Integrated Credit Modeling: From CCAR to CECL

Integrated Credit Modeling: From CCAR to CECL

SONER TUNAY

Q1 – Soner, thank you very much for taking the time to speak to us ahead of the Risk Americas 2017 Convention. You will be leading the pre-convention “Integrated Credit Modeling: From CCAR to CECL” Masterclass on May 22. Why is this an important topic area?

FASB in June of 2016 established the new standard to account for loan losses of financial firms.  Oncoming Current Expected Credit Loss (CECL) methodology would impact all financial institutions.

The primary impact of the change in methodology would be in regards to the loan loss reserves and capital ratios.  There is an ongoing concern about product strategy, pricing, profitability analysis, and capital planning all needing to align to the new loan loss reserving methodology.  There are many implications of CECL ranging from economic forecasting and modelling methodology to data management and public disclosures.  In the Masterclass, we will spend most of the time on scenario generation, credit loss forecasting, life-of-loan concept, and the integration of the risk and finance functions under CECL.

Q2 – You will be starting the masterclass by reviewing the last 5 years of CCAR, looking into models, loss forecasts, capital ratios, and regulatory feedback. How can professionals apply the lessons learned with CCAR to the upcoming CECL process? Without giving too much, what key examples will you be using?

In the last 5 years or so, due to the CCAR requirements, as an industry we have advanced our credit risk modelling methodologies and macroeconomic forecasting capabilities. Additionally, we have learned a great deal about building future balances, assessing liquidity positions, and ultimately evaluating the capital ratios by combining all of these topics.

CECL has many similarities to CCAR.  Lessons learned there would be immensely valuable.  However, the one area where CCAR did not receive adequate attention was the baseline estimates.  CCAR, by design, was a capital assessment exercise.  Therefore, most of the effort was spent on the stress scenarios, and similarly the models, are conservatively built to be sensitive to the macroeconomic shocks.

In the Masterclass, we will be spending a good amount of time on the baseline estimation; in particular, model development and model performance testing perspectives will be covered.

The convergence between loss forecasting for budgeting purposes, baseline, and CECL framework forces banks to spend more attention on modelling accuracy under the more benign scenarios.

Q3 – Last year you led the ‘PPNR Modeling and Scenario Analysis’ masterclass, which was very popular with attendees and sold out. This year’s class will once again review PPNR, what key items will you be addressing and how can professionals integrate PPNR with CECL models?

CCAR started the PPNR modelling paradigm.  As a result, given the remarkable effort from the banks, PPNR modelling has evolved significantly in the last few years.  Last year, we dwelled on the importance of credit and PPNR integration in modelling with applications.  Going forward, we expect to see further interest in PPNR models as they relate to balance forecasting.  Certain aspects of PPNR models, such as prepayments would further support the life-of-loan concern in CECL.

 Overall, the convergence in risk and finance functions, particularly the overlapping nature of the business activities such as loss forecasting, allowance setting, product strategy, and pricing would be further enhanced under CECL.

In this year’s class, we will cover all related activities, including scenario analysis, PPNR modelling, credit loss methodologies, and their particular applicability with the CECL standards.  We predict another highly interactive class due the elevated interest in the topic and a high turnout.

Q4 – You will also join us for the main Convention to discuss CCAR and economic capital. How can banks benefit from integrating these two areas?

We will be taking a critical view of the current EC methodologies and discuss how they can be updated with the recent advancements in credit modelling, particularly in relation to scenario-based models that are heavily used in CCAR.

Despite these new developments in credit risk modelling, applications of CCAR models at the enterprise level have been limited to capital adequacy and carried out mostly for regulatory compliance purposes.

EC models, which belong to a past generation of techniques and assumptions, have remained widely used in many business applications including internal capital allocation, profitability, and pricing considerations.

So far, there has been a limited debate about how to bridge the two worlds of EC and CCAR.  In our debate on EC vs. CCAR we will discuss the pros and cons of each approach with their applications in a bank and explore some ideas for potential integration between these two seemingly different methodologies.

Q5 – What is your current focus, and what key challenges are on the horizon for credit risk professionals and departments, over the next 6-12 months?

Most of the banking institutions which have been under the CCAR regime have advanced their models in the last few years.  We observe that the recent efforts have been based around the fine-tuning of these models.  There is also some work that has started regarding the day-to-day business applications of these models; albeit at a limited rate.

Industry; however, still faces a major challenge in applying stress testing models in benign environments and current baseline estimates.  There is still some additional work to be done on this area.

Additionally, CECL preparations started to consume a significant amount of time as we turn our attention to assessing gaps in our current modelling methodologies, data quality, and infrastructure for CECL standards.  We also started to reach out to our business partners to fully assess the impact of CECL and to be ready to respond to future changes in allowances.

Leave a Reply

Your email address will not be published. Required fields are marked *