Yale School of Management

Program on Financial Stability

Improving our understanding and management of systemic risk.

Authorities Urge Prudence in Loan Loss Accounting

May 22, 2020
: By Aidan Lawson, Arwin Zeissler, and Greg Feldberg

Access our live “financial-intervention” tracker, which is a single point-of-entry on the YPFS website to keep up with financial-crisis fighting interventions by central banks, fiscal authorities, and international organizations  here.

Banks and government authorities have tried to help struggling borrowers during the COVID-19 crisis, through widespread payment moratoria and modifications in the terms of loans.

But such efforts are costly to banks. Accounting standards require lenders to set aside reserves against borrowers who are unable to meet payments. Setting aside reserves diminishes banks’ income and equity. Nonpayment of interest also hits their income and cash flow.

Meanwhile, recently implemented accounting standards based on lifetime expected credit losses could require banks to sharply increase their loan-loss reserves, further impacting capital.

Of course, credit losses will likely be substantial during this severe economic crisis. Authorities would like to encourage banks to keep working with borrowers, while easing pressure on their earnings and capital to the extent possible.

This blog discusses measures that bank supervisors have taken in two categories:

  • Flexibility in classifying individual loans as troubled or nonperforming.

  • Easing the capital impact of the transition to lifetime expected loss accounting.

The blog then describes key concerns about the flexibility that supervisors have granted. 

The discussion illustrates the tradeoff supervisors face. On the one hand, they want to prevent the procyclicality inherent in regulatory and accounting standards from worsening an already severe credit crunch. On the other hand, they have a responsibility to maintain the integrity and transparency of financial-institution accounting for the long run.

Global regulators, through the Bank for International Settlements, recommend that any regulatory relief to banks should: (1) be effective in supporting economic activity in the long-run; (2) ensure that banks remain well-capitalized, liquid, and profitable; (3) avoid undermining the long-run credibility of financial policies. 

Policy options for flexibility in troubled-asset classification

Borrowers who are experiencing financial hardship may need the terms of their loans eased through restructurings or modifications. Authorities across the world have sought to help such borrowers through payment moratoria and regulatory relief. Loan forbearance, a common form of relief, typically lasts three to six months, in the hope that borrowers’ cash-flow issues will be temporary. (See YPFS’ discussion on forbearance for businesses and individuals)

But accounting standards typically require banks to reclassify modified loans into higher-risk categories, which means that they must set aside additional capital to compensate for the increased risk. Although a modification may increase the long-run likelihood that a borrower is able to repay, the increased provisioning and lower capital levels that result can disincentivize lenders to work with borrowers. 

To reduce that disincentive, governments and regulators have introduced a number of policies, including:

  • Guidance stating payment moratoria can be excluded from the number of “past-due” days for a loan.

  • Guidance stating that loans for borrowers benefiting from government assistance programs should not automatically be reclassified as nonperforming or in forbearance.

The Basel Committee on Banking Supervision (BCBS) has endorsed these strategies, so long as supervisors make sure banks use the flexibility prudently (see here, pp. 5). 

In the U.S., financial regulators issued an interagency statement on March 22, emphasizing that lenders should work with borrowers affected by the virus and providing guidance on accounting for loan modifications and reporting of past-due loan balances. 

On March 27, the U.S. Congress passed the CARES Act. Section 4013 of the Act gives financial institutions the option not to classify modified loans as troubled debt restructurings (TDRs), provided that the borrowers were adversely affected by COVID-19 and that they were current on their payments at the end of 2019. Section 4013 applies to modifications made between March 1 and the end of 2020, or later if the national emergency is still in effect. Under the CARES Act, lenders are also not required to report loans affected by Section 4013 as TDRs in reports to regulators or calculate impairment losses that would normally be associated with such modifications.

After the CARES Act passed, regulators revised the March 22 guidance to incorporate the language in Section 4013. The revised guidance was more flexible than the law. Even modifications that do not qualify under Section 4013 are still not automatically classified as TDRs. The regulators -- with the support of the independent accounting standard-setter, the Financial Accounting Standards Board (FASB) -- stated that “short-term modifications made on a good faith basis in response to COVID-19 to borrowers who were current prior to any relief are not TDRs” (see here, pp. 3). 

Many countries -- for example, Sri Lanka, Romania, Estonia, Kenya, Brazil, South Africa, Australia, Hungary, Nigeria, and Indonesia -- have explicitly stated that loans modified to help borrowers affected by COVID-19 should not be automatically classified as restructured. Romania’s rule applies to non-bank financial institutions (NBFIs) as well as banks. South Korea allows banks to not label loans to closed businesses as “sub-standard,” provided the owners have demonstrated sufficient repayment capabilities. 

Banks in the Philippines can receive similar relief, but only if they have used or are planning to use the central bank’s (BSP) rediscounting facility. Both Russia and Vietnam recommend that lenders assess loans affected by COVID-19 by using a date that precedes the pandemic. Russia also recommends that restructurings or reassessments should not show up on credit histories (see here).

Indian financial institutions are still required to provision for restructured loans. However, they are only required to set aside a provision of 10 percent of the amount outstanding of term loans and interest payments to working capital facilities that were deferred until May 31 (see here). These provisions will be phased in over the first two quarters of 2020.

Policy options for easing the impact of lifetime expected loss accounting

The implementation of new accounting standards is another source of increased loss provisioning. These standards require lenders to conduct forward-looking assessments of expected credit losses (ECLs) over the lifespan of each asset. The earlier “incurred loss” (IL) models allowed banks to set aside provisions only when it was probable that an asset would fail; they were “too little, too late” during the Global Financial Crisis (GFC). In contrast, the new lifetime expected-loss standards require banks to provision more and earlier. This could theoretically mitigate procyclicality by having banks build loss provisions for use during a downturn. 

Lifetime expected-loss models depend on each bank’s ability to accurately forecast and model expected losses. However, by definition, expected-loss models cannot prepare banks for unexpected events such as the COVID-19 crisis. For that reason, the potential capital effects of COVID-19 may be more pronounced than they would have been under an incurred-loss model (see here, pp. 5, 36-37). According to quarterly filings, provisions for credit losses for the six largest U.S. banks increased $20 billion in the first quarter of 2020 due to the combined impact of COVID-19 and their switch to expected-loss accounting. 

Table 1: Expected credit loss provisions of major U.S. financial institutions

Institution

Provision 

(Q4 2019)

Provision

 (Q1 2020)

Difference 

(Q4 2019 and Q1 2020)

JPMorgan Chase

$1.4 billion

$8.3 billion

$6.9 billion

Citigroup

$2.2 billion

$7.0 billion

$4.8 billion

Bank of America

$941 million

$4.8 billion

$3.9 billion

Wells Fargo

$845 million

$4.0 billion

$3.2 billion

Goldman Sachs

$336 million

$937 million

$600 million

Morgan Stanley

$53 million

$388 million

$335 million

Total

$5.8 billion

$25.4 billion

$19.7 billion

To mitigate the impact of expected-loss provisions on capital during this crisis, regulators have pursued the following options:

  • Flexibility in interpreting loss provisions.

  • Temporary sterilization of the effect of new rules on regulatory capital.

  • Allowing banks to temporarily suspend application of the new standard.

Background: CECL and IFRS 9

In the U.S., the Current Expected Credit Losses (CECL) standard was scheduled to come into effect for large, publicly traded banks in 2020. CECL does not specify a given method for measuring expected credit losses. It allows methods such as historical loan loss, roll-rate, and discounted cash flow analysis. Banks are only required to forecast conditions under a timeframe that is “reasonable and supportable.” Beyond that, they are allowed to use historical averages to model long-run behavior (see here, pp. 10 and here, pp. 30 - 31).

The new IFRS-9 standard, now common outside the U.S., is also based on expected losses, but works differently. It requires businesses to dynamically provision for expected losses based on the credit risk of the assets. The standard classifies assets in three stages. Lower-risk (stage 1) assets only require banks to provision for the next 12 months of expected losses (see here). Restructuring, renegotiating, lack of payment, and other behaviors can cause assets to be labeled as having a “significant increase in credit risk” (SICR), resulting in reclassification into high-risk groups (stages 2 and 3). 

The primary difference between the two standards is that CECL requires lifetime expected provisions for all loans and IFRS-9 requires them only for loans that have had a SICR and moved into Stage 2 or 3. IFRS-9 was implemented in 2018, which limits the tools that regulators can use.

 Table 2: Overview of IFRS-9 Expected Credit Loss (ECL) provisioning

ECL Classification

Threshold

Loss Provision

Interest Calculation

Stage 1 (Performing)

Immediately after origination.

12-month expected credit losses

Calculated on the gross carrying amount

Stage 2 (Underperforming)

If credit risk increases significantly and is not considered low.

Lifetime expected credit losses

Calculated on the gross carrying amount

Stage 3 (Nonperforming)

If credit risk increases to the point that it is credit-impaired

Lifetime expected credit losses

Calculated based on the amortized cost

Source: IFRS-9 information

Basel Committee recommendations during COVID-19

The sudden onset and uncertainty during the COVID-19 crisis amplifies the procyclicality of capital and accounting standards. To mitigate this, the Basel Committee has issued several recommendations on expected credit-loss accounting. 

In its guidance, the Basel Committee first emphasized flexibility in determining whether a SICR under IFRS-9 standards has actually occurred. If loans are affected by relief efforts, they should not automatically be reclassified as Stage 2 or 3 on that basis alone. 

Second, the Basel Committee expanded the use of transitional arrangements, which allow lenders to phase in the effect that ECL accounting would have on their Tier 1 capital. The Basel III capital agreement already includes transition arrangements that allow banks to phase in the capital effects of expected-loss accounting over five years. Because of COVID-19, the Basel Committee says that lenders now can either continue to use transition arrangements under the existing framework or defer the capital impact of expected-loss accounting until 2022; after that, banks can phase in the impact on a “straight line basis” over the following three years (see here, pp. 2-3). 

However, these transition arrangements have a limited effect (see here, pp. 6). The Financial Stability Institute argues that combining these arrangements with increased flexibility could provide more relief to lenders. However, they warn that doing so could widen the gap between accounting and prudential measures of capital. 

Many regulatory bodies -- such as the European Securities Markets Authority (ESMA) and European Banking Authority (EBA) -- have followed the Basel recommendation and stated that loans affected by economic recovery programs should not be considered as having a SICR and moved into high-risk groups. They emphasized that the high degree of uncertainty caused by COVID-19 could adversely affect estimated credit losses and damage the reliability of short-term forecasts (see here, p. 4). To resolve this issue, the European Central Bank (ECB) recommended that “banks give a greater weight to long-term macroeconomic forecasts evidenced by historical information when estimating long-term expected credit losses for the purposes of IFRS 9 provisioning policies” (see here). 

In the U.S., the CARES Act states that U.S. banks would not have to adopt CECL until the end of 2020 or the end of the national emergency, whichever is earlier. 

Also, U.S. financial regulators offered institutions that had already adopted CECL the option of delaying its effect on their regulatory capital ratios for two years. After two years, banks have a three-year transition period to phase in the capital impact (see here, pp. 4223). Banks that already adopted CECL can continue to follow the three-year transition period or adopt the new five-year plan instead. Regulators issued a policy statement on May 8 that provides flexibility for management in forecasting and estimating credit losses under CECL. 

The Bank of England (BOE) and Prudential Regulation Authority (PRA) first discussed IFRS-9 and echoed some of the same points as their Euro-area counterparts (see here). The PRA discussed many of these points in greater detail in a letter on March 26, specifically emphasizing the importance of using long-term trends when forecasting and not immediately classifying borrowers into stages 2 and 3. The PRA reiterated that the economic shock from COVID-19 “should be temporary, although its duration is uncertain;” it said that the likelihood of an increase in lifetime credit risk for most borrowers remained unchanged (see here, p. 6). The Australian Prudential Regulation Authority recommended similar post-ECL forecasting adjustments: “If the effects of covid-19 cannot be reflected in models, post-model overlays or adjustments will need to be considered.”

Countries recommending flexibility in calculating expected credit losses include Germany, South Korea, Brazil, New Zealand, Singapore, and Chile. Russia stated that loans affected by COVID-19 restructuring should not be classified as worse than stage II (underperforming). Similarly, Malaysia stated that temporary COVID-19-related shocks can be interpreted as only affecting 12-month expected credit losses, and not lifetime losses. 

Some central banks have supplemented the Basel Committee’s guidance on transitional arrangements by incorporating additional ways to stagger the capital impact that increased provisioning will have. The Central Bank of the UAE required banks to apply a “prudential filter” to IFRS-9 expected loss provisions. This filter allows banks to stagger any IFRS-9 provisions over five years to mitigate the impact on bank capital levels. The Bangko Sentral ng Pilipinas  implemented a similar five-year filter. 

Concerns

The fundamental tradeoff that regulators must consider when implementing these policies is how to mitigate procyclicality while retaining credibility. Credibility is hard to gain and easy to lose, especially in periods of economic stress. The more extensive the relief, the larger the tradeoff. 

Flexibility could erode comparability 

One concern is that flexibility will erode the comparability of financial statements, the main value of accounting standards. Temporary suspensions of already-implemented standards, such as the CECL measures in the CARES Act, may also reduce the comparability of financial statements across and within countries. Of the 17 countries and six international regulators that YPFS identified as having implemented some form of accounting relief, the U.S. and Romania are the only ones that suspended the adoption of new standards. See Table 3 for a full breakdown of country-based relief. CECL forbearance could cause global financial statement comparability issues if enough U.S. banks choose to maintain incurred-loss standards while the rest of the world continues to use lifetime-loss models, as they have since 2018

As it turned out, most major American financial institutions have not opted to use the CECL relief provided by the CARES Act. But most large banks and many small banks did take advantage of the regulatory guidance and delayed the regulatory capital effects of the transition to CECL. The largest U.S. financial institutions have all adopted CECL, and all but three of them -- Wells Fargo, State Street, and Bank of New York Mellon -- have elected to phase in the capital impact over the longer, five-year plan that the regulators made available in the guidance.

The very nature of expected-loss provisioning, which requires managers to make subjective judgments about the future course of the economy, can also erode comparability (see here). Varying national guidance on the application of IFRS-9, as well as increased flexibility in transitional arrangements, may alleviate market-based distress but could compromise comparability. Reduced financial statement comparability and increased heterogeneity across bank forecasts and estimations, on the other hand, can dampen procyclicality as financial institutions increase their provisions at different rates. 

Regulators may allow forbearance to last for too long 

 Prolonged forbearance had an adverse effect in both the Japanese banking crisis in the 1990s and the U.S. savings and loan (S&L) crisis in the 1980s.  

Japanese financial regulators liberalized their rules throughout the 1970s and ‘80s. Strong regulation had previously substituted for bank risk management. As controls were relaxed, it became apparent that financial institutions had not begun using modern risk management techniques (see here, pp. 257-259). 

The Japanese government responded with broad regulatory forbearance and flexibility, allowing banks to understate the extent of their problems (see here, pp. 49). Ultimately, lax loan classification and provisioning rules, continued dividend payments, competing regulatory authorities, and other factors unique to the Japanese banking system harmed the health of banks (see here, pp. 26-28). In the end, the banking system required aggressive government intervention, consisting of tens of trillions of yen in bank rescues and additional assistance. (For more, see the YPFS cases: Unnava 2020a, Unnava 2020b, and Unnava 2020c).

Prolonged forbearance also played an exacerbating role in the U.S. S&L crisis. S&Ls, or thrifts, lent money mainly for long-term, fixed-rate mortgages and funded themselves primarily with short-term deposits at interest rates that were capped by the government. As interest rates rose in the late 1970s and early 80’s, thrifts experienced deposit outflows and faced catastrophic losses because they could not raise their rates alongside commercial banks (see here). 

In response, the U.S. government relaxed accounting and ownership standards and passed legislation that relaxed regulations. Thrifts shifted from mortgages into riskier assets (see here, pp. 180). Large numbers of new thrifts were chartered, and many “goodwill mergers” took place, resulting in the overstatement of capital levels. Ultimately, the federal deposit insurer for the industry went bankrupt amidst a deluge of failures (see here, table 8). Many considered the regulatory forbearance to have been a core cause (see here, pp. 32). (For more, see the YPFS case).

The origins of these crises are very different from COVID-19. In these cases, the extensive, and in some cases exclusive, use of forbearance reflected “the hope that in time a changed economic environment would take care of the problem” (see here, pp. 49). While this rationale still may be a justification for forbearance in the COVID-19 crisis, these policies have complemented large-scale monetary and fiscal policy initiatives supporting both financial and nonfinancial organizations.

Table 3: Summary of International and National Reclassification and Accounting Relief Policies

 

Measure

Nation / Institution

Loan Reclassification

- Past-due day exclusions

Philippines, Egypt, India, Sri Lanka

- Encouraging reclassification flexibility

Australia, Russia (1) (2) (3) (4), Philippines, Nigeria, Ireland, ECB, United States (1) (2), Kenya, Romania, ESMA, India (1) (2), Sri Lanka (1) (2), South Africa, Ukraine, Korea, Estonia, EBA, European Commission, Malaysia

Accounting Changes

- Flexibility in interpreting loss provisions

Russia, ECB (1) (2), Kenya, Brazil, IASB, European Commission, South Africa, New Zealand, BCBS, ESMA, EBA, United Kingdom (1) (2), Korea, Singapore, Australia, Hong Kong, Chile, Germany, Malaysia

  - Sterilization of capital effects

BCBS, Philippines, UAE, United States

- Temporary suspension of new standards

IFRS (1) (2), Romania, United States

Accounting NPLs