October 21, 2014 Rick Hamilton 1 DISCLAIMER THE VIEWS EXPRESSED - - PowerPoint PPT Presentation

1

Economic Capital in the Age of CCAR October 21, 2014

Rick Hamilton

2

DISCLAIMER

THE VIEWS EXPRESSED IN THIS PRESENATATION ARE THOSE OF THE SPEAKER AND NOT NECESSARILY THOSE OF PNC FINANCIAL SERVICES GROUP, INC. OR ITS SUBSIDIARIES.

3

Agenda

• Defining the Risk Measures
• Information Content
• The Problems They Solve
• Reconciling The Stories

4

Agenda

• Defining the Risk Measures
• Information Content
• The Problems They Solve
• Reconciling The Stories

5

Typical Risk Quantification Process: Capital Management

Capital Requirements (x 12.5) Forecasted Capital Ratios Capital Resources New Volume Runoff Credit Migration Credit Losses Operational Losses Market Risk Losses Margin & Volume Spot Capital Ratios Forecast Assumptions Liquidity Impacts Tier 1 C Total RBC Tier 1 Leverage Standardized Pillar 1 Economic Cap. Tier 1 C Total RBC Tier 1 Leverage Standardized Pillar 1 Economic Cap. Idiosyncratic Quantitative Capital Adequacy Assessment CCAR Submission DFAST ICAAP Pricing & Performance

• Avail. Cap.
• Avail. Cap.

Macro Economic Scenarios

6

Measures of Capital Resources

Capital Requirements Precision/Sophistication Tier 1 Common Most conservative regulatory view

• f resources available to absorb loss

Tier 1 Somewhat broader regulatory definition of capital resources Total Risk Based Capital Broadest regulatory view of capital resources Internal Available Capital Firm’s own assessment of the resources available to absorb loss

7

Measures of Capital Requirements (Tail Risk)

Capital Requirements Precision/Sophistication Leverage Ratio Simplest view - assumes all assets have same risk Standardized Approach More detailed view – some differentiation by asset class. Basel II Pillar 1 Very granular view that relies upon a number of simplifying assumptions Economic Capital Very granular and precise estimate designed to reflect specific risk characteristics of the firm.

8

Forecasting Assumptions: Goals

• Project the firm’s capital ratios under

alternative, (typically severe) scenarios

• Scenarios are deterministic
• Scenarios can have multiple sources

Executive management/board driven Regulatory driven LOB driven

• There are key assumptions used in

forecasting that align with the

9

Agenda

• Defining the Risk Measures
• Information Content
• The Problems They Solve
• Reconciling The Stories

10

Spot Capital Ratios

Economic Capital Group

• A measure of the amount of potential loss a

firm could experience as a ratio of its available capital.

• Looks to ensure that the firm has sufficient

resources to absorb catastrophic losses.

• Primarily a gone concern perspective if one

assumes:

 Banks are not viable when they fall below their

minimum capital ratios.

 The losses implied by the capital calculations

reflect a market to market

11

Basel III Capital Ratios

• 100% of CET1 would be used by implied losses
• Tier 1 and Total RBC could be used to:

 Ensure depositors, FDIC and senior debt holders are paid  Provide a basis for recapitalization

Minimum capital ratios imply that at the 1/1000 year loss levels implied by BIII calculations, a bank would not be a viable entity

Hypothetical Bank, NA $ Ratio $

• Min. Ratio

Assets 10,000 BIII RWA 7,500 600 8.0% CET1 750 10.0% 525 7.0% Tier 1 850 11.3% 638 8.5% Total RBC 1,050 14.0% 788 10.5% Actual Buffer Minimums CET1 / RWA Loss 125% 38% 88% Tier 1 / RWA Loss 142% 35% 106% Total RBC / RWA Loss 175% 44% 131% Minimum Capital Required

12

Illustration of Economic Capital Ratios

• As with regulatory capital, EC limits could be structured to show

how much of the total loss will be absorbed by each level of capital.

• Economic capital is

the internal parallel to the BIII regulatory ratios – measures how large losses could be in the 1/1000, 3/10,000, etc. event.

• EC is more

comprehensive and precise than RC

• Ties well with

pricing

Hypothetical Bank, NA $ Ratio $

• Min. Ratio

Assets 10,000 BIII RWA 7,500 550 7.3% Avail Cap 1 800 10.7% 413 5.5% Avail Cap 2 900 12.0% 563 7.5% Total AC 1,100 14.7% 675 9.0% Actual Buffer Minimums AC1 / RWA Loss 145% 70% 75% AC2 / RWA Loss 164% 61% 102% Total AC / RWA Loss 200% 77% 123% Minimum Capital Required

13

CCAR

Economic Capital Group

• CCAR builds upon the solvency requirements
• f regulatory capital (or EC in some cases).
• CCAR (and other stress testing), answer the

questions:

• Does an institution have sufficient capital to ensure

it is:

 Perceived by markets as viable (non-zombie).  Can continue prudent lending.  Ensure depositors, FDIC and senior debt holders are paid

should failure occur.

 Provide a basis for recapitalization should failure occur.

14

CCAR Example

• In this example, bank is able to absorb downturn

scenario impacts on earnings and capital ratios.

• Maintains sufficient capital resources to repay

creditors.

Hypothetical Bank, NA $ Ratio Runoff / New Volume Credit Migration Market Risk Ops Risk NII / Fees / Expenses $ Ratio Assets 10,000 (1,000) (100) (5)

• 8,895
• BIII RWA

7,500 (750) 1,000 10 100

• 7,860
• Req. Capital

600 (60) 80 1 8

• 629
• CET1

750 10.0% (15) (75) (5) (25) (25) 605 7.7% Tier 1 850 11.3% (15) (75) (5) (25) (25) 705 9.0% Total RBC 1,050 14.0% (15) (75) (5) (25) (25) 905 11.5% Capital Resource Coverage Spot Projected Min CET1 / RWA Loss 125% 11%

• 30%
• 1%
• 5%
• 4%

96% 88% Tier 1 / RWA Loss 142% 13%

• 32%
• 1%
• 6%
• 4%

112% 106% Total RBC / RWA Loss 175% 17%

• 37%
• 1%
• 6%
• 4%

144% 131% Spot Stress Testing Impacts Projected Buffer Absorption

15

Agenda

• Defining the Risk Measures
• Information Content
• The Problems They Solve
• Reconciling The Stories

16

Capital Adequacy: Summary of Problems Solved

• Regulatory Capital: Regulatory perspective
• n a firm’s ability to repay depositors and

recapitalize without an impact on FDIC should firm fail.

• Economic Capital: Same as regulatory but

based upon management’s views of risk and resources

• CCAR (Stress Testing): Regulatory

perspective on firm’s ability to weather a periodic downturn.

17

How Far From Edge vs. How Far Of a Fall

• Reg. Cap and EC

(Resolution) CCAR (Cushion)

18

Other Uses?

• Economic Capital: “True” amount of capital

needed to support risk.

• Regulatory Capital: Capital cost of doing

business.

• CCAR (Stress Testing): Potential areas of

earnings volatility.

19

Pricing and Performance Management

• EC should drive prioritization of economic investment decisions
• Regulatory capital and CCAR cushions are additional costs to be
• considered. Multiple ways to incorporate. Complex, nuanced and

imperfect.

• Top of house vs. bottom-up

Hypothetical Bank, NA

$ Capital Requirements (10% Minimum Hurdle Rate) EC ROEC RC RORC RORC + CCAR RORC +CCAR Asset 1 10.0 15.0% 11.0 13.6% 1.0 12.5% Asset 2 20.0 10.0% 22.0 9.1% 0.5 8.9% Asset 3 15.0 9.0% 12.0 11.3% 2.0 9.6% Asset 4 5.0 20.0% 4.0 25.0% 1.0 20.0% Asset 5 5.0 25.0% 9.0 13.9% 0.5 13.2% Total 55.0 12.9% 58.0 12.2% 5.0 11.3%

Reprice, restructure, exit Economics good – how to handle reg cost.

Exposure to stress scenario

20

Unexpected Loss

• EC models typically can produce an

unexpected loss result for various confidence levels.

• CCAR is an unexpected loss estimate

based upon one or more adverse scenarios.

• Both can be used to help manage

portfolio concentrations and earnings volatility

21

Unexpected Loss: Different Perspectives

• Each model provides a different perspective on how much

unexpected losses could impact earnings.

• Useful in managing concentrations, hedging risks and managing

capital.

• Different perspectives on the same question

Hypothetical Bank, NA

Dollar Earnings-at-Risk EC Model UL CCAR Adverse CCAR Severe Adverse Asset 1 0.5 0.8 1.5 Asset 2 0.6 0.4 0.8 Asset 3 3.0 1.5 3.0 Asset 4 1.0 0.8 1.5 Asset 5 1.0 0.4 0.8 Total 6.1 3.8 7.5

22

Different Perspectives on Same Problem

• No models are right. Some are useful.
• Different perspectives are needed to get

a reliable perspective on risk.

23

Different Perspectives on Same Problem: Hypothetical Bank

• In statistical terms, EC, RC, CCAR are all estimates of the firm’s true loss

distribution.

• Understanding the difference between them can lead to insights into the

firm’s risks.

CCAR Adverse Other Scenarios CCAR Severe Adverse CCAR Expected

Implied CCAR loss distribution

EC Event (99.9, 99.93, etc) EC Expected UL (66%ile) 80%ile 70%ile

EC Loss Distribution “True” Loss Distribution

24

Agenda

• Defining the Risk Measures
• Information Content
• The Problems They Solve
• Reconciling The Stories

25

How Does One Reconcile EC and CCAR

• We understand that the loss distributions are

different.

• But are they really comparable?

 Different loss distributions  Different time horizons  Different migration matrixes  Different discounting assumptions  Different balance sheets  Different correlation assumptions  PD/LGD correlation

• How useful is it to say EC says X and CCAR X/2

if you don’t understand the details?

26

Decomposition and Comparison

• To make sense of these numbers, one

needs to break them down into component parts and carefully contrast the assumptions.

• Why bother?

Challenge core assumptions by

playing one off against the other

Monitor performance of risk frame

works at useful level of detail

27

New Volume

Finding Common Ground: Hypothetical Bank

• Start with the core assumptions: 1 year loss.

EC

Fees

Economic Loss Accounting Loss

Migration Sharpe

CCAR – 3 Year 1 Year

• Conf. Level.

Common Ground but Different Assumptions

Expected PDs, LGDs, EADs Credit Migration / Default Corr. LGD Correlation

28

Comparing Hypothetical Bank’s Assumptions

Assumption Observations and Reasons for Differences PDs, LGDs, EADs Are the expected values the same or different? PIC vs. TTC? If different, are variances reasonable? Credit Migration / Default Correlation Macro-variable based vs. RiskFrontier. How do the models treat industries, products, geographies differently. PD/LGD Correlation Are the model using the same drivers or collateral segmentations?

• Ideally quantify by running models with similar settings
• Goal: Are the differences consistent with one’s understanding of

the portfolio risk.

29

Illustrations of Hypothetical Bank’s Conclusions

Model Observations and Reasons for Differences PD Differ PDs used for EC use a hybrid cycle sensitivity while CCAR is close to PIC. Given that we are in recovery, EC EL will be somewhat higher than CCAR baseline. Impact is most apparent for asset classes with long cycles like CRE. Overall variance in EL is in line with expectations. Correlation – CRE CRE correlation assumptions as measured by the EC/EL multiple are consistent on

• average. However, CCAR does not go to MSA
• r property type level so direct comparison is

not possible. PLC PLC models for EC and CCAR leverage the same data but are producing very different

• results. Deeper analysis of variances

required.

30

New Volume

Finding Common Ground: Hypothetical Bank

• How do we compare confidence levels?

EC

Fees

Economic Loss Accounting Loss

Migration Sharpe

CCAR – 3 Year 1 Year

• Conf. Level.

Common Ground but Different Assumptions

Expected PDs, LGDs, EADs Credit Migration / Default Corr. LGD Correlation

How do we convert from 99.XX to CCAR CL?

31

Mapping CCAR Scenarios to EC Loss Distribution

• To compare EC and CCAR, a confidence level mapping is helpful
• How does one move between the Moody’s Factor Correlation model and

CCAR?

CCAR Severe Adverse EC Event (99.9, 99.93, etc) EC Expected 80%ile 70%ile

EC Loss Distribution

? ? ?

32

Macro Economic Variable Based Mapping

• One approach is to attempt to estimate the CL associated with the CCAR

scenario and map to RiskFrontier loss distribution

• Another is to leverage the output from RiskFrontier to align CCAR

scenario macro forecasts with those implied by the RiskFrontier model which requires macro outputs and simulation details.

CCAR Adverse 1 in 20 Year Event GDP: -7.1% EC Event (99.9, 99.93, etc) EC Expected 80%ile 70%ile

EC Loss Distribution

GDP: -3 GDP: -4 GDP: -5 GDP: -4.5 GDP: -6 GDP: -8 GDP: -7.3 GDP: -9.2 GDP: -12.1 85%ile

33

Potential Learnings from Exercise

Model Observations and Reasons for Differences Greater Confidence in EC EC estimates tail. Impossible to really know if the tail estimate is correct. If EC loss at lower confidence level can be mapped to CCAR scenarios, could improve confidence in EC

• utput.

Challenge EC and/ or CCAR Assumptions If the EC scenario that maps to CCAR is either much higher (99.95%) or lower (60%) than expected, then underlying correlation assumptions for both models should be challenged.

34

Closing Observations

• The capital management playing field is much more

complex and requires more analytical horsepower.

• There is no “right” measure of risk – multiple views are

valuable

• EC is still the best tool for determining the required

economic returns on the firm’s capital and aligning them with the firm’s aggregate capital holdings.

• Regulatory capital and the associated CCAR driven

buffers are a cost of doing business.

• Understanding the linkage between EC, RC and CCAR

will help a firm with the task of optimizing capital.

35

DISCLAIMER

THANK YOU

THE VIEWS EXPRESSED IN THIS PRESENATATION WERE THOSE OF THE SPEAKER AND NOT NECESSARILY THOSE OF PNC FINANCIAL SERVICES GROUP, INC. OR ITS SUBSIDIARIES.