Addressing imperfect results in your PPNR CCAR Models

Addressing imperfect results in your PPNR CCAR Models


Jonathan and Ryan, please tell the Risk Insights’ readers about yourself and your professional experiences.

WES: I’m a Managing Director in Novantas’s Global Treasury and Risk practice, and I oversee the team’s Balance Sheet Modeling practice. Given industry demand, our main focus has been on PPNR modeling for stress testing, but we are actively involved in introducing the best practices of stress testing (statistical rigor, macroeconomic sensitivity, conceptual clarity, etc) into all other aspects of asset/liability management. It’s been an honor to work with so many 15-19 and 15-18 banks, and tracking their progress as the CCAR and DFAST process increasingly becomes BAU.

RYAN: I’m a principal in Novantas’s Global Treasury and Risk practice, specializing in PPNR modeling for CCAR and DFAST and broader balance sheet management. Since the genesis of CCAR I’ve assisted banks ranging from global megabanks to regional retail banks to custody banks to FBOs in the refinement, development, and execution of models to support capital stress testing. At Novantas, I sit on our internal PPNR Stress Testing Steering Committee as well as head PPNR Research and Development. I have also served banks as a consultant for First Manhattan Consulting Group where I focused on risk-adjusted profitability, reporting, and management, funds transfer pricing, liquidity, and balance sheet management.

Your presentation at the Stress Testing USA Congress focused on PPNR CCAR models. Why do you believe this is a key talking point at the Congress?

One of the key talking points was addressing how we can get real value out of the many millions of dollars of investment we as an industry have made in CCAR more generally, and PPNR modeling specifically.

The forecasting of balances, rates, and fees under various macroeconomic regimes has been one of the most challenging components of the CCAR effort for many banks. We’re thrilled to see the industry has moved this from a perennial time and money-sink into a relatively well-solved problem.

As a result, this is some of the most advanced modeling ever done on banks’ balance sheets, but right now it is stuck in the land of regulatory requirements. Well-constructed balance sheet models can be invaluable to Treasurer and CFOs looking to add quantitative rigor to their ALM – and in a way that was not technically feasible in the past.

Why is it so important for FIs to build benchmark models and what are the benefits of gaining a comparison across the industry?

The SR-15-18/19 guidance from the end of last year made it clear that benchmark models are not a regulatory requirement for smaller CCAR institutions. Nevertheless, we have heard from multiple regional bank clients that they are looking for ways to sense-check and provide context for their own bank-specific results.

The obvious application of benchmark models are through the challenger modeling process. They provide an alternative means of estimating and projecting performance – whether in scope, such as a top-down instead of a bottom-up approach, or using different techniques, such as panel regressions. The models can then be used as a back-up model when quantitative methods fail in bank-specific portfolios, or, more commonly, as a top-of-the-house check on the reasonableness of aggregated results and support for model overlays.

However, another crucial application of benchmark models is the qualitative benefit they provide, particularly in the narrative of the stress testing results.

Benchmark models provide invaluable context for framing discussion of banks’ stress test results. The Fed runs their own models and has a view of where each institution fits into the industry – banks should have a perspective here as well. For banks with more vanilla portfolios that track the overall industry closely, benchmark models can be a guidepost for what reasonable growth and macroeconomic sensitivity are, absent the idiosyncrasies of their specific portfolios. For more complex institutions with unique portfolios and compositions, benchmark models are the first step in explaining where the bank’s performance is deviates from industry trends and importantly why this is the case.

How can Novantas assist financial institutions with imperfect results in their PPNR CCAR models?

Novantas has a large, experienced team that has spent years discussing, researching, and solving the many problems that come with PPNR CCAR models. This allows us to be at the forefront of the latest methodologies and applications.

One of the most common ways we help financial institutions is by applying this expertise to assist them in building models that close the gaps on their largest limitations – whether they are models for the bank’s own portfolios, or benchmark models on alternate or industry data. This includes building models to meet not just the needs of stress testing, but unified models that also address business planning and ALM needs.

However, regardless of the techniques used, there are inevitably portfolios or components that are not well-suited to purely quantitative modeling. We also help many banks in developing qualitative estimations, overlays, and narratives which all help to address imperfections in the pure statistical models. Our framework for developing and documenting these elements ensures they are rigorously documented and clearly understandable, while maintaining a concise structure to avoid any unnecessary governance burdens.

How do you see the role of the stress testing professional changing over the next 6-12 months?

The recent comments from the Fed accelerated what we saw happening over the next year. 15-19 banks will be required to maintain CCAR capabilities without the qualitative review. 15-18 banks may now be subject to increased scrutiny as regulators are able to focus their efforts on only the largest institutions. Neither set of banks can afford to continue to fund massive consulting engagements to meet regulatory demands without seeing and business benefit.

Stress testing professionals need to increasingly provide a means of getting business benefit out of stress testing. This means models can’t just be regulatory box-checking exercise; they need to be useful to other constituents throughout the bank. The two biggest areas we see for opportunity are in business planning and ALM. In order to do this, stress testing professionals need to increasingly be the liaison between model validation, the business, and Treasury in order to build and manage models that can meet the requirements of all three.


Our 6th annual Risk Americas 2017 Convention will be taking place on May 23-24 and will feature a dedicated track on Stress Testing and Model Risk. Register before the agenda is announced for the heavily reduced rate of only $999.

Leave a Reply

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