The $700 Billion Credit Question for Banks

Written by

It’s the $700 billion question: How bad could it get for banks?

That’s the maximum amount of losses that the Federal Reserve modeled in a special sensitivity analysis in June for the nation’s 34 largest banks over nine quarters as part of its annual stress testing exercise.

Proportional losses could be devastating for community banks, which also tend to lack the sophisticated stress testing models of their bigger peers and employ a more straight-forward approach to risk management. Experts say that community banks should draw inspiration from the Fed’s analysis and broad stress-testing practices to address potential balance sheet risk, even if they don’t undergo a full stress analysis.

“It’s always good to understand your downsides,” says Steve Turner, managing director at Empyrean Solutions, an asset and liability tool for financial institutions. “Economic environments do two things: They tend to trend and then they tend to change abruptly. Most people are really good at predicting trends, very few are good at forecasting the abrupt changes. Stress testing provides you with insight into what could be the abrupt changes.”

For the most part, stress testing, an exercise that subjects existing and historical balance sheet data to a variety of adverse macroeconomic outlooks to create a range of potential outcomes, has been the domain of the largest banks. But considering worst-case scenarios and working backward to mitigate those outcomes — one of the main takeaways and advantages of stress testing — is “unequivocally” part of prudent risk and profitability management for banks, says Ed Young, senior director and capital planning strategist at Moody’s Analytics.

Capital & Liquidity
The results of the Fed’s sensitivity analysis underpinned the regulator’s decision to alter planned capital actions at large banks, capping dividend levels and ceasing most stock repurchase activity. Young says bank boards should look at the analysis and conclusion before revisiting their comfort levels with “how much capital you’re letting exit from your firm today” through planned distributions.

Share repurchases are relatively easy to turn on and off; pausing or cutting a dividend could have more significant consequences. Boards should also revisit the strategic plan and assess the capital intensity of certain planned projects. They may need to pause anticipated acquisitions, business line additions and branch expansions that could expend valuable capital. They also need to be realistic about the likelihood of raising new capital — what form and at what cost — should they need to bolster their ratios.

Boards need to frequently assess their liquidity position too, Young says. Exercises that demonstrate the bank can maintain adequate capital for 12 months mean little if sufficient liquidity runs out after six months.

Credit
When it comes to credit, community banks may want to start by comparing the distribution of the loan portfolios of the banks involved in the exercise to their own. These players are active lenders in many of the same areas that community banks are, with sizable commercial and industrial, commercial real estate and mortgage portfolios.

“You can essentially take those results and translate them, to a certain degree, into your bank’s size and risk profile,” says Frank Manahan, a managing director in KPMG’s financial services practice. “It’s not going to be highly mathematical or highly quantitative, but it is a data point to show you how severe these other institutions expect it to be for them. Then, on a pro-rated basis, you can extract information down to your size.”

Turner says many community banks could “reverse stress test” their loan portfolios to produce useful insights and potential ways to proceed as well as identify emerging weaknesses or risks.

They should try to calculate their loss-absorbing capacity if credit takes a nosedive, or use a tiered approach to imagine if something “bad, really bad and cataclysmic” happens in their market. Credit and loan teams can leverage their knowledge of customers to come up with potential worst-case scenarios for individual borrowers or groups, as well as what it would mean for the bank.

“Rather than say, ‘I project that a worst-case scenarios is X,’ turn it around and say, ‘If I get this level of losses in my owner-occupied commercial real estate portfolio, then I have a capital problem,’” Turner says. “I’ll have a sense of what actions I need to take after that stress test process.”

A key driver of credit problems in the past has been the unemployment rate, Manahan says. Unemployment is at record highs, but banks can still leverage their historical experience of credit performance when unemployment hit 9.5% in June 2009.

“If you’ve done scenarios that show you that an increase in unemployment from 10% to 15% will have this dollar impact on the balance sheet — that is a hugely useful data point,” he says. “That’s essentially a sensitivity analysis, to say that a 1 basis point increase in unemployment translates into … an increase in losses or a decrease in revenue perspective to the balance sheet.”

After identifying the worst-case scenarios, banks should then tackle changing or refining the data or information that will serve as early-warning indicators. That could be a drawdown of deposit accounts, additional requests for deferrals or changes in customer cash flow — anything that may indicate eventual erosion of credit quality. They should then look for those indicators in the borrowers or asset classes that could create the biggest problems for the bank and act accordingly.

Additional insights

  • Experts and executives report that banks are having stress testing conversations monthly, given the heightened risk environment. In normal times, Turner says they can happen semi-annual.
  • Sophisticated models are useful but have their limits, including a lack of historical data for a pandemic. Young points out that the Fed’s sensitivity analysis discussed how big banks are incorporating detailed management judgement on top of their loss models.
  • Vendors exist to help firms do one-time or sporadic stress tests of loan portfolios against a range of potential economic forecasts and can use publicly available information or internal data. This could be an option for firms that want a formal analysis but don’t have the time or money to implement a system internally.
  • Experts recommend taking advantage of opportunities, like the pandemic, to enhance risk management and the processes and procedures around it.

Six Things To Know About CECL Right Now

Written by


CECL-11-13-18.pngMany banks began the transition to CECL in earnest when the final version was issued in 2016. While banks are in various stages, some are already working through more nuanced aspects of the transition.

Many lessons have been learned from actual CECL implementations, and here are some tips to assist bank directors as they guide management through the transition.

1. The quantitative impact of CECL adoption may be less straightforward than initially expected. Even before the final CECL standard was issued, industry observers tried to predict just how much the allowance would increase upon adoption. In truth, it will be almost impossible to estimate the impact of the transition for an individual institution. The actual impact will depend upon many bank-specific factors, the estimation method, the length of the reasonable supportable forecast, the size of today’s qualitative adjustment, and management’s outlook, to name a few. Additionally, some banks with short-term portfolios have been surprised to discover the CECL estimate may be lower than the current allowance due to a shift from an estimate based on a loss emergence period to one that considers the next contractual maturity date.

2. CECL may result in a requirement to manage model risk for unsuspecting institutions. Similar to reserving practices today, banks are employing a variety of approaches. General trends include the largest institutions employing statistical software to build custom in-house models, while the smallest institutions favor a less complex approach that relies on adjusting historical averages. Many institutions who are not using models are relying on “correlations” to support their adjustments. However, this practice needs to be managed carefully, as per regulatory definition, any method that applies a statistical approach, economic, financial, or mathematical theory to derive a quantitative estimate is considered a “model.” Therefore, using a correlation – regardless of whether it is identified in a spreadsheet, vendor solution, or anywhere else – to quantify the impact of a factor is by definition a model, and subject to model risk management. Institutions taking this approach to CECL should carefully consider the scope of model risk management, and avoid accidentally creating or misusing models.

3. Qualitative adjustments will still be necessary. Regardless of the method used to estimate the impact of forecasted conditions, there will still be a need to apply expert judgment for factors not considered in the quantitative (modeled) estimate. Even the most sophisticated models used by the largest banks will not consider every factor. Further, many banks prefer the flexibility to exercise judgment in their reserving process. While it’s not yet clear which factors the industry will use or how to quantify the lifetime impact, as it relates to regulatory and auditor oversight, the level of scrutiny around qualitative adjustments will not decrease from existing practice. Again, accidentally creating models is particularly important given the scrutiny on management judgment and the overall impetus to quantify it.

4. Think beyond compliance. One of the overarching goals of CECL is to better align credit loss measurement with underwriting and risk management practices. The transition to CECL presents banks with an opportunity to have unprecedented insight into the credit portfolio. For example, a comparison between the CECL estimate and the interest margin can provide insight into underwriting practices. But this can only happen if banks take a holistic approach to the transition and make the necessary investment in systems and reporting.

5. Reporting and analytics will be more important than ever. Bank directors will be responsible for answering shareholder questions related to the CECL reserve, which will be sensitive to changes in forecasted conditions. As a key constituent of the disclosures and internal management reports, bank directors have a responsibility to ensure a proper reporting framework is in place – one that integrates the data inputs and quantifies the change in expected credit losses at the instrument level. Attribution reports, for example, will be especially helpful in explaining why the allowance changed because they isolate and quantify the impact of individual variables affecting the reserve.

6. Be prepared for an iterative process, even after adoption. Translating the conceptual to operational can reveal unintended consequences and further questions. The industry has continued to work through implementation concerns since the final version was issued in 2016, including several meetings of the CECL Transition Resource Group. Industry best practices will evolve well after initial adoption.

A Practical Guide for CECL Implementation

Written by


CECL-1-12-18.pngBy now, most community bankers are familiar with the Current Expected Credit Loss standard (CECL), which was issued by the Financial Accounting Standards Board in June of 2016 as a new standard for the recognition and measurement of credit losses for loans and debt securities. However, your bank may be struggling with applying its theoretical concepts. We’ve put together a few simple steps to help you start your implementation process.

Form an implementation team.
CECL implementation cannot be the responsibility of just one or two people. It requires a team that should include:

  • A chief financial officer or equivalent who has knowledge of loan loss accounting and basic modeling capabilities;
  • A chief audit executive or equivalent to identify key controls necessary to the new process;
  • A chief credit officer with deep knowledge of the loan portfolio and related documentation;
  • And a chief technology officer to assist with data gathering and retention.

We advise documenting the members of your team, and briefly summarizing their skill sets and roles in implementing CECL.

Confirm your implementation deadline.
The deadline for implementation of CECL is based on whether or not the bank is considered a Public Business Entity (PBE), unless the institution is a Securities and Exchange Commission registrant. It is important to periodically re-evaluate, document, and receive concurrence from auditors and regulators regarding the bank’s status as a PBE. The American Institute of Certified Public Accountants’ (AICPA) Technical Questions and Answers (TQA) document can help institutions with this determination. Based on this document, most non-SEC registrants will not qualify as a PBE, so most institutions will be expected to implement CECL by December 31, 2021. For SEC registrants, the standard will go into effect one year earlier, in December 2020.

Establish a simple project plan.
A CECL project plan does not need to be voluminous in order to be effective. Start with a single page implementation timeline as a foundation. Next, break the project into manageable segments. For near-term deadlines, record specific tasks and dates. Assign broader timeframes to latter segments to allow sufficient time in the event that there are changes in the bank’s operations, such as an acquisition or PBE classification updates.

Understand CECL’s impact.
It is important to quantify the impact of CECL by understanding industry reserve levels compared to current accounting rules. For the historical loss component of the allowance, which will be the base component for this new standard, current industry data shows the following:

  • Most financial institutions use between a three- and five-year average annual loss rate to compute the historical loss component of the allowance.
  • Based on quarterly call report data, the average three-year net charge-off rate for all bank loans from December 31, 2014 to December 31, 2016 was 0.49 percent. The average five-year net charge-off rate was 0.68 percent.
  • The percentage of allowance to loans for the historical loss component for all banks over $1 billion in assets was 1.24 percent as of September 30, 2017.

Under current accounting rules, this data would suggest that the industry believes incurred losses in the loan portfolio are 0.56 percent to 0.75 percent worse than the average of the last three to five years of actual charge-offs. This could indicate there may be some excess in current reserve levels, which could reduce your previous expectation of the impact of CECL on your institution.

Start retaining available data and use it for modeling.
Consider what models can be built with information that is readily available to most community banks, such as a standard loan trial balance, the history of net charge-offs by loan number and a watch list for set periods of time. Starting with a limited number of data points and simple models can help banks gain familiarity with modeling basics, and identify modeling flaws and potential additional data point requirements.

Effective models such as discounted cash flow, vintage analysis, migration analysis and static pool analysis can be built with these limited reports. The important step in data retention is to ensure core system reports are maintained for a period of time. This will ensure when you begin your modeling efforts, you will have the data necessary to start to build your model.

Even though the implementation date is a couple years away, it is important for institutions to get started with data collection and modeling efforts, as there will be unforeseen challenges along the way. The sooner an institution begins its modeling efforts, the sooner it can identify potential additional data requirements, and the potential impact of this new standard to the balance sheet and income statement.

Using Big Data to Assess Credit Quality for CECL

Written by


CECL-4-7-17.pngThe new Financial Accounting Standards Board (FASB) rules for estimating expected credit losses presents banks with a wide variety of challenges as they work toward compliance.

New Calculation Methods Require New Data
The new FASB standard replaces the incurred loss model for estimating credit losses with the new current expected credit loss (CECL) model. Although the new model will apply to many types of financial assets that are measured at amortized cost, the largest impact for many lenders will be on the allowance for loan and lease losses (ALLL).

Under the CECL model, reporting organizations will make adjustments to their historical loss picture to highlight differences between the risk characteristics of their current portfolio and the risk characteristics of the assets on which their historical losses are based. The information considered includes prior portfolio composition, past events that affected the historic loss, management’s assessment of current conditions and current portfolio composition, and forecast information that the FASB describes as reasonable and supportable.

To develop and support the expected credit losses and any adjustments to historical loss data, banks will need to access a wider array of data that is more forward-looking than the simpler incurred loss model.

Internal Data Inventory: The Clock is Running
Although most of the data needed to perform these various pooling, disclosure and expected credit loss calculations can be found somewhere, in some form, within most bank’s systems, these disparate systems generally are not well integrated. In addition, many data points such as customer financial ratios and other credit loss characteristics are regularly updated and replaced, which can make it impossible to track the historical data needed for determining trends and calculating adjustments. Other customer-specific credit loss characteristics that may be used in loan origination today might not be updated to enable use in expected credit loss models in the future.

Regardless of the specific deadlines that apply to each type of entity, all organizations should start capturing and retaining certain types of financial asset and credit data. These data fields must be captured and maintained permanently over the life of each asset in order to enable appropriate pooling and disclosure and to establish the historical performance trends and loss patterns that will be needed to perform the new expected loss calculations. Internal data elements should focus on risks identified in the portfolio and modeling techniques the organization finds best suited for measuring the risks.

External Economic Data
In addition to locating, capturing, and retaining internal loan portfolio data, banks also must make adjustments to reflect how current conditions and reasonable and supportable forecasts differ from the conditions that existed when the historical loss information was evaluated.

A variety of external macroeconomic conditions can affect expected portfolio performance. Although a few of the largest national banking organizations engage in sophisticated economic forecasting, the vast majority of banks will need to access reliable information from external sources that meet the definition of “reasonable and supportable.”

A good place to start is by reviewing the baseline domestic macroeconomic variables provided by the Office of the Comptroller of the Currency (OCC) for Comprehensive Capital Analysis and Review (CCAR) and Dodd-Frank stress testing (DFAST) purposes. Because regulators use these variables to develop economic scenarios, these variables would seem to provide a reasonable starting point for obtaining potentially relevant historic economic variables and considerations from the regulatory perspective of baseline future economic conditions.

Broad national metrics—such as disposable income growth, unemployment, and housing prices—need to be augmented by comparable local and regional indexes. Data from sources such as the Federal Deposit Insurance Corporation’s quarterly Consolidated Report of Condition and Income (otherwise known as the call report) and Federal Reserve Economic Data (FRED), maintained by the Federal Reserve Bank of St. Louis, also can be useful.

Data List for CECL Compliance

critical-internal-data-elements-small.png

Looking Beyond Compliance
The new FASB reporting standard for credit losses will require banks to present expected losses in a timelier manner, which in turn will provide investors with better information about expected losses. While this new standard presents organizations of all sizes with some significant initial compliance challenges, it also can be viewed as an opportunity to improve performance and upgrade management capabilities.

By understanding the current availability and limitations of portfolio data and by improving the reliability and accuracy of various data elements, banks can be prepared to manage their portfolios in a way that improves income and maximizes capital efficiency.

Fair Value and the Allowance for Credit Losses: What Does the Future Hold?

Written by


These two topics, near and dear to bankers, are in the process of being addressed by the Financial Accounting Standards Board’s (FASB) financial instruments project. The financial instruments proposal, issued in May 2010, was a monster–fair value, impairment and hedging all rolled into one.

Adding to the complexity, remember that FASB is trying to converge with the International Accounting Standards Board (IASB)–but IASB is using timetables that differ from FASB’s. In this post, I’m setting aside hedging to focus on the two areas that affect everyone: fair value and the allowance for credit losses.

Fair Value

red-pin-finances.jpgFASB’s past proposal to carry most financial instruments (loans, securities and deposits) on the balance sheet at fair value was received with little enthusiasm, for two primary reasons. First, robust market data doesn’t exist for many of those instruments, raising concerns about the reliability of the fair values that would be used. Second, the proposal did not take into account management’s intent–when management does not intend to ever sell most of those instruments, what is the point in recording them at fair value?
 
The good news is that FASB has reconsidered its initial proposal. The board has moved away from broadly requiring fair value for most financial instruments. Instead, the determination of whether an instrument is carried at fair value will depend on (1) the characteristics of the financial asset and (2) the business strategy. The result is three categories:


Fair Value–Net Income: Measured at fair value with all changes in fair value recognized in net income. It includes items held in trading or for sale.

  1. Fair Value–Other Comprehensive Income: Measured at fair value with qualifying changes in fair value recognized in other comprehensive income. It includes financial assets for which the business objective is investing with a focus on managing risk exposures and maximizing total return, typically characteristics of an investment portfolio category.
  2. Amortized Cost: Measured at historical cost and for assets, evaluated for impairment. It includes financial instruments for which the business strategy is managing the instruments through a lending, borrowing or customer financing activity, typically characteristics of a loan portfolio category. This category would also include those liabilities (deposits) that the bank intends to hold.

If you think this overall model looks similar to what we have today, you are correct. However, there is an important shift. Today, the accounting model is driven primarily by the form of the instrument. That is, we follow one model if the asset is deemed to be a loan and another model if the asset is deemed to be a security. The shift is really toward one model, but then drivers are the marketability of the instrument (Is there a readily available market for the asset?) and management’s intent (What is the plan for the asset? Is the intent to hold the asset and collect the cash flows the typical intent for a loan? Or is the intent to manage interest rate risk and liquidity, which is typical for an investment portfolio?).
 
While there are many nuances with this area of the project, the key point is that FASB has moved away from essentially requiring fair value for the majority of the balance sheet. Stay tuned–FASB plans to make its final decisions in the third quarter of 2011.

Allowance for Credit Losses

Near and dear to bankers is the allowance for loan and leases losses (ALLL), an area in which FASB’s financial instruments project seeks to make some changes. For decades, we have struggled with the accounting in this area in large measure because of the confusion about what the allowance does and doesn’t represent. Today, losses cannot be recorded until they are probable and incurred. Those are words of art meaning that a loss has indeed happened and that a future event will likely confirm the loss. In other words, the allowance does not represent all possible or expected losses in the portfolio.

However, these concepts are being reevaluated. In late January, FASB and the IASB published a joint proposal for comment that is focused primarily on loans evaluated on a pool basis. Think about that in the context of loans that are not individually flagged being problematic. The proposal seeks to change the allowance from a probable and incurred loss model to a more forward-looking model–that is, closer to an expected loss model.

However, the two boards differ. FASB favors an approach that looks to the foreseeable future but not one that includes losses over the life of the portfolio. IASB favors an approach that takes expected losses over the portfolio and records those losses over time. The new proposal is really a hybrid of the two approaches. Essentially, bankers would have to determine the outcome of both approaches and record the lesser of the two amounts. So, for pools, the allowance for credit losses would be calculated based on the lower of a time-proportional amount (the IASB model) or losses expected to occur in the foreseeable future, which can’t be less than 12 months (the FASB model).

Unsurprisingly, the comment letters did not yield a consensus of opinion. So, with a goal of reaching a consensus by late June, the boards are back to the drawing board. Depending on the outcome, we may see yet another exposure draft. Stay tuned–more fun to come!

Can You See ThiS????