# LO 56.3: Explain challenges in modeling a bank’s revenues, losses, and its balance

LO 56.3: Explain challenges in modeling a banks revenues, losses, and its balance sheet over a stress test horizon period.
Current stress tests are based on macro-scenarios (e.g., unemployment, GDP growth, the HPI). One concern is how to translate the macro-risk factors employed in stress testing into micro (i.e., bank-specific) outcomes related to revenues and losses. Supervisors need to map from macro-factors into intermediate risk factors that drive losses in specific products and geographic areas. Although not limited to these products, geographic differences are especially important in modeling losses in both commercial and residential real estate lending.
Credit card losses are particularly sensitive to unemployment figures. For example, unemployment was 12.9% in Nevada in July 2011, 3.3% in North Dakota, and the national unemployment rate was 9.1%. Credit card loss rates varied dramatically from region to region during this period. The geographic diversity with respect to macro-factors makes a one-size-fits-all stress testing regime less meaningful.
Geography is not the only difference supervisors must contend with. Risks affect different asset classes in different ways. For example, during recessions people buy fewer automobiles overall. Flowever, if a person needs a car during a recession, he is more likely to buy a used car. Thus, if default rates increase, loss given default (LGD) (i.e., loss severity) may not increase as much.
The business cycle also affects different industries at different times. Consider the airline industry versus the healthcare industry during a recession. Airplanes are collateral for loans to airlines. If the airline industry is depressed, the bank gets stuck with collateral that is very difficult to sell except at extremely depressed prices. Healthcare is somewhat recession-proof but that doesnt mean the bank can transform an airplane it is stuck with into a hospital. These factors increase the difficulty of mapping broader macro-factors to bank-specific stress results.
Modeling revenues over a stress test horizon period is much less developed than modeling losses. The 2009 SCAP did not offer much clarity on how to calculate revenue during times of market stress. The main approach to modeling revenue is to divide a banks total income into interest and non-interest income. The yield curve can be used to estimate interest income, and it can reflect credit spreads during stress testing scenarios; however, it remains unclear how bank profitability is directly influenced by the net impact of changing interest rates. Estimating noninterest income, which includes fees and service charges, is even more difficult to model. This is alarming given the steady increase in noninterest income among U.S. banks.
C h a l
l e n g e s i n M o d e l
i n g t h e B a l a n c e S h e e t
The typical stress test horizon is two years. Over this period, both the income statement and balance sheet must be modeled to determine if capital is adequate post-stress. Generally speaking, capital is measured as a ratio of capital to assets. There are different types of capital (e.g., Tier 1 and Tier 2) but in general (and for the sake of simplicity), capital can be
2018 Kaplan, Inc.
Page 251
Topic 56 Cross Reference to GARP Assigned Reading – Schuermann
defined as common equity. Risk-weighted assets (RWA) are computed based on the Basel II risk weight definitions. For example, agency securities have a lower risk weight than credit card loans.
In a stress model, the beginning balance sheet generates the first quarters income and loss from the stressed scenario, which in turn determines the quarter-end balance sheet. At that point, the person modeling the risk must consider if any assets will be sold or originated, if capital is depleted due to other actions such as acquisitions or conserved as the result of a spin-off, if there are changes made to dividend payments, if shares will be repurchased or issued (e.g., employee stock or stock option programs), and so on. These decisions make modeling the balance sheet over the stress horizon quite difficult. The stress model doesnt determine if it would be a good time to sell a subsidiary or lower dividend payments.
The challenges of balance sheet modeling exist under both static and dynamic modeling assumptions. The bank must maintain its capital (and liquidity) ratios during all quarters of the stress test horizon. At the end of the stress horizon the bank must estimate the reserves needed to cover losses on loans and leases for the next year. This means that a two-year horizon stress test is actually a three year stress test (i.e., aT-year stress test requires the bank to estimate required reserves to cover losses forT+1 years).
S t r e s s T e s t C o m p a r i s o n s
Disclosure was a significant feature of the 2009 SCAP. It disclosed projected losses for each of the 19 participating banks for eight asset classes. It also disclosed resources the bank had to absorb losses other than capital (e.g., pre-provision net revenue and reserve releases if available). This high level of disclosure created transparency. It allowed investors and the market to check the severity of stress tests and to comprehend stress test outcomes at the individual bank level. Before the 2009 SCAP, banks only reported realized losses, not forecasted losses (i.e., possible losses given the stress scenario).
The 2011 CCAR required only that macro-scenario results be published, not bank level results. This differed dramatically from the 2009 SCAP requirements. The market had to figure out whether a bank had passed the test or not (i.e., market participants had to do the math themselves). For example, if a bank increased its dividend, it was assumed by the market to have passed the stress test. However, the 2012 CCAR disclosed virtually the same amount and detail of bank level stress data as the 2009 SCAP (i.e., bank level loss rates and losses by major asset classes). The regulatory asset classes are: 1. First and second lien mortgages.
2. Commercial and industrial (C&I) loans.
3. Commercial real estate loans.
4. Credit card lending.
5. Other consumer loans.
6. Other loans.
Page 252
2018 Kaplan, Inc.
Topic 56 Cross Reference to GARP Assigned Reading – Schuermann
One of the key contributions of the CGAR was that in both 2011 and 2012 the CCAR required banks to submit the results of their own scenarios, both baseline and stress, not just supervisory stress test results. The Fed also reported dollar pre-provision net revenue (PPNR), gains and losses on available-for-sale and held-to-maturity securities, and trading and counterparty losses for the six institutions with the largest trading portfolios. These firms were required to conduct the trading book stress test. The numbers that were reported were supervisory estimates, not bank estimates, of losses under the stress scenario.
In contrast, the 2011 European Banking Authority (EBA) Irish and 2011 EBA European wide stress tests, both disclosed after the CCAR, contained considerable detail. In the Irish case, the report contained a comparison of bank and third party estimates of losses. The EBA data was available in electronic, downloadable form. Ireland needed credibility, having passed the Committee of European Bank Supervisors (CEBS) stress test in July 2010 only to need considerable aid four months later. In general, the faith in European supervisors was harmed and only by disclosing detailed information on bank-by-bank, asset-class, country, and maturity bucket basis could the market interpret the data and draw its own conclusions about individual bank risks. Figure 2 summarizes the differences among the various stress test regimes.
2018 Kaplan, Inc.
Page 253
Topic 56 Cross Reference to GARP Assigned Reading – Schuermann
Figure 2: Comparison of Macro-prudential Stress Tests
Stress Test SCAP (2009). All banks with $100 billion or more in assets as of 2008 year end were included. bank level projected required to raise losses and asset/ product level loss rates. six months. The Disclosure Methodologies Tested simple scenarios with First to provide three dimensions, GDP growth, unemployment, and the house price index (HPI). Historical experience was used for the market risk scenario (i.e., the financial crisisa period of flight to safety, the failure of Lehman, and higher risk premia). A one-size-fits-all approach. Findings 19 SCAP banks were six months. The$75 billion within undercapitalized banks actually raised \$77 billion of Tier 1 common equity and none of the banks were forced to use the Treasurys Capital Assistance Program funds.
CCAR (2011)
CCAR (2012)
In recognition of one-size- fits-all stress testing, CCAR asked banks to submit results from their own baseline and stress scenarios. Banks were again asked to submit their own baseline and stress test results.
EBA Irish (2011)
Similar in design to EBA Europe 2011.
Only macro scenario results were published.
Similar in detail to SCAP 2009 bank level and asset/ product level loss rates disclosed. Comparison of bank and third party proj ected losses; comparison of exposures by asset class and geography. Data is electronic and downloadable.
EBA Europe (2011). [formerly the Committee of European Bank Supervisors (CEBS)] 90 European banks were stress tested.
Specified eight macro-factors Bank level (GDP growth, inflation, projected losses. unemployment, commercial Comparisons of exposures by asset and residential real estate price indices, short and class and geography. long-term government rates, Data is electronic and downloadable. and stock prices) for each of 21 countries. Specified over 70 risk factors for the trading book. It also imposed sovereign haircuts across seven maturity buckets.
After passing the 2010 stress tests, 2011 stress tests revealed Irish banks needed 24 billion. Greater disclosure in 2011 resulted in tightening credit spreads on Irish sovereign and individual bank debt. Eight banks were required to raise 2.5 billion.
Page 254
2018 Kaplan, Inc.
Topic 56 Cross Reference to GARP Assigned Reading – Schuermann
The key benefit of greater disclosure is transparency. Transparency is especially important in times of financial distress. However, during normal times, the costs of disclosure may outweigh the benefits. For example, banks may window dress portfolios, making poor long-term investment decisions to increase the likelihood of passing the test. Traders may place too much weight on the public information included in stress test disclosure and be disincentivized to produce private information about financial institutions. This harms the information content of market prices and makes prices less useful to regulators making policy decisions.
One thing to note is that prior to the CCAil 2011 requirements, all supervisory stress tests imposed the same scenarios on all banks (i.e., a one-size-fits-all approach to stress testing). In recognition of the problem, the 2011 and 2012 CCAR asked banks to submit results from their own scenarios in addition to the supervisory stress scenario in an attempt to reveal bank-specific vulnerabilities.
2018 Kaplan, Inc.
Page 255
Topic 56 Cross Reference to GARP Assigned Reading – Schuermann
Ke y C o n c e pt s
LO 56.1 After the 20072009 financial crisis, it was clear that traditional risk measures such as regulatory capital ratios were insufficient. Supervisory stress-testing became an important risk-assessment tool at that point.
The goal of stress testing is to assess how much capital and liquidity a financial institution needs to support its business (i.e., risk taking) activities.
The 2009 U.S. bank stress test, known as the Supervisory Capital Assessment Program (SCAP), was the first macro-prudential stress test after the 20072009 financial crisis.
Disclosure was a significant feature of the 2009 SCAP. This high level of disclosure lead to transparency and allowed investors and the market the ability to check the severity of the stress tests and the outcomes of the stress at the individual bank level.
In 2011, CCAR required only macro-scenario results be published, not bank level results, differing significantly from the 2009 SCAP requirements. The 2012 CCAR disclosed virtually the same amount and detail of bank level stress data as the 2009 SCAP. The EBA Irish and the EBA Europe required significant disclosures as well. The disclosures were needed to increase trust in the European banking system.
LO 56.2 One of the challenges regulators face is designing coherent stress tests. The sensitivities and scenarios must be extreme but must also be reasonable and possible (i.e., coherent). Problems are inherently multi-factor, making it more difficult to design a coherent stress test.
LO 56.3 Current stress tests are based on macro-scenarios (i.e., unemployment, GDP growth, the HPI). One concern is how to translate the macro-risk factors employed in stress tests into micro (i.e., bank specific) outcomes related to revenues and losses. Supervisors must be able to map from macro-factors into intermediate risk factors that drive losses in specific products and geographic areas.
In a stress model, the starting balance sheet generates the first quarters income and loss from the stressed scenario, which in turn determines the quarter-end balance sheet. The bank must maintain its capital (and liquidity) ratios during all quarters of the stress test horizon, typically two years.
Page 256
2018 Kaplan, Inc.
Topic 56 Cross Reference to GARP Assigned Reading – Schuermann
C o n c e pt C h e c k e r s
1.
2.
3.
4.
5.
Which of the following changes in stress testing was not the result of the 2009 Supervisory Capital Assessment Program (SCAP)? A. Banks are now required to provide the results of their own scenario stress tests. B. Stress scenarios are now broader in nature. C. Stress testing now focuses on the whole firm. D. Stress testing now focuses on revenues, costs, and losses.
Piper Hook, a bank examiner, is trying to make sense of stress tests done by one of the banks she examines. The stress tests are multi-factored and complex. The bank is using multiple extreme scenarios to test capital adequacy, making it difficult for Hook to interpret the results. One of the key stress test design challenges that Hook must deal with in her examination of stress tests is: A. multiplicity. B. efficiency. C. coherence. D. efficacy.
Greg Nugent, a regulator with the Office of the Comptroller of the Currency, is presenting research on stress tests to a group of regulators. He is explaining that macro-variable stress testing can be misleading for some banks because of geographical differences in macro risk factors. He gives the example of the wide range of unemployment rates across the U.S. following the 20072009 financial crisis. Which type of loan did Nugent most likely identify as having losses tied to unemployment rates? A. Residential real estate loans. B. Credit card loans. C. Commercial real estate loans. D. Industrial term loans.
A risk modeler has to make assumptions about acquisitions and spinoffs, if dividend payments will change, and if the bank will buy back stock or issue stock options to employees. These factors make it especially challenging to: A. get a CAMELS rating of 2 or better. B. determine if the bank has enough liquidity to meet its obligations. C. meet the Tier 1 equity capital to risk-weighted assets ratio. D. model a banks balance sheet over a stress test horizon.
One of the key differences between the 2011 CCAR stress test and the 2011 EBA Irish stress test is that: A. the CCAR did not require banks to provide results from their own stress the CCAR did not require banks to provide results from their own stress scenarios. the EBA Irish did not find any banks in violation of capital adequacy requirements.
B.
C. the CCAR required disclosure of macro-level, not bank level, scenario results. D. the EBA Irish allowed for 1-year stress horizons.
2018 Kaplan, Inc.
Page 257
Topic 56 Cross Reference to GARP Assigned Reading – Schuermann
C o n c e pt Ch e c k e r A n s w e r s
1. A The 2009 U.S. bank stress test, known as the Supervisory Capital Assessment Program (SCAP), was the first macro-prudential stress test after the 2007-2009 financial crisis.
The 2011 CCAR, not the 2009 SCAP, required that banks provide results of their own stress scenarios along with supervisory stress scenarios.
2. C One of the challenges of designing useful stress tests is coherence. The sensitivities and
scenarios must be extreme but must also be reasonable or possible (i.e., coherent). Problems are inherently multi-factored, making it more difficult to design a coherent stress test. Hook is dealing with the possibly incoherent results of the banks stress tests.
3. B Credit card losses are particularly sensitive to unemployment figures. For example,
unemployment was 12.9% in Nevada in July 2011, 3.3% in North Dakota, and the national unemployment rate was 9.1%. Credit card loss rates varied dramatically from region to region during this period. Residential mortgages are affected by unemployment as well but people are generally more likely to quit paying credit card bills before mortgages.
4. D
In a stress model, the starting balance sheet generates the first quarters income and loss from the stressed scenario, which in turn determines the quarter-end balance sheet. At that point the person modeling the risk must consider if any assets will be sold or originated, if capital is depleted due to other actions such as acquisitions or conserved as the result of a spin off, if there are changes made to dividend payments, if shares will be repurchased or issued (e.g., employee stock or stock option programs), and so on. This makes it challenging to model the balance sheet over the stress horizon.
5. C The 2011 CCAR required banks to provide results from their own stress scenarios but the EBA Irish did not. After the 2011 EBA Irish tests, 24 billion was required to increase the capital of several banks. The 2011 CCAR, unlike the SCAP and the 2012 CCAR, only required the disclosure of macro-level scenario results. The EBA Irish did not change the stress horizon from two years to one year.
Page 258
2018 Kaplan, Inc.
The following is a review of the Operational and Integrated Risk Management principles designed to address the learning objectives set forth by GARP. This topic is also covered in:
G u i d a n c e o n Ma n a g i n g O u t s o u r c i n g Ri s k
Topic 57
E x a m F o c u s
This short and nontechnical topic begins by examining the general risks arising from a financial institutions use of service providers. It then provides details on the key elements of an effective service provider risk management program. For the exam, focus on the three broad areas of due diligence. Also, be familiar with the details from the numerous contract provisions that should be addressed with third-party service providers.