An account of your accounting: Four major data issues surrounding CECL
Last year, the Financial Accounting Standards Board (FASB) issued several major documents: Accounting Standards Update (ASU) 2016-13, and Financial Instruments—Credit Losses (Topic 326): Measurement of Credit Losses on Financial Instruments. That adds up to Accounting Standards Codification (ASC) 326—and many finance professionals, so used to the world of numbers, understandably puzzled over the alphabet soup of letters and difficult-to-recite titles.
Here’s the bottom line: This represents a significant change in accounting policy—one with a big impact on how entities report loss allowances for their portfolios.
And here’s how the new standard takes shape: Current Generally Accepted Accounting Principles (GAAP) require that credit losses are recognized with an “incurred loss” methodology. This means banks and other financial institutions report credit losses only after these losses are determined probable, or have already occurred. This inevitably delays credit losses reporting. By contrast, the new accounting standard introduces a Current Expected Credit Loss (CECL) model. This is based on the following:
- expected losses over the life of the loan
- leveraging historical information
- expected results, and
- economic trends.
ASC 326 will affect any institution that carries financial instruments on their balance sheet. The standard will be required beginning Dec. 15, 2019 for public entities that file with the Securities and Exchange Commission (SEC), or Dec. 15, 2021 for private entities and not-for-profits. FASB will allow early standard application after Dec. 15, 2018. (The standard requires entities to apply a cumulative adjustment through retained earnings in the first reporting period).
Both 2019 and 2021 might seem far away. But in reality, financial institutions need to start preparing immediately for CECL if they haven’t already done so. Meanwhile, financial institutions that implement CECL should consider the following issues:
Data diversity: You may need numerous additional data fields besides what is currently stored. Ratings over time, contractual data, loan performance, and items related to borrower behavior might be required; this depends on the methodology employed. Current systems must be evaluated in terms of available data. Meanwhile, data requirements and gap resolution will also depend on the portfolio under consideration. Credit cards, automobile, and student loans mau call for data perhaps not used before.
Expected loss, by definition, is forward looking. This requires modeling loan behavior or cash flows for a reasonably long period—and through economic cycles. A fundamental starting point for projections is to look back and build appropriate assumptions about the future. The look-back period needs to be long enough to capture portfolio behavior under varying economic conditions.
Therefore, data requirements become more stringent and many banks will grapple with this data issue. Data exists at most institutions but not necessarily in a format ready to use for loss forecasting purposes. It might reside in disparate systems, lack quality control and in some instances exist only in loan documents.
Institutions must undertake data gathering with an understanding of gaps, availability and overall quality. Assembling a task force to address these issues is critical. Institutions also need to sweat the details of storage aspects; data lineage and integrity become important as do the controls around extraction and aggregation. Finally, data storage and architecture need to be clearly understood. Incurring costs in this process will prove beneficial in the long run as it avoids costly re-estimations and course correction due to faulty data.
Methodology and models: CECL does not prescribe any specific methodologies or complexity levels required for loss estimates. Some methodologies used under the incurred loss model approach can still apply. These include:
- roll rate
- migration analysis, and
- probability of default/loss given default (PD/LGD)
Modeling tools or platforms need not be complex. Spreadsheet-based models can still work so long as appropriate controls apply. But it’s important for tools to adequately utilize data that concerns past events—and correctly incorporate macroeconomic conditions for future loss projections. For many entities, the challenge of reporting under the new standard may come through the need to apply an effective analytical approach, or sound professional judgment to make accurate estimates. But modeling will evolve into an easier exercise once you address your data challenges.
Documentation demonstration. CECL presents a fundamental shift towards an understanding of losses in the future. Thus it makes sense that certain nuances come to the forefront:
- the forecasting period,
- approach and assumptions beyond a reasonable, supportable forecasting period, and
- use of management judgement vis-a-vis macroeconomic conditions.
Volatility in loss estimates may increase and some assumptions must be made—which means documenting every element of the loss estimation process to address data, methodology, assumptions, overlays, attribution analysis and controls. Nor is this a one-time process. Documentation should become part of ongoing corporate culture.
Good governance, critical controls. Understand this: CECL is not just an accounting and financial reporting level issue isolated to the CFO. Rather, CECL is a regulation that permeates all parts of a business. Accurate, effective adoption of the standard requires a holistic approach: a streamlined processes with collaboration between all parts of the organization. CECL will inevitably affect business models and structures of financial instruments. Many banks will also face governance and risk management challenges during the implementation phase.
With any new accounting standard—let alone one as significant as CECL—entities must prepare for increased scrutiny from regulatory agencies. In turn, entities must adopt an appropriate level of governance and oversight around the estimation process. Controls should address data, models, output and overlays. CECL success calls for detailed planning, a strong project management office and forward- looking technology. Investments for ongoing allowance calculations and reporting will prove critical.
In the meantime, perhaps the confusion behind all those acronyms won’t go away soon. Yet in the end, only one letter grade counts in regard to outstanding preparation and performance: “A.”
Want more Banking Strategies? Sign up for our free newsletter!
Dr. Raman Mandapaka is a managing director and heads the Quantitative Risk Management practice at Navigant Consulting. Hemant Pradhan, associate director in Navigant’s Finance and Accounting practice also contributed to this article.