Stress Testing Deposits: Setting 2014 CCAR Priorities

Regulatory  stress testing focus is  increasingly turning to  deposits. Will the   techniques you employed for  the  recent submission pass muster in  the  next round? As we  discuss below, this  should be  top-of-mind for  businesses as  well as  treasury.

Pencils down! For larger banks, at least.  With  yesterday’s  submission  of the latest round of stress tests (allow us to welcome the regional banks joining the party), the first order of business for CCAR leaders may be a well-deserved vacation. Shortly thereafter, it will be time to identify areas of improvement for the July CCAR submission.

One thing that’s clear is that banks cannot rest on their laurels. The Fed continues to set expectations of a constantly evolving stress testing capability. Relying on 2013 practices will be insufficient for the 2014 submission. A challenger process, where past practices are constantly improved, is the favored advancement means. So what practices should be challenged this year?

For many banks, deposit models will be a priority. Why? In part because deposits have not been the focus to date. In the early years of stress testing, the criticality of loan losses and capital shortfalls overshadowed deposit balance and rate forecasts—and appropriately so. But with a stabilized credit environment and clearer expectations of acceptable capital distributions, robust deposit modeling is the new stress testing frontier.

The potential for rising rates also makes deposit modeling a priority. In rising (and higher) rate environments,

deposit cost is a more significant driver of PPNR (Pre-Provision Net Revenue) and a larger differentiator between banks. For example, when rates last rose from 2004 to 2006, the median cost of deposits for CCAR banks rose 150 bp from 101 bp to 251 bp, and the spread between the 25th and 75th percentile cost of deposits nearly doubled from 31 bp to 52 bp (see the graph below). In addition, the unprecedented growth in certain corporate deposit segments has some experts concerned about deposit balance stability in the event of an economic recovery, adding an undercurrent of liquidity questions to the immediate PPNR ask.

Cost of deposits for traditional CCARR banks (bp)


Which components of your deposit modeling would benefit from a challenger approach? In our experience helping several banks, CCAR executives typically focus on improvements to previous submission rounds. They address frustrations such as:

How can these problems be fixed? Some banks require a soup-to-nuts approach  to  redesigning  (or  design- ing  from  scratch)  deposit  stress  test- ing models. If deposit stress testing has not been a priority component of your CCAR to date, a complete redesign may be in order, such as:



Others need only targeted challenges. From our discussions with the original CCAR banks, the most common areas they feel they need to advance are:

1. Improve the Quality and Quantity of the Modeling Dataset.

Every bank wishes it had a multi- decade, account-level, daily-frequency, consistently-householded history of deposit balances and associated interest rate and earnings credits. None do. Instead, banks are forced to make trade-offs, such as:

2. Advance the List of Independent Variables.

Here the most common need for advancement is with pricing terms— actual pricing (inclusive of things like earnings credit rates on commercial deposits) as well as pricing relative to competitors. Banks without pricing terms in their models are hamstrung when they wish to tweak pricing to fine-tune balance forecasts—without a pricing term, there is no objective link between rate and balance, and the ability to overlay models with anticipated pricing actions is undermined.

Beyond price, most banks have just scratched the surface on independent strategic variables like distribution coverage, product structure and evolution, organizational focus, and marketing spend—often because institutional memory can complicate a bank’s ability to define these independent variable datasets with sufficient accuracy. Finally, some banks have much more advanced practices related to deriving additional macroeconomic variables (e.g., M2) from the set provided by the Fed—while these derivations can be difficult, they are immensely useful and help foster forecast consistency

3. Agree on Appropriate Time Series Modeling Parameters.

One of the adverse consequences of increasingly aggressive model validation units is their tendency to adopt a “my way or the highway” approach to time series modeling. This is most prevalent in the modeling debate be- tween level balance modeling and rate of change modeling. In practice, fears of non-stationarity and multi-collinearity (in English: fears that the modeling technique will violate statistical tenets) drive some banks to pursue one approach at the exclusion of the others, building a “pure” model at the expense of more plausible causal business relationships. To us, the science is important, but so is the experience that leads to integrating an understanding of what is actually happening in the business

4. Step Up Business Involvement.

Too many banks allow deposit stress testing to occur in a vacuum isolated from the business; the business is brought in at the end to sense check the results, but given tight CCAR timeframes, often it is too late to have a material effect. This is a key area for improvement. Because the base case stress test results should align with the business forecast, their involvement is critical early on to understand the data, help develop hypotheses that will inform which independent variables should be considered (especially strategic variables), determine whether there are results that should be excluded or de-emphasized from the modeled dataset, etc. Looping in the business in the right manner is a key determinant of a successful outcome

5. Integrate with Other Deposit Analytics.

Aligning the base case CCAR result with the forecast is now a table stake. Leading banks are pushing the envelope on integration—they aim for a nirvana state where they use the same set of deposit analytics (data, model- ers, etc.) to:

This nirvana state is possible, as is the case in a number of major banks that we have supported. However, it requires defining a comprehensive set of competencies that will be part of this “new reality” and assembling the internal and external resources necessary to get it done.

All in all, this is quite a task list for 2014. But help is out there.

At FMCG, our obsession with deposits dates back to the early 1980s.
For decades we have been helping banks realize the profitability of deposits—first through business unit profitability dissection, then through funds transfer pricing and term liquidity premium curve development, then through our best-of-breed Precision-PriceSM deposit pricing tool. We have spent much of the last two years working for some of the largest banks in the world helping them improve their deposit stress testing capabilities—including two of the major U.S. money centers. Banks that have worked with us have found that by bringing our deposit knowledge (from treasury and business perspectives), our modeling capabilities, and our model validation experience, we are able to materially accelerate their achievement of success. We arrive with sets of hypotheses to test and the “been there, done that” experience of having helped numerous others develop these tools—so we are able to help your staff come up the learning curve quickly. If you would like to compare notes on deposit stress testing practices, feel free to give us a call.

This issue was written by Peter Gilchrist, head of risk management, and Andrew Frisbie, head of deposit analytics. The intellectual property described herein has been developed collectively by numerous officers and other staff of the firm over the last 35 years.

Direct comments or inquiries to, or to the authors
at (212) 557-0500.

©2014, First Manhattan Consulting Group. FMCG is a leading management consulting firm to the financial services industry. Other white papers are available at

Please wait...

Subscribe to our Technical Publication!

Want to be notified when Compass is published? Enter your email address and name below and sign up to our mailing list!
Please wait...

Download the PDF File