Abstract blue network lights

In 2011, the Federal Reserve System (FED) and the Office of the Comptroller of the Currency (OCC) jointly published SR 11-7 which was related to guidance on model risk management and model validations. The guidance is intended for use by banking organizations to assess their organization’s management of model risk. Since 2011, financial institution models have been widely used, however, the results from most of those models have not directly impacted the organization’s financial statements. The adoption of Accounting Standards Update (“ASU”) No 2016-13, Financial Instruments – Credit Losses (Topic 326) that amended accounting for current expected credit losses (CECL) has changed that. Now, model estimates and the possible adverse consequences of decisions based on models being incorrect can directly impact financial statements of all financial institutions. As a result, in 2020, the OCC, FDIC, Federal Reserve, and NCUA issued a joint final interagency policy statement related specifically to the allowance for credit losses (ACL). This statement describes validation considerations related to the CECL estimation process and the importance of obtaining an independent validation of your CECL estimate to properly verify the models used within the organization.

Model validation is the set of processes and activities intended to verify that models are performing as expected and in line with their design objectives and business uses. An effective model validation challenges the model’s data, design, assumptions, and estimates, assesses if proper governance has been established, and finally back-tests the assumptions to actual performance. All model components from the front-end model governance to the back-end reporting analytics and everything in between (data inputs, assumptions, forecasting) should be part of the validation process.

SR 11-7 describes the key elements of a comprehensive validation as:

Evaluation of conceptual soundness:

assessing the quality of the model design and construction as well as the documentation supporting the methods used and variables selected.

Ongoing monitoring:

confirming that the model is appropriately implemented and is being used and performing as intended.

Outcomes analysis

comparing model outs to corresponding actual outcomes. 

Additionally, model validations should be implemented on an ongoing basis to allow management to identify new potential risks and limitations as the organization and the economic landscape continue to evolve over time. As part of sound model risk management framework, regulators recommend that financial institutions have their CECL models validated (usually every 12-24 months) as the risks associated with CECL model results have a larger possible adverse impact to the financial institution’s financial statements than other regulatory required models such as ALM, BSA/AML or Fair Lending that may be validated on a less frequent basis and do not directly impact entity’s financial statements 

Baker Tilly has completed over 100 model validations since CECL was introduced and our approach adheres to all applicable model validation frameworks described above. Aside from checking the regulatory box, our model validations provide insights into the model and its processes, model governance and internal controls to help our clients identify data issues or gaps and user challenges and enhance transparency of the model functionality. Additionally, we provide recommendations to strengthen current CECL frameworks to align more strategically with the financial institution operations. Management and users of CECL models can gain full confidence in the model that was chosen and its appropriateness to their institution.  

Key takeaways from our completed model validation include:
  • Model governance and documentation is just as important as the final CECL estimate. Financial institutions need to own the assumptions and the development and application of those assumptions within the model. 
  • Data extract, transfer and load processes need to be re-evaluated on an ongoing basis to ensure correct mapping of contractual data inputs into the model, especially on cash flow specific CECL models.  
  • Sensitivity analyses and stress-testing of key assumptions should be implemented on an ongoing basis to identify impacts of changing economic landscapes to the bank’s future CECL estimates. 
Ivan Cilik
Sean Statz
Team of professionals working on computer
Next up

DCAA’s focus on audits of contractor business systems highlighted in its latest report to congress