Integration of stress testing and economic capital: Industry approaches

Integration of stress testing and economic capital: Industry approaches

By Tally Ferguson, Director of Enterprise-Wide Risk, BOK Financial

Can you please tell the Risk Insights readers a little bit about yourself, your experiences and what your current professional focus is?

I have been in the financial industry for over 30 years, almost always in a “second line” role.  I started as an international bank examiner with the Federal Reserve Bank of New York.  There I saw Bankers Trust give birth to RAROC and JP Morgan create RiskMetrics.  The swaps market was five years old when I started.  When I left bank supervision after a decade, derivatives had not yet evolved to financial weapons of mass destruction, although I got to witness some impressive income waxes and wanes from derivative transactions.

After the Fed, I served two years as a regulatory consultant for Ernst & Young and helped clients implement numerous regulatory initiatives including comprehensive risk management programs and interest rate risk initiatives.  There, one of my clients was the Bank of Oklahoma (“Bok”).  In 1996, BOk had need of my capital markets and regulatory skills.  I had family in Oklahoma, so I joined them.  There, over the past 22 years, I served multiple 2nd line of defense roles, generally connected to capital markets, market risk, and model risk management.  Currently I am the Director of Enterprise Risk Management, responsible for enterprise wide market risk monitoring, model risk analysis and validation, corporate risk governance and coordinating the corporate insurance program.

What, for you, are the benefits of attending a Congress like the ‘Stress Testing USA Congress’?

The main benefit of Center for Financial Professionals’ congresses is networking.  Here you meet old friends and make new ones who have experiences and insights that they are willing to share.  This sharing helps delegates improve their job performance through new skills or better practices.  This synergy from multiple experts’ sharing ideas is difficult to replicate.

Secondarily, the material itself is eye opening.  CFP spends a great deal of time surveying industry practitioners to learn their priorities and see what challenges financial institutions face today.  Reviewing the agenda is a valuable radar for seeing what issues are coming down the pike.

Finally, CFP congresses grant access to FinTech firms in a considerably less aggressive but more informative way than the cold calls we all get.  From time to time, the open, honest and professional vendor presentations at CFP conferences lead to valuable relationships.

What are key considerations that need to be made when integrating stress testing and economic capital?

The first question we need to ask is do we still believe in economic capital, or is it a relic from a pre-financial crisis time?  Forged in the mines of Basel II’s Pillar 2, economic capital was created to measure unexpected loss.  In theory, an economic capital model condenses the behavior of all assets, liabilities, income and expense items into one large copula of distributions.  Each distribution has a mean or expected value that informs how much income a product or service must return.  Each distribution has a probability density function that allows the user to calculate the change in value for a given cumulative probability.  The industry referred to this calculated change in value as “unexpected loss,” and it called the expected value “expected loss.”  The industry named the given cumulative probability “solvency standard.”

Regulators intended advanced internal rating based banks to determine capital using a very high solvency standard.  Regulators required theoretically sound and complicated models for determining each product and service distribution.  The financial crisis of the Great Recession showed these capital calculations to be too low.  Since then, “stress testing” crowded out “economic capital” in the regulatory lexicon and industry emphasis.

So, do we still believe in economic capital?  Perhaps not as a regulatory capital measure.  However, as I’ll discuss at our Stress Testing USA conference at the close of the first day, we can still believe in economic capital for classical financial economic decisions.  Namely, to measure risk and determine if return is sufficient for that risk.  If we make this leap, we can identify key integration decisions.  These are the considerations upon which I will elaborate in New York:

1.       What scenario should we manage to?  Capital should be set up to protect against Armageddon.  Limit structure and required return, among other measures, must be more practical.  These might employ a much lower solvency standard.

2.       What stress tests are available in our tool box?  Stress tests run the gambit from CCAR to Liquidity Coverage Ratio calculation to Value at Risk.  Most are set up for specific functions.  We should integrate consistently with the purpose of the stress test.  For example, a liquidity stress like the liquidity coverage ratio could be used to form the stress cost of funds in an economic capital model.

3.       Match quantitative approach to measurement accuracy.  Financial Institutions vary in their measurement expertise.  To rely on a modeled component with a high degree of confidence needs two components.  They need complete, consistent and accurate history, and they apply to a product whose behavior can be approximated with a formula or quantitative relationship.  Elsewhere, well considered management overlays are needed.

How do you deal with uncertainty in accurately assessing capital and resource requirements for effective set-up?

Mostly, this goes back to matching quantitative approach to measurement accuracy.  One way to combat uncertainty is to arbitrarily add a number like the capital conservation buffer.  Another approach that I’ll discuss on November 6 is to look more carefully at “uncertainty.”  Of what are we uncertain?  Our models’ predictions?  Economic events?  Customer behavior? Regulatory change?  Technological innovation?  Each warrants a different approach.  To deal with model predictions, widen the solvency standard.  For economic events, apply DFAST/CCAR models that equate economic events to pretax, pre-provision net revenue.  For customer behavior vary model assumptions.  The other two sources of uncertainty require a Palantir.  Absent that, attending congresses like Stress Testing USA are a close second.

Here’s another thing to consider regarding uncertainty.  Recognize interrelationships.  I concede that in times of stress, historic correlations breakdown.  Nevertheless, there are fundamental relationships that generally hold.  Bond prices rise when rates fall, volatility increases option values, economic downturn leads to lower rates.  Also, regulators and government overreact to a crisis.  Recognizing fundamental relationships can help build a diversified set of products and services whose uncertainties might be offsetting.

How can financial institutions benefit from leveraging models in capital planning? What potential modelling problems could they come across?

Mostly, institutions benefit from using a tool they spend time and money creating, validating and defending to regulators.  Our DFAST and CCAR solutions integrate systems across products to arrive at a single measure of capital under three economic scenarios.  Looking back at how we can integrate stress testing with economic capital and how we can deal with uncertainty, benefits of a DFAST/CCAR system jump out.  For example, these give a built-in way to determine the impact to earnings and capital from economic scenarios.  DFAST/CCAR validation efforts provide techniques for determining conceptual soundness of models and giving credible challenge discipline to second lines.  While DFAST is no longer required for financial institutions under $100 billion, the tools that make up DFAST remain useful in a post DFAST world.

Naturally, the DFAST/CCAR model challenges inure to any application for which they are leveraged.  Take economic scenarios, for example.  Regulators offer a pretty wide range of economic variable, but these might not cover scenarios appropriate for regulatory capital.  I’ll use the example of energy prices to illustrate this challenge at the congress.

How do you see the stress testing regulations evolving over the next 6-12 months?

Passage of senate bill 2155 (Economic Growth, Regulatory Relief, and Consumer Protection Act) combined with regulatory guidance on due dates for DFAST submissions tell us how stress testing regulations will evolve for midsize banks.  Emphasis will shift from prescribed scenarios and detailed documentation to model infrastructure and bespoke capital planning.  I expect regulators will be more interested in how banks incorporate idiosyncratic stress testing into their strategic plans.  I doubt economic capital will get much attention from them; however, I do expect more attention on risk return analysis, and economic capital could play an important role here.

You may also be interested in our Stress Testing USA Congress…

Sign up for your free account: