The CECL vision

The CECL vision

Venkat Veeramani the SVP of Risk Strategy and Analytics at Wintrust Financial Corporation shares his knowledge with us on understanding the CECL end vision to effectively prepare for implementation. 
Can you tell the Risk Insights’ readers about yourself and your professional experiences?

Venkat is currently SVP Risk Strategy and Analytics at Wintrust Financial Corporation. He is an accomplished enterprise-wide analytics thought leader and a subject matter expert on risk management life-cycle. He has successfully led several high visibility risk and financial analytical initiatives at multinational and midsize financial institutions. He is a published author and frequent speaker on topics related to game theory, risk & financial analytics and creation of data-driven business intelligence. He has previously worked at Morgan Stanley, Discover Financial Services and HSBC.

 

Can you provide an overview as to the key considerations when developing a CECL compliant reserving methodology and how this differs from current methods?

The current incurred loss based reserving methodology is like using your rear-view mirror to drive forward. On the other hand, CECL forces financial institutions to consider a lot more information, including incorporating forward-looking information into the allowance process. Incorporating forward-looking information is similar in concept to the annual or semi-annual regulatory stress test except that CECL requires more frequent runs and disclosures. Additionally, financial institutions need to generate their own economic scenarios for estimating reserves under CECL.

Things to consider are:

  • Data – availability of clean, accurate data over a longer time period,
  • Analytical Resources – analytical resources for developing quantifiable estimates of institution-specific factors and external factors, including macroeconomic, financial markets and interest rates that drive lifetime expected credit losses,
  • Strategic Resources – resources for generating forward-looking projections of institution-specific and external factors,
  • Systems – a flexible integrated system that has internal-controls over data, qualitative factors, quantitative factors, forward-looking projections, scenario analysis and disclosures.
  • Processes and Procedures – end-to-end CECL effective challenge processes and procedures.

 

How can institutions ensure that ‘most likely’ economic scenarios can be updated quarterly?

A typical stress test process kicks into high gear with the release of the stress scenarios from the Federal Reserve. Similarly, I would expect the forward-looking economic scenarios to kick-start the CECL ALLL process. For CECL, financial institutions need to generate their own forward-looking institution-specific scenarios. For starters, each institution needs to identify key internal and external portfolio drivers and establish a database of past and current data points of those drivers. Using those data points, forward-looking economic scenarios could be developed and added to the database. The ‘most likely’ economic scenario could be selected from a range of forward-looking economic scenarios via a committee or working group. Establishing a database of current and past scenarios could help in explaining quarter to quarter changes to internal and external stakeholders. From an effective challenge standpoint, institutions may end up using a combination of internal and external sources for generating institution-specific economic scenarios. Since the provision expenses are front loaded under CECL, from an impact analysis standpoint, institutions may find it beneficial to run CECL ALLL analysis on a more frequent basis than quarterly.

 

What are the potential impacts of qualitative components to reserving methodologies, and how can institutions ensure reflective numbers without too high a qualitative sway?

Since CECL has a quantitative tone to it, arriving at the final ALLL numbers using a qualitatively dominant methodology may not fly. However, solely relying on quantitative models to produce outputs for CECL purposes may introduce more volatility. I don’t think there is a prescribed proportion of quantitative and qualitative factors contribution that needs to be used in estimating provision expenses. Several factors including market conditions, portfolio size, composition, complexity and materiality determine the use of quantitative vs qualitative factors. Sensitivity analysis could be used in documenting the rationale behind inclusion/non-inclusion of factors in estimating the provision expenses.

 

Finally, what challenges do you foresee with CECL implementation over the coming years and how can institutions best plan meet deadlines?

Since CECL requires inputs from multiple functions outside of the traditional finance and accounting functions within an institution, establishing a CECL implementation team is a necessary first step. Developing a CECL implementation plan and performing CECL gap analysis are other things that will help institutions meet deadlines. For quarter-end reporting, provision estimation process needs to be completed within a few days. Therefore, establishing an integrated system for running end to end CECL processes in days is vital for a successful CECL implementation. While some elements of the regulatory stress test processes are highly transferable in operationalizing CECL, in my opinion, CECL won’t be as simple as running DFAST quarterly.

Because of the disclosure requirements, frequency of estimation runs and granularity of analysis, even seasoned stress test professionals may find the CECL process tougher to negotiate, at least in the initial years. Additionally, as the front loading of provision expenses under CECL has the potential to impact lending dynamics and profitability calculations, institutions should be prepared to tackle unexpected or unknown challenges.

Venkat will be speaking at the CECL Congress 2017, October 11-12 in New York City. He will address key topics including: System requirements and capabilities, road map for implementation and estimating impact for initial shock. 

 

//]]>