By Grigoris Karakoulas, President, InfoAgora
Could you please tell our readers a little bit about yourself, your experience, InfoAgora and what your current professional focus is?
I am the president and founder of InfoAgora that, since 2004, has been providing prescriptive customer analytics, RegTech solutions (CECL/IFRS9/IRRBB/Basel III) and model risk management services to financial services organizations. Prior to founding InfoAgora, I was working at CIBC as Vice President, Customer Behavior Analytics, responsible for customer decisioning and credit risk measurement solutions across the bank. I have also been Adjunct Professor in the Department of Computer Science at the University of Toronto. We have deployed our loss forecasting and stress testing solution to Canadian banks for compliance to IFRS9 accounting standards. Leveraging on this expertise we are currently focusing on assisting US banks to implement a methodology that is CECL compliant and appropriate to the size, complexity, and risk profile of the bank’s exposures.
What, for you, are the benefits of attending a conference like the CECL Congress and what can attendees expect to learn from your session?
The CECL standards are principles-based allowing individual institutions to make key decisions on governance, data and modeling methodology. Thus, the CECL Congress can help attendees learn from peers and industry experts like ourselves about best practices on those key decisions and how to implement them in a transparent and traceable way.
In your opinion, what is the best practice for ensuring successful CECL implementation?
From our experience with implementation of IFRS9 in banks of different size and complexity, we recommend the following steps for successful CECL implementation
- Establish governance and oversight
- A team with representation from functional areas across the organization including: finance, risk, credit, IT, internal audit, and financial reporting
- A plan for validating the segmentation and loss estimates during development as well as during on-going monitoring
- Engaging the external auditors as early as possible in the project and keeping them apprised of major decisions and milestones. Since the CECL framework is principles-based certain decisions in the loss estimation could be subjective and hence interpreted differently.
- Allowing at least two quarters of parallel runs with the new allowance method in order to assess operational efficiency, financial reporting, and impact on capital.
- Identify gaps in internal data and supplement them with external data where necessary. This is a critical step and a big challenge for smaller, data constrained banks. The integration with external data should be carefully conducted.
- Select the loss modeling method according to the financial asset type and available data. Although the vintage method has been shown to perform the best compared to other methods, in terms of accuracy and counter-cyclicality of expected credit loss, it may not be feasible due to data limitations. No matter what the method is, the latter should lend itself to explaining q/q changes in allowances.
- Develop scenario(s) that are relevant to the geography of the financial asset and can therefore be used for reasonable and supportable forecasts for losses. This is again particularly important for regional/community banks and credit unions due to their local geography relative to big banks
Why is it important to integrate CECL into forecasting and stress testing processes?
Such integration is important for two main reasons. First, forecasting and stress testing processes for capital adequacy are forward looking similarly to CECL. It is therefore important that the scenario(s) employed for CECL are consistent with the company-run scenarios for stress testing and capital planning. The CECL loss is going to be an input in the capital planning process. Secondly, although CECL does not require more than one scenario, using a few scenarios can reduce q/q volatility in the allowance. It also better captures the skewness of the loss distribution and reduces the impact from error in a single forecast of the macroeconomic conditions.
What are the challenges and opportunities Financial Institutions face in regards to modeling decisions for CECL and how can institutions overcome these challenges?
In regards to modeling decisions for CECL, some of the challenges include:
- Segmentation granularity
- Selection of ECL calculation method
- Linkage to forward-looking scenario(s)
- Extension to life of loan
- Products with non-contractual maturities
- Adjustment for built-in bias and uncertainty in data
- Integration with Q factors
To deal with the above challenges a bank may have to follow a different approach depending on financial asset type and available data.
How do you see CECL impacting the financial risk landscape over the next 6-12 months?
The implementation of CECL could have significant effects on the financial risk of an institution and the industry as a whole. In addition to capital, it could have an impact on the risk appetite and the types of credit products institutions are willing to offer. The accompanying disclosures will provide all stakeholders a better assessment of the financial risk across institutions.