Mind the Gaap: US banks brace for $50–100bn capital hit

New loan loss accounting regime could shrink US banks' Common Equity Tier 1 ratios by 25–50bp

Platform step
Step into the unknown: banks are still evaluating the impact of Cecl

  • The US Financial Accounting Standards Board's Current Expected Credit Loss (Cecl) standard, which takes effect in 2020, requires banks to recognise lifetime expected losses on loans at the point of origination.
  • Fitch, the rating agency, estimates loan loss reserves for US banks would increase by between $50 billion and $100 billion if Cecl were implemented today. That would translate into a 25-50bp hit to common equity ratios.
  • Higher loan reserves and the uncertainty inherent in long-term forecasts could push banks to adopt a more conservative lending strategy.
  • As expected losses tend to be higher for longer-term loans, banks may look to shorten the duration of their loan portfolios.
  • Most banks are still assessing the impact of Cecl. The largest 33 US banks, which are subject to the Federal Reserve's annual stress test, have built conditional loss forecasting models that will give them a head start.

Some are calling it the most significant accounting change for the banking industry in over four decades.

The Current Expected Credit Loss (Cecl) rule, which was finalised by the US Financial Accounting Standards Board (FASB) on June 16, will require banks to immediately set aside the full amount of expected credit losses over the lifetime of a loan once it is originated. Currently, credit losses are only recognised in financial statements at the point of impairment.

"Cecl has changed the accounting concept of a loss contingency," says Xihao Hu, head of accounting, tax, planning and finance shared services at TD Bank in New Jersey. "This is going to be a sizeable change to the banking book. Essentially, a significant portion of the balance sheet will be based on forecasts beyond the traditional loss confirmation period."

The implications could be far-reaching. The shift from an incurred loss to an expected loss model will force banks to hold higher reserves against their loan portfolios, which some bankers say equates to an increase in equity capital requirements by stealth.

"The fact that we will likely have more reserves could impact capital planning," says the head of accounting at a mid-size US bank with a large credit portfolio.

Credit rating agency Fitch estimates US banks would need to increase their loan loss reserves by $50 billion to $100 billion if Cecl were implemented today. This would shave 25 to 50 basis points off tangible common equity ratios as retained earnings – which, along with common stock, contribute to a bank's Common Equity Tier 1 capital – would be diverted to boost loan reserves.

Bankers say this would ultimately drive them to be more conservative in their lending, with an emphasis on making short-term loans that are easier to model for losses. "It will be interesting to see if there is a pullback in lending because banks don't want to book those upfront losses," says the head of accounting at a second North American bank with a large US consumer lending business.

A properly executed Cecl model will always result in a recorded loss that is greater than a properly executed incurred loss model
Mike Lundberg, RSM US

The growing unease among bankers is not shared by the FASB, which downplays the significance of Cecl. "No accounting standard can change the economics of a loan. All it can do is reflect the economics differently," says Hal Schroeder, a board member of the Norwalk, Connecticut-based standard setter. "This standard just changes the timing of when you recognise the loss, not the absolute loss that you incur."

Schroeder suggests Cecl is a straightforward change aimed at getting accountants and risk managers on the same page. "If you talk to people inside the institutions that manage credit risk and to people that manage the accounting, they didn't speak the same language," he says. "This standard aligns how risk managers think in terms of expected loss with accountants."

With current US Generally Accepted Accounting Principles (Gaap), the disconnect between risk managers and accountants is certainly real. "A typical chief risk officer would probably be evaluating loans based on what expected lifetime credit losses are," says Sydney Garmong, a partner in the assurance professional practice at auditing firm Crowe Horwath in New York. "That doesn't marry up well with financial reporting because we don't record expected credit losses on day one. That's not the way that current Gaap works."

However, the FASB's solution may come at a steep price. "Banks are going to have to hold more capital against their portfolios, which increases the cost of credit," says Mike Lundberg, a partner in the US financial institutions practice at RSM US, an auditing firm based in Minneapolis.

Out with the old

Cecl will replace the current accounting standard – Financial Accounting Standard 5, which has been in place since 1975 – beginning in the first quarter of 2020, or in 2019 for early adopters.

FAS 5 is based on an incurred loss model, which requires loan losses to be reported if a liability has been incurred by the date of the financial statements, or if a loss is probable and the amount can be reasonably estimated. In contrast, Cecl applies an expected loss model, which incorporates loss estimates based on historical data, current credit conditions and reasonable and supportable forecasts of future performance. As a result, banks will have to report potential losses tied to future events that are expected to occur over the life of the loan, not just those incurred up to the reporting date.

"The credit losses standard operates under the premise that banks don't lend to perfection, and if you make enough loans, it's likely that you took some risk," says FASB's Schroeder. "This standard requires you to reflect that risk on day one."

Credit losses will be recorded in financial statements through an allowance, which will be calculated as the difference between the asset's amortised cost – the loan's book value – and the amount that is expected to be collected.

Mind the GaapCecl also removes the so-called probability threshold enshrined in FAS 5, which states that a bank should only recognise a future loss if it is probable and the amount can be reasonably estimated. "The Cecl model removes that threshold, so even if the risk of loss is remote, it will still be recorded," says Lundberg at RSM US.

The upshot is that banks will almost certainly have to reserve more capital against their loan portfolios, Lundberg says: "A properly executed Cecl model will always result in a recorded loss that is greater than a properly executed incurred loss model."

Lundberg gives the example of a consumer loan portfolio that is heavily exposed to local employment levels. Under FAS 5, a bank would only recognise loss events, such as job cuts, that occur before the financial reporting date. Cecl requires the bank to forecast the likelihood of future job cuts over the remaining lifetime of the loan portfolio and adjust its reserves accordingly. "Under Cecl, we would say that unemployment has averaged 8%, but we think it's going to 10% over the next 30 months, so we would need to add more reserves because we are projecting a future increase in loss events," he says.

In addition to higher baseline allowances for loan losses, Cecl also exposes banks to the volatility inherent in economic forecasting, which will further complicate capital planning. "This presents a new paradigm," says Mike Gullette, vice president for accounting and financial management policy at the American Bankers Association. "If banks manage it well, it could be an improvement for them. If they don't, it could be very frustrating because their capital could swing back and forth."

Unintended consequences

The amount of regulatory capital banks hold to absorb unexpected losses is influenced at least in part by the amount of reserves they hold to cover expected losses. The precise impact of Cecl on capital planning is hard to quantify, but some warn the outcome may not be to regulators' liking.

For instance, assume a bank currently has a capital ratio of 10% – representing a 2% buffer over an 8% regulatory minimum requirement – and loan loss reserves of 0.5% under the incurred loss model. If the bank's loan loss allowance increases to 1% following the transition to Cecl, the immediate effect will be to lower its capital ratio to 9.5% as retained earnings are diverted to bolster loan reserves.

Management would then need to decide whether to raise extra capital to restore its 2% buffer. However, as the higher loan reserves increase the bank's loss absorption capacity by 0.5%, they may conclude the net impact is neutral and that no further action is necessary.

"Although many supervisors may hope each dollar of additional loan loss allowances will increase banks' ability to absorb one more dollar of losses, such a favourable outcome is unlikely," says Larry Wall, executive director of the Center for Financial Innovation and Stability at the Federal Reserve Bank of Atlanta. "Instead, banks are likely to perceive the extra loss absorbing ability in their loan loss allowance as at least partially reducing their need to maintain a capital buffer above the regulatory requirements."

The increased loan loss allowance does not improve a bank's ability to absorb other types of losses, such as operational or trading losses, while the smaller regulatory capital buffer diminishes its ability to opportunistically deploy capital to potentially profitable ventures.

As far as we know, the banking regulators aren't going to change their capital requirements. They're not going to provide capital relief just because now you have a bigger loan allowance
Jonathan Prejean, Deloitte Advisory

Regulators will be watching developments closely. FASB's Schroeder says Cecl has the full backing of US banking and securities regulators. "There was some concern by banks that it's fine for the FASB to say what the standard means, but what really matters is what the field examiners say it means," he says. "We have been talking to state banking examiners and the federal banking agencies to make sure there's a consistent message of what the standard means."

That message is already filtering down to banks. "It's going to be raised to the top from the regulators' perspective," says the mid-size US bank's head of accounting.

The transition to Cecl in the first quarter of 2020 will have an immediate balance sheet impact in the form of a 'cumulative effect adjustment' to retained earnings, which will be diverted to cover the loan loss allowance. "There will be a day when an individual bank switches and they'll have to record the impact of the change on that day through a cumulative effect adjustment, which bypasses the income statement and goes straight to equity," says Lundberg.

Bankers worry that in their zeal to promote higher loss absorbing capacity, regulators will force all banks to raise their loan reserves to the highest common denominator, and leave regulatory capital minimums unchanged. "If they see that JP Morgan is producing a higher allowance than another bank for the same portfolio using a different methodology, they may force the second bank to raise its reserves," says the accounting head.

Few expect bank supervisors to provide relief in the form of lower regulatory capital minimums to offset higher loss reserves. "As far as we know, the banking regulators aren't going to change their capital requirements," says Jonathan Prejean, a director at Deloitte Advisory in Washington, DC. "They're not going to provide capital relief just because now you have a bigger loan allowance. They think that's the way it's supposed to be. Capital is for unexpected losses, and reserves are for expected losses."

The long view

The potential implications for earnings and capital planning could have a profound impact on lending activity. "Once you are in a business-as-usual state, new originations are going to hit current period earnings immediately, which penalises companies with higher growth," says John Erdmann, head of US accounting at TD Bank in New Jersey. "If banks are very concerned about the profit and loss [impact], this could result in a bias towards more conservative loan growth targets."

Risk managers say Cecl will focus attention on the duration of loan portfolios. "Contractual life for different products is going to drive the time horizon, which will then drive the difference between how reserves are calculated today versus how they will be calculated under Cecl," says the head of risk architecture at a US regional bank. "Banks will be looking at their product mix, what the cost is, and what adjustments they need to make to optimise their portfolios and their returns to shareholders."

What we've done with CCAR will help us build models for Cecl more easily than non-CCAR banks
Xihao Hu, TD Bank

One obvious effect is that banks will become reluctant to extend longer-term loans, as the probability of incurring losses tends to increase with the length of the loans. "The duration of your credit portfolio has an impact on both the complexity of the model and the end result," says Lundberg. "This will at the very least put downward pressure on the duration of credit. If you've got a five-year portfolio, and you can shrink it down to a three-year portfolio, that may be helpful from a model complexity and reserves standpoint."

Most banks are still assessing the impact of Cecl and have no tangible plans to change their capital planning or lending strategies at this time. "Before we even get to the point of estimating our impacts, we are stepping back and determining what models we should use and whether we should change models, and where we should use third-party supported models to come up with our estimate," says the head of accounting at a second mid-size bank in the US Midwest. "Certain models may be better than others, and that's what we're looking at now. We're still in the exploratory stage."

According to a survey of bank executives conducted in early July by technology firm Sageworks, two-thirds of respondents have begun discussing Cecl internally, while 22.9% have not taken any action yet, and 11.3% are not sure where to begin.

The US regional bank's head of risk architecture, who was previously in charge of credit model development, was recently given the task of implementing the firm's Cecl programme. The bank has established a number of work streams on loss modelling, data and technology and accounting policy, among others. Of these, loss modelling is the most important, he says.

A matter of principle

The FASB has made it clear that Cecl is a principles-based accounting rule, and banks will have a great deal of latitude in how they model expected loan losses. "You can use a discounted cashflow methodology, a loss rate methodology, a roll rate methodology, or a historic rate methodology," says the risk architecture head. "They're not telling banks what they have to do, which is especially important for smaller banks. There are a lot of community banks that will never be able to develop the type of loss forecasting analytics that the top 30 or 40 banks in the country have."

The 33 largest US banks that are subject to the Federal Reserve Board's annual Comprehensive Capital Analysis and Review (CCAR) have built conditional loss forecasting models based on economic scenarios for stress testing, which will give them a head start with Cecl. "What we've done with CCAR will help us build models for Cecl more easily than non-CCAR banks," says TD Bank's Hu.

However, there is a question mark around whether banks' CCAR models will meet the rigorous control standards required for financial reporting under the Sarbanes-Oxley Act. "A lot of stress testing models that banks are thinking about using have not been through a full Sarbanes-Oxley process," says Chris Boyles, advisory managing director at accounting firm KPMG. "That would be an area to get internal audit to come to the table."

The risk architecture head at the US regional bank has similar concerns. "To the extent that banks leverage work that supports CCAR, the type of controls and production environment that they'll require for Cecl in order to pass the Sarbanes-Oxley controls will be a higher standard," he says.

Meanwhile, some banks worry that Cecl might make the annual CCAR exercise more difficult. Although the work done for stress testing will help firms to model loan losses, CCAR puts the emphasis on extreme scenarios, while Cecl is purely concerned with expected or baseline losses.

"For CCAR, we have to submit multiple loss forecasting scenarios as part of a larger capital plan. Although the base forecast is one scenario, the most important by far for CCAR is the severely adverse scenario," says the risk architecture head. "For Cecl, the only scenario that matters is the base scenario, which generates the expected losses for the loan and investment portfolios."

The worry is that loss allowances required for Cecl will balloon when subjected to the Fed's adverse and severely adverse scenarios, putting further pressure on capital requirements. "It will also put pressure on CCAR results, because CCAR stresses Gaap-based financial results, so whatever we're doing – such as an additional loss reserve put on under Cecl – can be further exacerbated in the CCAR results under adverse and severely adverse scenarios," says Hu at TD Bank.

Only users who have a paid subscription or are part of a corporate subscription are able to print or copy content.

To access these options, along with all other subscription benefits, please contact info@risk.net or view our subscription options here: http://subscriptions.risk.net/subscribe

You are currently unable to copy this content. Please contact info@risk.net to find out more.

You need to sign in to use this feature. If you don’t have a Risk.net account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an individual account here