Moodys Analytics Risk Practitioner Conference 2014 Quantitative - - PowerPoint PPT Presentation

SLIDE 1

Moody’s Analytics Risk Practitioner Conference ‐ 2014

Quantitative Modeling Approaches for Mid‐size Institutions

Thomas L Thomas Quantitative Portfolio Manager City National Bank

SLIDE 2

Development of Commercial Stress Testing Process

1/20/2015

2

CNB Idiosyncratic Scenarios – Under Development CRE PPR Model Investor Properties – tested

Owner Occupied Properties – Under Development Land & Construction – Under Development

Macroeconomic Scenario – “Fed Test " Bottom Up Industry Segment Macro Scenario – Moody’s Top Down Macro Scenario CSUN Scenario Analysis Specific Shock Testing “Breakage Points” ‐General Sensitivity Analysis (Shock Testing) in CreditManager Migration Analysis Spreadsheet +/‐ PD % (TARP Stress Testing) Credit Manager Worst Case Transition Matrix

We chose to take an incremental approach:

• 1. There are many stress

testing techniques including:

• Shock Testing
• Proxy Analysis
• Scenario Analysis
• Econometric Modeling
• 2. Staring with simpler models

first and then moving to more complex econometric models

• Helps understand your data

and loss behavior

• Sets parameters (boundaries'

for later modeling)

• Acts as a challenge and

reasonableness check

SLIDE 3

Important to Map DFAST Process Flow

3

As the graphic above illustrates, the credit capital stress testing modeling process consists of the following three main steps:

• Data extraction and preparation
• Applying stress test models
• Export results to Fed reporting templates

Helps understand potential data issues and explain the process and data flow to both regulators as well as senior management.

SLIDE 4

Econometric Approach ‐ Regression Analysis

• Econometrics is a branch of economics that uses the “science and art of building models which

utilizes a set of quantitative tools to construct and test mathematical representations of the real world.”

• More specifically the branch attempts to unify economics with math and statistics to make

forecasts and empirically test economic theory.

• While the analyst may employ several different mathematical and statistical techniques to build

his or her econometric model, regression is often the starting point.

• The regression is represented through a modification of the formula for a straight line by adding

an error term (e): = a+B1X1+ e

SLIDE 5

Regression Approaches

Two general regression approaches: Introspective Counterfactual Queries and Prospective Counterfactual Queries.

• Introspective Counterfactual Queries ‐ The goal in this case is to determine the

values that are driving the dependent variable or imply casual relationships.

• Here statistical methods are used to estimate the strength of the causal relationships.
• The independent variables are said to be efficient when exhibiting good fit via statistical

measures like the adjusted R‐squared and statistically significant “F” and “t” tests.

• Prospective Counterfactual Queries ‐ Take the form what value would the

dependent variable take if the independent variable X were set to a specific value. Such models are simply conditional expectation models that are more in the line of traditional curve fitting techniques.

• If the question we are trying to answer is one of conditional expectations, and not

causal, then we do not have to be as concerned whether other causal factors are present.

• Measurements of goodness of fit, like R‐squared, Adjusted R‐squared, F and t statistics

are sufficient to estimate the best fit line given a specific value of “X.

• Due to feedback issues in the independent variables – we believe a Prospective

Counterfactual or “curve fitting” approach is best.

SLIDE 6

Issues to Consider

• Multicollinearity –
• In causal models is an implicit assumption in many regression models that the

relationship between the independent (exogenous regression variables) are assumed to be independent.

• In addition, the relationship between the independent variables and errors

between predicted values and the actual values are also assumed to be independent.

• These can be checked by looking at the sings in the regression coefficients and

through test like variance inflation factor test (VIF) to diagnose potential multicollinear variables.

• Stationarity –
• That is the independent variables do not change dramatically over time thereby

skewing the regression forecast.

• Plotting trends and using histograms to view the independent variable’s

distribution help identify if the data is stationary.

• If the data is not stationary – utilizing transposition techniques like “Tuckey’s

Ladder” can be used – the simplest is using the period‐over‐period change in the data for your regression.

SLIDE 7

Choosing the Appropriate Horizon

• It is important to choose the correct time horizon to model the behavior you are

interested in examining.

• The model above illustrates a regression analysis of predicted charge‐off rates using

Federal Reserve data to proxy the impact to LGD during a stress event.

• Note the first analysis includes a long‐time horizon incorporating an expansionary

period between 2000 and 2005. This pulls the regression downward as a result the actual charge‐off rates exceed the regressions upper 95% confidence level.

• When we only model the stressed recessionary period – the actual charge off

rates and forecast now fit within our confidence level and mimic the correct expected behavior.

SLIDE 8

Picking Independent Variables

Comm. Delinquency Nominal GDP Growth Real Disposable Income Growth Unemployment Rate CPI Inflation Rate 3-Month Treasury Yield 10-Year Treasury Yield BBB Corporate Yield Dow Jones Total Stock Market Index Market Volatility Index (VIX) House Price Index Commercial Real Estate Price Index Ln 3-Month Treasury Yield Ln 10-Year Treasury Yield Ln BBB Corporate Yield

• Comm. Delinquency

1 Nominal GDP Growth

• 0.3433

1 Real Disposable Income Growth

• 0.3306

0.4913 1 Unemployment Rate 0.9555

• 0.19
• 0.3141

1 CPI Inflation Rate

• 0.2171

0.5125 0.3789

• 0.1835

1 3-Month Treasury Yield

• 0.6818

0.2487 0.2423

• 0.7253

0.1623 1 10-Year Treasury Yield

• 0.7184

0.1802 0.3042

• 0.7955

0.1258 0.6355 1 BBB Corporate Yield 0.0195

• 0.7384
• 0.3239
• 0.145
• 0.465
• 0.0165

0.3292 1 Dow Jones Total Stock Market Index

• 0.0962

0.3358 0.2813

• 0.1503

0.411 0.342 0.0198

• 0.4909

1 Market Volatility Index (VIX) 0.4638

• 0.8034
• 0.5257

0.3412

• 0.5521
• 0.4333
• 0.3286

0.6681

• 0.5194

1 House Price Index

• 0.4465

0.2791 0.2096

• 0.4167

0.2098 0.4702 0.1687

• 0.2929

0.6338

• 0.5115

1 Commercial Real Estate Price Index

• 0.0423
• 0.1383

0.0235

• 0.1451

0.1392 0.235

• 0.1253
• 0.0576

0.7393

• 0.0585

0.7206 1 Ln 3-Month Treasury Yield

• 0.8974

0.219 0.313

• 0.9431

0.1868 0.8337 0.8313 0.1412 0.1587

• 0.4135

0.4425 0.1316 1 Ln 10-Year Treasury Yield

• 0.7054

0.1626 0.3002

• 0.7822

0.1224 0.6156 0.9901 0.3322 0.0125

• 0.3351

0.2018

• 0.1078

0.8318 1 Ln BBB Corporate Yield

• 0.0207
• 0.695
• 0.2822
• 0.1893
• 0.4156

0.0247 0.389 0.9952

• 0.4733

0.6251

• 0.2745
• 0.0554

0.1946 0.3929 1

Correlation Matrix

• We chose to utilize only the Fed Scenario Variables – This kept it more manageable

and reduced the need for additional forecasts.

• Use common sense when picking variables. For example if you are modeling

commercial loan defaults, one would expect the house price index to add little value. Indeed, the housing index exhibited a low correlation 45% compared to 10‐year treasury which exhibited 72% correlation.

• Use a correlation matrix (something easily created in Microsoft Excel) to help choose

relevant variables and confirm which variables to exclude.

 Keep in mid the signs to make sure the are institutively correct with expected economic behavior.

SLIDE 9

Example Stepwise Regression

Commercial Factors

R2 3‐Month Treasury 68.39% 10‐Year Treasury 58.24% VIX Index Only 26.78% Nominal GDP Growth 14.66% CPI Inflation Rate 10.24% Dow Jones Total Index Only 4.89% BBB Corporate Yield 3.35%

Consumer Factors

R2 Unemployment Only 93.56% Real Disposable Income Only 14.90%

Multi‐Factor Regression

R2 Change Change to Base

3‐Month, 10‐Year Treasury (BASE) 70.70% 3‐Month, 10‐Year Treasury, VIX 72.28% 1.58% 1.58% 3‐Month, 10‐Year Treasury, VIX,GDP 71.87% ‐0.40% 1.18% 3‐Month, 10‐Year Treasury, VIX,GDP, DowJones 74.70% 2.82% 4.00% 3‐Month, 10‐Year Treasury, VIX,GDP, DowJones,BBB 74.77% 0.07% 4.07% 3‐Month, 10‐Year Treasury, VIX,GDP, DowJones,BBB, Comml RE 79.38% 4.62% 8.68% 3‐Month, 10‐Year Treasury, VIX, Unemployment 96.33% 16.95% 25.64% 3‐Month, 10‐Year Treasury, VIX, Unemployment, Dow Jones 97.27% 0.93% 26.57% 10‐Year Treasury, VIX, Unemployment, Dow Jones 97.34% 0.07% 26.64% All Factors 97.97% 1.64% 27.27%

This is an example of stepwise regression utilizing Federal Reserve delinquency data as a proxy for PD behavior

• Note in this process we first begin by regressing each of the individual “Fed”

scenario factors against delinquencies.

• We then go through a process of combining various factors to obtain the “Best

Fitting Regression Line.”

SLIDE 10

Testing the Results Scenario Change

20,000,000 40,000,000 60,000,000 80,000,000 100,000,000 120,000,000 140,000,000 160,000,000 180,000,000 200,000,000 Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8 Q9

<‐‐‐‐Loss $$ ‐‐‐‐>

Moody's Test Portfolio

2012Stress 2013Severe 2014Severe

18% ‐28% ‐18%

• When picking a vendor model – make sure you understand the impact from changing variables.
• The analysis above shows the expected loss results from changes to the Fed Variables for the severely

adverse case.

• The graph above highlights the results from Moody’s Test Case Portfolio for each of the Severely Adverse

Scenarios (Note in 2012 there was only a base case and stress case scenario provided by the Federal Reserve).

• The credit losses estimated by the RiskCalc Plus model were 18% lower for 2014 stress test compared to

the 2013 stress test and 28% lower when compared to the 2012 stress case.

• The results suggest that in the 2014 Severely Adverse Scenario the economic factors are similar to the

economic factors in the 2013 Severely Adverse stress test, but recover much more quickly over the last 5 quarters of the stress test.

10

SLIDE 11

Macro Factor Comparison

• When we compare the macroeconomic variables utilized by RiskCalc Plus, we observe a
• verall improvement between the 2013 and 2014 Severely Adverse Scenarios.
• One would expect then to see some reduction in forecast losses.
• For example, the Unemployment rate is substantially lower in the 2014 stress test

suggesting fewer firms are expected to default thereby reducing the number of people unemployed.

• A similar reasoning can be used for all of the above factors.

8 8.5 9 9.5 10 10.5 11 11.5 12 12.5 F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 F13

Unemployment

2013 2014 0.5 1 1.5 2 2.5 F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 F13

10Y Treasury

2013 2014 3 3.5 4 4.5 5 5.5 6 6.5 7 F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 F13

BBB Corporate

2013 2014 6000 8000 10000 12000 14000 16000 F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 F13

DowJones Index

2013 2014 10 20 30 40 50 60 70 80 90 F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 F13

VIX

2013 2014 3.00 3.50 4.00 4.50 5.00 5.50 6.00 F1 F2 F3 F4 F5 F6 F7 F8 F9 F10 F11 F12 F13

BBB 10Yr Treasury Spread

2013 2014

11

SLIDE 12

Understand the impact to internal variables ‐ CNB Balance Sheet Mix Example

9/30/2008 9/30/2009 9/30/2010 9/30/2013

LOAN OUTSTANDING TOTALS LOAN OUTSTANDING TOTALS LOAN OUTSTANDING TOTALS LOAN OUTSTANDING TOTALS

1 34,275.5 42,304.1 38,961.4 90,022.4 47,718.3 2 726,686.5 785,059.3 708,929.8 1,794,998.4 1,009,939.1 3 893,349.7 1,019,598.1 806,631.2 1,360,747.4 341,149.4 4 1,982,097.6 1,536,847.1 1,242,879.8 2,698,935.4 1,162,088.2 5 2,462,053.8 1,785,202.8 1,773,974.3 3,269,869.0 1,484,666.2 6 1,262,103.9 1,106,293.5 1,277,892.1 1,406,674.5 300,381.1 7 392,855.2 437,795.9 255,081.8 109,041.8 (328,754.1) 8 389,012.7 1,023,939.5 849,355.2 342,189.2 (681,750.3) 9 81,364.7 101,476.4 25,743.5 2,732.1 (98,744.3) PG 4,007,607.6 4,288,463.1 4,392,936.0 5,265,791.3 977,328.2 Not rated** 47,011.8 41,510.5 46,157.1 225,131.3 183,620.8 Total $12,278,419.1 $12,168,490.2 $11,418,542.4 $16,566,132.8 $4,397,642.5 9/30/2008 9/30/2009 9/30/2010 9/30/2013

LOAN % TOTALS LOAN % TOTALS LOAN % TOTALS LOAN % TOTALS

1 0.28% 0.35% 0.34% 0.54% 0.20% 2 5.92% 6.45% 6.21% 10.84% 4.38% 3 7.28% 8.38% 7.06% 8.21%

• 0.16%

4 16.14% 12.63% 10.88% 16.29% 3.66% 5 20.05% 14.67% 15.54% 19.74% 5.07% 6 10.28% 9.09% 11.19% 8.49%

• 0.60%

7 3.20% 3.60% 2.23% 0.66%

• 2.94%

8 3.17% 8.41% 7.44% 2.07%

• 6.35%

9 0.66% 0.83% 0.23% 0.02%

• 0.82%

PG 32.64% 35.24% 38.47% 31.79%

• 3.46%

Not rated** 0.38% 0.34% 0.40% 1.36% 1.02% Total 100.00% 100.00% 100.00% 100.00% 0.00%

Difference between Peak Loss Exposure 2009 and 2013 Difference between Peak Loss Exposure 2009 and 2013

Risk Rating Risk Rating 1 0.35% 2 6.45% 3 8.38% 4 12.63% 5 14.67% 6 9.09% 7 3.60% 8 8.41% 9 0.83% PG 35.24% Not rated** 0.34%

Outstanding Balances by Risk Grade September 2009

1 0.54% 2 10.84% 3 8.21% 4 16.29% 5 19.74% 6 8.49% 7 0.66% 8 2.07% 9 0.02% PG 31.79% Not rated** 1.36%

Outstanding Balances by Risk Grade September 2013

• The Severely Adverse stress test loss rates (2.95% vs. 2.40%) are approximately 19% lower that what CNB experienced over

the recent economic downturn.

• CNB’s current loan portfolio balance mix is a major contributing factor to the expected decline in credit losses.
• The above tables and pie charts compare CNB’s book balances between our worst year 2009 and the most recent quarter‐

end used in our stress Test (September 2013).

• Note that CNB has experienced a significant improvement in credit quality in grades 6,7,8,9 compared to the period just

preceding 2009‐2010. 12

SLIDE 13

Use External Comparisons for Reasonable Testing ‐ Loss Scalars Example

First Lien Mortgages 2.35% 6.05% 2.58 0.12% 0.30% 2.48 ELC and Lien Mortgages 3.48% 9.40% 2.70 0.81% 3.88% 4.82 Commercial & Industrial Loans 3.29% 6.50% 1.98 3.53% 3.37% 0.95 Commercial Real Estate Loans 5.94% 7.85% 1.32 5.33% 2.80% 0.53 Credit Cards 13.27% 15.85% 1.19 4.22% 8.78% 2.08 Other Consumer 3.65% 4.10% 1.12 3.78% 4.94% 1.31 Average 1.82 Average 2.03

Loss Scalar Analysis ‐ Actual Losses Vs. Stress Test Loss Rates

CNB 2014 Severly Adverse Case Loss Scalar ASSET CLASSIFICATION Median 2‐ Year Worst Case Loss CCAR Banks 2009‐2010 2013 CCAR Median Average Loss Rates Loss Scalar CNB Loss Rate 2009‐2010

• One way to measure the reasonableness of our loss projection is to compare our loss rates to those of the CCAR Stress Test.
• In this comparison we would expect our loss rates to be similar in scale to those experienced by the CCAR banks that have been

performing macro economic stress testing over the last several years and have much more modeling experience.

• Note on average the CCAR Stress Test Loss rates are generally higher than what the CCAR banks experienced over the stress period

between 2009 – 2010. This suggest the CCAR stress scenarios are more severe than what was experience in the recent economic downturn.

• Our Severely Adverse Stress Test scenario suggest we may experience losses that are on average 2 times greater than what CNB

experienced between 2009‐2010 compared to 1.8 times for the CCAR banks.

• The reduced loss scalar for CRE is primarily attributable to the reduced concentration of ADC loans
• This result suggest our Severely Adverse credit loss forecast is reasonably scaled up for such a severe economic event.

13

SLIDE 14

Impact Analysis

• One way to estimate the impact of the balance sheet mix and the changing scenarios is to run the

period‐end balance sheet through the Moody’s model for the same historical time period and same stress scenario and measure the differences.

• This acts like a controlled experiment where we change one variable – in this case balances or scenario,

hold everything else, constant and then measure the impact.

• To keep as many variables constant as possible, we utilized the same PDs and LGDs in each test.
• In addition, we utilized the 2013 Severely Adverse Scenario because we began this test before the Fed

released the 2014 scenarios.

• We also tested at a total portfolio level since GL and Fed Class code roll‐ups were different between the

two periods.

Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8 Q9 Total 2 Yr Loss Rate Delta 2009 Historical (2009 Balances) 44,022 37,418 33,179 33,023 29,833 26,027 22,802 21,601 23,917 271,822 2.23% 2009 Historical (2013 Balances) 33,517 28,476 25,482 26,065 23,839 20,949 18,533 17,871 20,261 214,992 1.77% ‐20.91% Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8 Q9 Total 2 Yr Loss Rate Delta CCARSevere2013 (2009 Balances) 50,136 60,849 73,421 80,705 85,746 83,042 75,955 67,066 57,403 634,324 5.21% CCARSevere2013 (2013 Balances) 40,597 54,791 68,400 80,178 83,892 78,333 69,069 59,101 49,522 583,883 1.77% ‐7.95% Balance Issue Risk Rating

14

SLIDE 15

Challenger Model

• Another way to examine the reasonableness of the stress test results is to compare the

Moody’s results against our CSUN Challenger Model.

• The CSUN Challenger Model develops PD scalars base on Federal Reserve delinquency
• data. 1
• While the quarterly results are somewhat different since the two models employ

different modeling variables, the forecast credit losses and loss rate are generally within the same overall range.

15

SLIDE 16

Graphical Comparison Severely Adverse Scenario

2.00% 3.00% 4.00% 5.00% 6.00% 7.00% 8.00% 9.00% 2012Q4 2013Q1 2013Q2 2013Q3 2013Q4 2014Q1 2014Q2 2014Q3 2014Q4

Unemployment

StressPDUnEmp+1STD BaseStress StressPDUnEmp‐1STD 3.00% 3.50% 4.00% 4.50% 5.00% 5.50% 6.00% 6.50% 7.00% 7.50% 2012Q4 2013Q1 2013Q2 2013Q3 2013Q4 2014Q1 2014Q2 2014Q3 2014Q4

10‐Year Treasury

StressPD10Y+1STD BaseStress StressPD10Y‐1STD 3.00% 3.50% 4.00% 4.50% 5.00% 5.50% 6.00% 6.50% 7.00% 7.50% 2012Q4 2013Q1 2013Q2 2013Q3 2013Q4 2014Q1 2014Q2 2014Q3 2014Q4

BBB

StressPDBBB+1STD BaseStress StressPDBBB‐1STD 3.00% 4.00% 5.00% 6.00% 7.00% 8.00% 2012Q4 2013Q1 2013Q2 2013Q3 2013Q4 2014Q1 2014Q2 2014Q3 2014Q4

DJX

StressPDDJX+1STD BaseStress StressPDDJX‐1STD 2.50% 3.50% 4.50% 5.50% 6.50% 7.50% 2012Q4 2013Q1 2013Q2 2013Q3 2013Q4 2014Q1 2014Q2 2014Q3 2014Q4

Vixhigh

StressPDVixhigh+1STD BaseStress StressPDVixhigh‐1STD

• The graphs above illustrated the results to a +/‐ one standard deviation change in each individual macro variable.
• Unlike the Base Case which exhibits generally linear downward trends, as one would expect the resulting PDs

rapidly increase in the first part of the stress scenario, peak around the mid part, and then decline during the remaining forecast quarters.

• All factors exhibit a doubling in absolute PD rates as the stress is increased; however, the volatility of the spread to

a one standard deviation change is greater for some of the factors, such as unemployment and DJX.

• The model also exhibits a more stable relationship for the 10‐year Treasury / BBB Credit Spread the VIX index in

that the volatility around these measures narrows slightly compared to the base case.

16

SLIDE 17

Sensitivity Analysis

• We also conducted a Sensitivity Analysis in order to analyze the model’s results

and determine if the model results were directionally consistent with our expectation for a given change in a macroeconomic variable input.

• Like the impact analysis above we conducted a controlled experiment where we

changed one macro variable one standard deviation up or down while holding all the other independent variable in the model constant.

• The independent variables we tested were the independent macro variables

employed in the vendor model. They include:

• Unemployment Rate
• BBB Bond Rate
• 10 – Year Treasury Rate
• VIX Index (VIXhigh)
• Dow Jones Index (DJX)

17

SLIDE 18

• Although the BBB bond rate is a macroeconomic variable employed in the Moody’s model, the Moody’s model actually

measures the “credit spread” between the BBB bond rate and the 10‐year treasury rate. 2

• If the bond rate increases relative to 10‐Year Treasury, the credit spread widens suggesting the market is pricing for perceived

increased credit risk. Hence we would expect an increase in PDs.

• Conversely, if BBB decreases relative to 10‐Year Treasury, the perceived risks are lower and PDs decline.
• As one would expect a one standard deviation increase in the BBB rate results in an increased PD forecast by the Moody’s

model.

• However, the results between a one standard deviation change upward Vs. a one standard deviation downward in the BBB rate

are asymmetrical – The model exhibits a larger change in PDs for a one standard deviation decrease that by a one standard deviation increase by roughly 0.59 times.

• This suggests that in the Moody’s Model, BBB market rates are slightly more sensitive to downward default rates and re‐

pricing the credit spread then when defaults increase.

1.50% 2.00% 2.50% 3.00% 2012Q4 2013Q1 2013Q2 2013Q3 2013Q4 2014Q1 2014Q2 2014Q3 2014Q4

BBB

StressPDBBB+1STD BaseStress StressPDBBB‐1STD

2012Q4 2013Q1 2013Q2 2013Q3 2013Q4 2014Q1 2014Q2 2014Q3 2014Q4 StressPDBBB+1STD 2.90% 2.78% 2.63% 2.51% 2.38% 2.25% 2.15% 2.05% 1.94% BaseStress 2.79% 2.62% 2.47% 2.35% 2.23% 2.10% 2.00% 1.91% 1.81% StressPDBBB‐1STD 2.62% 2.36% 2.21% 2.10% 1.97% 1.83% 1.74% 1.66% 1.55%

18

Sensitivity Analysis ‐ BBB Bond Rate Example

SLIDE 19

Graphical Comparison Severely Adverse Scenario

2.00% 3.00% 4.00% 5.00% 6.00% 7.00% 8.00% 9.00% 2012Q4 2013Q1 2013Q2 2013Q3 2013Q4 2014Q1 2014Q2 2014Q3 2014Q4

Unemployment

StressPDUnEmp+1STD BaseStress StressPDUnEmp‐1STD 3.00% 3.50% 4.00% 4.50% 5.00% 5.50% 6.00% 6.50% 7.00% 7.50% 2012Q4 2013Q1 2013Q2 2013Q3 2013Q4 2014Q1 2014Q2 2014Q3 2014Q4

10‐Year Treasury

StressPD10Y+1STD BaseStress StressPD10Y‐1STD 3.00% 3.50% 4.00% 4.50% 5.00% 5.50% 6.00% 6.50% 7.00% 7.50% 2012Q4 2013Q1 2013Q2 2013Q3 2013Q4 2014Q1 2014Q2 2014Q3 2014Q4

BBB

StressPDBBB+1STD BaseStress StressPDBBB‐1STD 3.00% 4.00% 5.00% 6.00% 7.00% 8.00% 2012Q4 2013Q1 2013Q2 2013Q3 2013Q4 2014Q1 2014Q2 2014Q3 2014Q4

DJX

StressPDDJX+1STD BaseStress StressPDDJX‐1STD 2.50% 3.50% 4.50% 5.50% 6.50% 7.50% 2012Q4 2013Q1 2013Q2 2013Q3 2013Q4 2014Q1 2014Q2 2014Q3 2014Q4

Vixhigh

StressPDVixhigh+1STD BaseStress StressPDVixhigh‐1STD

• The graphs above illustrated the results to a +/‐ one standard deviation change in each individual macro variable.
• Unlike the Base Case which exhibits generally linear downward trends, as one would expect the resulting PDs

rapidly increase in the first part of the stress scenario, peak around the mid part, and then decline during the remaining forecast quarters.

• All factors exhibit a doubling in absolute PD rates as the stress is increased; however, the volatility of the spread to

a one standard deviation change is greater for some of the factors, such as unemployment and DJX.

• The model also exhibits a more stable relationship for the 10‐year Treasury / BBB Credit Spread the VIX index in

that the volatility around these measures narrows slightly compared to the base case.

19