Integrating CECL and DFAST Risk Management

By Oleg A. Blokhin, David W. Keever, and Chad C. Kellar, CPA
| 3/31/2017
Integrating CECL and DFAST Risk Management

As banks work to comply with the enhanced credit management, financial modeling, and forecasting requirements stemming from recent changes to accounting standards and regulatory guidance, some are discovering they can streamline compliance by integrating their data monitoring, management, and governance programs.

The transition to the new current expected credit loss (CECL) model for estimating credit losses, as spelled out by the Financial Accounting Standards Board (FASB), is presenting many banks and other lending organizations with some significant data management challenges. These challenges are in addition to the ongoing data issues many banks must address as they work to comply with Dodd-Frank Wall Street Reform and Consumer Protection Act stress-testing (DFAST) requirements.

Of course, there is no direct statutory or regulatory relationship between CECL, which is an update to current accounting and financial reporting standards, and DFAST, which is a regulatory requirement. Nevertheless, the similarities in their data requirements and forecasting methods suggest that many organizations could benefit by combining – or at a minimum coordinating – the necessary data management and governance components of their CECL and DFAST projects.

Growing Credit Data Challenges

The parallel CECL and DFAST requirements for measuring, managing, and monitoring credit risk are presenting banks with a number of new challenges. Many of these challenges are the result of the enhanced modeling and forecasting regimens that must be implemented.

The objectives of these regimens obviously are worthwhile – and very similar. In the case of the FASB’s CECL calculations, the purpose is to provide investors and other stakeholders with a forward-looking view of lifetime credit risk. The objective of the DFAST requirements is to alert regulators and the public to risks to financial services organizations’ long-term stability and sustainability.

The expanding loan-level data and reporting requirements associated with these efforts are, in many cases, exceeding banks’ existing data sources and data management capabilities. The extensive data that is needed and used for models and analysis is driving the need for enterprise credit data management.

Currently, in many organizations, the data collected is siloed and not easily integrated across the enterprise. At the same time, the costs of acquiring, monitoring, and governing data and models are substantial and growing. This situation often places severe demands on departmental resources and operating costs.

Fortunately, there are substantial opportunities for integrating data, processes, controls, and models.

CECL and DFAST Compared

To understand where CECL and DFAST data requirements coincide – and, equally important, where they do not coincide – a brief review of the two systems is helpful. At the most fundamental level, the differences between CECL and DFAST can be summarized as follows:

Issued by FASB in June 2016, the CECL method prescribed by the final standard on impairment is the most significant change in decades for financial services companies. In general, the application of CECL will move the measurement of credit losses from an incurred loss basis to an expected loss basis by requiring that entities consider more forward-looking information than is permitted under current U.S. generally accepted accounting principles and recognize the expected lifetime losses on financial assets upon origination. For public business entities (PBEs) that meet the definition of an SEC filer, the standard becomes effective in fiscal years beginning after Dec. 15, 2019, including interims within that fiscal year. PBEs that are non-SEC filers will adopt effective the first quarter of the fiscal year beginning after Dec. 15, 2020. All other entities will not adopt until the fourth quarter of the fiscal year beginning after Dec. 15, 2020. DFAST is a financial reporting requirement authorized by Section 165(i)(2) of the Dodd-Frank Act and implemented by federal banking regulatory agencies. Under DFAST, banks with total consolidated assets of more than $10 billion are required to conduct annual stress tests using economic scenarios provided by the Federal Deposit Insurance Corporation (FDIC). Beginning in 2016, all covered banks with between $10 and $50 billion in assets were required to submit the results of their company-run stress tests to the FDIC by July 31 and to publish those results between Oct. 15 and Oct. 31. Covered financial services companies with $50 billion in assets or more are required to submit the results of their company-run stress tests to the FDIC by April 5, 2017, and to publish those results between June 15 and July 15, 2017.
CECL is a method for assessing, accounting for, and reporting expected credit losses in a bank’s loan and lease portfolios as well as other assets held at amortized cost. The objective is to avoid overstatement of an asset’s value by requiring the recognition of the expected remaining lifetime credit losses, including the recognition of lifetime credit losses upon origination of a financial asset. The standard will impact earnings directly. DFAST is designed to assess the forecasted impact that “future shock” scenarios would have on a bank’s financial statements. The objective is to identify capital and operating risks in order to maintain appropriate capital levels to mitigate the risk of failure. Unlike CECL, this regulation includes future originations.
CECL is to be calculated and reported quarterly and reflects the risk of loss of only the assets outstanding as of the measurement period. Detailed disclosures are required in audited financial statements as well as interim periods for entities filing with the SEC. DFAST is calculated and reported annually or semi-annually, depending on a bank’s size, with only minimal public disclosure of the results of severely adverse scenarios.

Despite these clear distinctions, the data management, financial modeling, and forecasting processes associated with CECL and DFAST do share some important similarities as well. CECL and DFAST compliance methods both:

  • Employ comparable credit modeling techniques
  • Look to similar sets of historical data and use those for developing forecasts of future activity
  • Employ similar techniques for segmenting loan portfolios, grouping assets into pools with similar characteristics
  • Require that forecasts be reasonable – that is, based on how the bank currently is doing business or planning to do business
  • Require that forecasts be supportable – that is, based on verifiable data and information the bank has available
  • Require that banks must have certain governance structures, processes, and controls in place to support compliance with the various data management, forecasting, and reporting requirements
  • Often engage internal resources with similar skills and capabilities, and apply technology that performs similar functions

Financial Modeling Approaches

As mentioned, one of the most significant similarities between CECL and DFAST calculations involves the use of similar financial modeling techniques. These models are designed to explore various scenarios to help management estimate the likely effects of possible future events, such as changes in interest rates, general economic conditions, collateral values, and other variables. These modeling techniques can be either “top-down” or “bottom-up” in nature.

Top-down modeling frameworks implicitly assume that a bank’s portfolio characteristics remain broadly constant over the relevant time period being tested. So when the current portfolio characteristics being used for CECL calculations differ from the bank’s historical characteristics, the model forecasts must be adjusted. Similarly, when the DFAST scenarios issued by regulators differ from historical patterns, the model forecasts must be adjusted.

Such adjustments typically are accomplished through direct management overrides. Advanced modeling techniques can assist in this process by providing loss-rate estimates according to grade and industry sector. This process allows appropriate adjustments to more accurately reflect the effects of changes in portfolio composition. Performing these advanced modeling and forecasting processes can be costly and resource-intensive. Integrating CECL and DFAST projects can help mitigate these costs.

Bottom-up or loan-level modeling frameworks segment a bank’s loan portfolio according to various factors such as product type, geographic area or region, specific loan terms and conditions, collateral, and other variables. This type of modeling involves gathering additional data elements regarding historical portfolio and economic performance in order to calculate the probability of default, potential loss in the event of default, value at risk, and other critical outcomes.

In the case of DFAST forecasting, these calculations involve the use of quarterly data, while CECL calculations use data covering the life of the loan. So while the specific data requirements vary somewhat, they are nevertheless fundamentally similar.

Identifying Data Integration Opportunities

While the opportunities to integrate data management processes, modeling techniques, and governance structures are numerous, identifying the specific confluence points is not always immediately obvious. Achieving the benefits of this integration requires a systematic plan for developing cross-functional data teams and systems.

The ultimate goal of the plan is to integrate the CECL and DFAST projects by aligning the relevant planning and maintenance cycles; sharing data, models, and methods; and engaging in common challenge, validation, and audit activities. Although the specifics will vary from one organization to the next, most banks will find similar confluence points in which CECL and DFAST projects intersect and can be integrated. These opportunities typically will include:

  • Assessments and model development. This involves aligning the CECL risk identification and data inventory processes with DFAST gap assessment and model development activities
  • Resources and technology. This involves matching the resource capabilities and enabling technology needed for CECL calculations with the software solutions used for preparing DFAST and related stress-testing activities.
  • Governance, challenge, and validation. This is coordinating model risk and governance oversight, along with shared internal audit, validation, and assessment activities.

In the coming years, the finance, credit, risk, treasury, and internal audit functions all will need access to integrated data sources. Having enterprise data that is linked to all areas – including allowance for loan and lease losses (ALLL) calculations, stress testing, concentration monitoring, reporting, and analytics – will become an important operational advantage.

By converging related data activities, and by integrating separate CECL and DFAST data management, acquisition, and governance processes, financial services companies can gain efficiencies. Even more important, the formation of cross-functional data teams can help these companies align their methods so they produce consistent, workable, and credible forecasts of portfolio performance.

Contact us

Oleg Blokhin
Dave Keever
Chad Kellar
Chad Kellar