The global financial crisis highlighted the need for timely response to and financial reporting of credit losses on loans and other financial instruments.
Current U.S. Generally Accepted Accounting Principles (GAAP) account for credit impairment using an “incurred loss” model, which does not require the recognition of credit loss until the loss is probable or has been incurred. Many have argued that the incurred loss model recognizes too few losses too late and suggest a more forward-looking “expected loss” approach for recognizing credit losses. It is also expected that the developed forward-looking approach could increase the loss reserve level significantly and make the loss reserve level volatile.
To address this issue, the Financial Accounting Standards Board (FASB) and the International Accounting Standards Board (IASB) worked together to devise a common solution. This collaboration continued through July 2012, after which their paths separated.
The approach pursued by FASB and IASB prior to 2012, and by IASB since, is referred to as the “three-bucket impairment” model. It utilizes two different expected loss windows to determine the credit impairment, depending on the extent of credit deterioration. For assets within Bucket 1, an organization will need to recognize lifetime losses expected in the next 12 months. For assets within Buckets 2 and 3, an organization will need to recognize losses expected throughout the lifetime of those assets.
Stakeholders expressed significant concerns to FASB that the three-bucket impairment model could be difficult to operate and audit. Particularly, the threshold to separate the assets using 12-months’ and lifetime expected loss windows is ambiguous. Organizations would need to design thresholds for different asset types and collect significant documentation to justify the decisions and thresholds for each type of asset.
Due to these and other concerns, FASB reached a decision to amend the proposed model and simplify the methodology. IASB continued to develop a three-bucket model.
In December 2012, FASB proposed a Current Expected Credit Loss (CECL) methodology to replace the incurred loss model. FASB is expected to release its final impairment credit accounting standard soon, along with detailed guidelines for the CECL model. The new standard is expected to become effective for public companies by December 2018. For private organizations, the effective date could be extended to December 2020.1
Below is a summary of the CECL methodology proposed by FASB:
- Forward-looking and lifetime credit loss forecasts: Under the existing incurred loss model from FASB (as well as IASB), expected credit loss is primarily driven by past events. In contrast, in the proposed CECL methodology, expected credit losses must reflect current conditions and take into account broader information covering the foreseeable future that could affect the financial assets’ remaining contractual cash flows.
- No threshold to separate impaired and non-impaired loans: Existing loss reserve accounting standards (FAS 5 and FAS 114) include different measurements of credit losses for non-impaired and impaired financial assets. In contrast, CECL requires that organizations forecast credit losses using the same approach, regardless of the current credit quality of the asset.
- Simplified measurement approach: The CECL model would replace the multiple impairment models for different assets, such as loans, receivables, purchased credit impaired loans and hold-to-maturity ebt securities, that currently exist in U.S. GAAP.
Challenges and Opportunities
As previously mentioned, FASB and IASB amended the proposed impairment accounting standard to make sure the reported reserve reflects any change of portfolio quality and the overall internal and external environment on a timely basis. However, the amendment also raises various challenges to the industry, in several areas:
- Organizations need to update their analytical methodologies to generate forward-looking and lifetime loan loss forecasts.
To do that, organizations will need to update or recalibrate their current loss estimation approach to meet the requirements. One of the potential lifetime loss forecasting methodologies is vintage analysis. It was discussed by the Federal Reserve Bank of St. Louis in a webinar in October 2015.2
However, vintage analysis has many limitations. One key limitation is that organizations will need to assume that loans originated in recent years will behave similarly to loans originated several years earlier, even though the underwriting standards, business strategy and macroeconomic environment have changed significantly. For example, for 15-year mortgages, organizations may need to use the vintage curves from more than 10 years ago to estimate lifetime losses for loans originated two years ago. Additionally, historic data points must be available to support this vintage analysis.
In addition, it will be challenging to include sufficient predictive information under vintage analysis. Because vintage analysis is based on long-term historical losses from multiple pools segmented by static risk factors, such as loan-to-value (LTV) and FICO score, the outputs from the vintage analysis may not be sensitive to changes in the current environment. Furthermore, organizations may not be able to obtain reliable loss forecasts if the portfolio is over-segmented by multiple risk factors, because the number of loans in each static pool will become too small to yield stable results.
Because of the above limitations, management may need to adjust the vintage analysis with appropriate supporting information that vintage analysis doesn’t capture, such as current or future acquisition strategy change or underwriting standard change.
For organizations with smaller and less complex portfolios, vintage analysis could be an appropriate approach to generate lifetime loss forecasts, because management does not need to review each line of business and make an adjustment. For organizations with complex portfolios covering multiple jurisdictions and product lines, model frameworks that can integrate information from the present and the foreseeable future would be more efficient and appropriate choices. Such frameworks might include, for example, probability of default (PD), loss given default (LGD) and exposure at default (EAD) models built based on multinomial regression or transition matrix approaches.
An organization should consider its internal resources, portfolio size and portfolio complexity to determine the appropriate modeling methodology.
- Organizations need to collect data with granular information for longer periods of time.
To support the new accounting standards, organizations also need to assess their existing data and IT systems and take steps to enhance their data warehouses and IT systems accordingly. For example, to form lifetime loss forecasts using vintage analysis, an organization will need long-term loan-level data to construct the vintage curves. Below are samples of loan-level variables that may be needed for vintage analysis. Depending on the types of products and the model framework, more data should be collected in the historical database to support CECL calculations.
- Organizations need to redesign their processes and systems.
Organizations may need to redesign their loss reserve processes and systems based on the new accounting standard to reflect changes in methodologies and asset classification.
For example, the existing accounting standard separates impaired and non-impaired assets and requires different measurements for these two groups. However, the new accounting standard will require that the same measurement be applied to all assets, regardless of whether the asset is impaired or not. Thus, organizations will need to make process updates to remove the impairment classification.
In addition, the different assets under the new accounting rule, such as troubled debt restructurings (TDR), purchased credit-impaired (PCI) financial assets and held-to-maturity assets, will need to be reclassified under the new estimation methodology.
Our Point of View
The updated GAAP impairment accounting standard will impact the entire financial services industry, from global banks to community banks, credit unions and non-core financial services organizations with financing arms, such as car companies offering financing in-house. In addition, the new standard will impact multiple internal functions and business lines and require collaborative efforts inside an organization to meet the requirements.
It is expected that the industry will have approximately two to four years to implement the updates. Organizations should have enough time to prepare for the implementation, but they need to begin the planning process now in order to meet the expected timeline.
One of the first steps that an organization should take is to perform a gap assessment to identify areas that need to be improved and to estimate the impact on the loss reserve level based on the CECL method. After the gap assessment, an organization can lay out implementation plans and identify the required resources based on the outputs of the gap assessment. With a thorough and collaborative gap assessment, an organization can prioritize implementation tasks and avoid duplicated efforts from other internal projects.
For example, some organizations have implemented Basel or stress testing programs in the last few years. Their models, data, systems and processes may have been significantly enhanced, such that they could be leveraged for CECL implementation. Smaller organizations, whose Basel or stress testing efforts have been more limited, may find that more areas need to be enhanced and more significant efforts needed to meet the requirements.
How We Help Companies Succeed
Protiviti’s dedicated professionals can help clients identify the key areas for improvement to meet CECL requirements. We apply our expertise and experience in allowance and credit risk models, process design, data acquisition/sourcing, system implementation and model validation to these identified areas to design customized solutions and provide the appropriate resources to execute or support the bank’s execution of the solution.
In addition, Protiviti’s model risk team can assess the institution’s models and provide recommendations to calibrate or re-design the models. Our model risk team also provides independent validation services to assess compliance of the models with CECL requirements and consistency with industry standards.
Our data management team can help institutions enhance their overall data infrastructures, streamline data collection processes and ensure proper governance over the required data assets. These data improvements will be needed to meet the requirements of CECL methodologies and produce timely financial disclosure reports.
Last but not least, Protiviti has a global network of resources that can help institutions with international exposures to implement and validate their implementation programs under the requirements of both FASB’s and IFRS’s updated impairment accounting standards.
Content Contributed by:
Managing Director and Global Leader, Data Management and Advanced Analytics
Managing Director and Practice Leader, Model Risk Management
Director, Model Risk Management
Managing Director, Business Intelligence & Data Governance