low credit-risk or investment grade at the reporting date) — lifetime expected credit losses are recognised. Lifetime expected credit losses are the expected shortfalls in contractual cash flows, taking into account the potential for default at any point during the life of the loan. Twelve-month expected credit losses take into account the potential for default in the next 12 months.

An entity would need to track credit deterioration and apply judgement in determining whether there has been a significant increase in credit risk — for example, credit deterioration is considered significant if an entity would now originate or purchase the financial asset on significantly different terms. The entity should compare the probability of a default (PD) on the loan at the reporting date with the PD at initial recognition. However, this is not easy. The PD for the remaining life of the loan at the reporting date (for example, two years if three years have already passed on a five-year instrument) should be compared to the PD expected at initial recognition for the last two years of its maturity (that is, for years four and five). This requirement can be operationally challenging and entail huge investments in technology. Further, tracking information may not be available for existing loan portfolios with respect to PDs. In India, public sector and small banks may struggle to comply with this requirement.

Under the ED, there is a rebuttable presumption that the lifetime expected credit losses criterion is met if contractual payments are more than 30 days past due. This can pose a significant problem for Indian banks where lack of credit and contractual discipline could lead to a lifetime expected credit loss review, even when the borrower is otherwise absolutely credit-worthy. The Basel definition of default — that is, more than 90 days past due (180 days for retail mortgages) — could be a pragmatic solution.

An entity would need to discount the expected credit losses to the reporting date. The recognition of time value of money will result in recognition of expected credit losses even when the entire loan amount is recovered after the contractual due dates. Further, expected credit losses are based on a probability weighted approach, which is not a best- or worst-case estimate. In other words, all loans will have some PD and some credit losses on initial recognition — albeit a very small amount, even for an absolutely good loan.

Interestingly, the IASB model and the US FASB (Financial Accounting Standards Board) model have differences. Due to concerns raised by FASB’s constituents on the model’s complexity, its proposed expected loss model uses a single measurement approach, under which an entity is required to recognise a full lifetime credit loss on initial recognition of the loan.

There are no established industry practices or methods to determine 12-month or lifetime expected credit losses. IASB, Basel or an expert advisory panel will have to provide more application guidance and practical expediencies so that banks in less-sophisticated economies may find it easier to implement the new standard.

The Reserve Bank of India will have to decide whether it is willing to give up its formula-driven NPA provisioning for a more universal and principle-based provisioning. Experience tells us that the RBI formula results in a much smaller provision for credit losses compared to the current “incurred loss model” of IASB. The expected credit loss model will make it still worse, as it will result in earlier recognition of credit losses. Huge capital will have to be pumped in to maintain capital adequacy. Considering these issues, Indian banks should be given at least three years to implement the standard.

The author is Partner in a member firm of Ernst & Young Global

COMMENT NOW