The Center for Financial Professionals interviewed Stevan Maglic, SVP, Risk Analytics at Regions Bank. He shared his insight on leveraging stress testing for CECL Implementation.
Stevan, can you tell the Risk Insights’ readers about yourself and your professional experiences?
I’ve got over 20 years of experience as a practitioner in risk management and line-of-business functions at Merrill Lynch, BMO, ABN AMRO, and most recently at Regions Bank. I have worked extensively developing and implementing economic capital and other credit-related models. I have also worked in the credit portfolio management space, developing and implementing credit trading and risk mitigation strategies. Over the last seven years, I have focused on developing champion/challenger stress testing methodologies through multiple generations of development and refinement. Increasingly my focus has been on the integration of stress testing with economic capital, risk rating, valuation and other methodologies across the firm into a comprehensive framework. Over the course of my career, I have spent considerable amount of time in front of regulators, model validation, audit, and internal stakeholders, defending methodologies and methodological choices.
How can financial institutions best leverage stress testing methods when working towards CECL implementation?
Use of stress testing infrastructure is a natural choice because CECL requires a forward view of loss expectations. Most institutions with a stress testing program have both economic forecasts as well as the models needed to translate their economic view into loss expectations. To the extent that the stress testing models perform adequately, most of the work has already been done. Indeed, stress testing methods have already been socialized and validated internally, as well as by regulators.
What are some of the benefits of stress testing professionals having a home in the CECL department?
The primary benefit is that many institutions have made tremendous efforts in the development of stress testing infrastructure – in terms of both models and technical staff – and the level of expertise is now very high. Stress testing professionals have been working on a closely related topic for a number of years already, so they bring a great deal to the table when it comes to CECL. Arguably, stress testing professionals are ideally suited to support CECL implementation.
What challenges may professionals come across when applying models used for stress testing, towards CECL implementation?
There are several types of challenges. First, due to the scrutiny that the allowance process receives, allowance teams and auditors understandably have a natural preference for a process that is simple and transparent. Unfortunately, econometrically-driven stress testing models can be very involved. As a result, modelling teams need to ensure that the model users – and auditors as well – do not view their work as a “black box”. Second, in terms of the models themselves, stress testing models are calibrated to be sensitive to adverse economic conditions and tend to overestimate losses under benign economic conditions. For this reason, additional test work and calibration may be required in order to successfully implement a stress testing model for CECL. Third, in terms of impact on CECL reserve estimates, one concern is the degree to which minute changes in quarter-to-quarter economic outlook drive inappropriate or unintuitive changes in the reserves. Since changes in reserve levels effect the net income, it would be very undesirable for reserves to be whipsawed by noisy quarter-to-quarter changes in economic forecasts.
How do you see the role of the risk management professional within the CECL department changing over the next 6-12 months?
Good question. My guess is that we will see more integration of related activity, and that is a good thing. Accountants and statisticians are now talking and there is a good deal to learn from one another. Organizationally, the distinction between allowance and stress testing teams should go away. From a practical perspective, the activities are related, so it makes sense. However, as firms look to improve efficiency and reduce costs, the overhead associated with model development and validation may ultimately favour fewer multipurpose models that can support both stress testing and allowance activity. In terms of process, stress testing teams have lived at a very high modelling standard. However, as with most risk models, reporting accuracy is not as critical. On the other hand, allowance teams tend to work with simpler models, but at a much higher standard of accuracy for accounting purposes. With CECL, both high modelling and accounting standards will be needed so risk management professionals will need to make sure they have a very tight process in place as they implement CECL.
Stevan will be presenting at our CECL 2017 Event. Giving his insight regarding “For CCAR/DFAST institutions, what current synergies from stress testing may be leveraged for CECL implementation?”. He will examine key topics such as: Infrastructure, reliance on models for stress testing and support via backtesting.