utilizing dual pit/ttc ratings to support ifrs9 … dual pit/ttc ratings to support ifrs9 expected...

Utilizing Dual PIT/TTC Ratings to Support

IFRS9 Expected Loss Calculations

CRC – University of Edinburgh Business School

Dr. Scott D. Aguais - Managing Director, Aguais & Associates Ltd.

[email protected]

August 4, 2015 – DRAFT – FINAL

2Aguais & Associates Ltd. CRC - IFRS9

Point-in-Time Methodology & IFRS9 Challenges

AgendaAgenda

• Overview – Key Points – PIT IFRS9 Challenges & Historical Background

• Overview - PIT-TTC Framework & Methodology Required for Developing PIT Measures

Systematic Credit Cycles, PIT/TTC Measures, Model Calibration & Batch Processing

Utilizing PIT Measures to Project ‘ECL’ for IFRS9 for Retail, Commercial & Wholesale

• Summary – Integrated IFRS9 & Stress test Architecture & Key Points in the Presentation



Key Points in this PIT/TTC Dual Ratings IFRS9 ChallengesKey Points in this PIT/TTC Dual Ratings IFRS9 Challenges

1) Systematic Credit Cycles Exist & Can be Measured – Motivates PIT-TTC

Distinctions

2) Evolving Regulatory Agenda – TTC for Basel II – PIT for IFRS9 & Stress Testing

3) IFRS9 Modelling – Commercial, Corporate & Retail Require Different Approaches

4) Substantial Accuracy Implications Motivate PIT – Wholesale ECLs can vary across

the credit-cycle between starting at a peak or trough by a factor of 3-5 times !!!

5) IFRS9 & Stress Testing Require One Integrated Batch Solution & Architecture


4. Multiple Business Objectives:

• Capital (TTC)• IFRS9/Provisioning (PIT)• Stress Testing (Stress PIT)• Risk Appetite (TTC)

• Sanctioning (TTC)• Pricing (PIT)• Early Warning (PIT)• CVA (PIT)

Risk & Regulatory Drivers Motivate PIT/TTC Ratings

Dual PIT/TTC Ratings Support Multiple Business GoalsDual PIT/TTC Ratings Support Multiple Business Goals

Models/Ratings

Models/Ratings

Data

Process

1. Regulatory Compliance:

• Support portfolio-wide PIT/TTC methodology development • Provide integrated & unified IFRS9 & Stress Testing solutions

• E2E support for internal model governance & regulatory approval

3. Increase Returns:

• Batch Processing: competitive advantage• Accurate Asset Valuations/Early Warning• Optimize Model accuracy & Capital Costs

2. Know Your Risk:

• E2E PIT/TTC implementation• Dual PIT/TTC ratings framework• Advanced Credit Cycle Measures• Integrated PD/LGD/EAD Solution


Primary PIT/TTC Drivers - Risk Ratings & Regulatory Evolution

Dual Ratings & Credit Benchmarking – Represent New Paradigms in Risk ManagementDual Ratings & Credit Benchmarking – Represent New Paradigms in Risk Management

Single Internal RatingSingle Internal Rating

90s 2002-07 2014 2015-20

Regulatory Evolution

Basel II(TTC)

Basel II(TTC)

Stress Testing(PIT)

Stress Testing(PIT)

IFRS 9(PIT)

IFRS 9(PIT)

Internal AIRB Models Determine

TTC Rating

Internal AIRB Models Determine

TTC Rating

Post B2 – ‘RW Conundrum’

Leads to ‘AERB’ Model Calibrations

Post B2 – ‘RW Conundrum’

Leads to ‘AERB’ Model Calibrations

Credit BenchmarkingAcross Bank’s

Internal Ratings

Credit BenchmarkingAcross Bank’s

Internal Ratings

Full External & Internal

Benchmarking

Full External & Internal

Benchmarking

Basel IBasel I


Systematic Credit Cycles Require PIT/TTC Distinctions

Dual Ratings (PIT & TTC) - Required for IFRS 9 & Stress Testing Compliance/AccuracyDual Ratings (PIT & TTC) - Required for IFRS 9 & Stress Testing Compliance/Accuracy

‘Point‐in‐time risk parameters (PDpit and LGDpit) should be forward looking projections of default rates and loss rates and capture current trends in the business cycle. In contrast to through‐the‐cycle parameters they should not be business cycle neutral. PDpit and LGDpit should be used for all credit risk related calculations except RWA under both, the baseline and the adverse scenario. Contrary to regulatory parameters, they are required for all portfolios, including STA and F‐IRB.’

EBA – ‘Methodology EU-wide Stress Test 2014’Version 1.8, 3 March 2014, P 26


Regulatory Evolution Also Motivates Benchmarking

Dual Ratings & Benchmarking – Represent New Paradigms in Risk ManagementDual Ratings & Benchmarking – Represent New Paradigms in Risk Management

Use External Agency

Ratings & Default to

Develop Corp PD Models

‘AgencyReplication’

Use External Agency

Ratings & Default to

Develop Corp PD Models

‘AgencyReplication’

1990s

BenchmarkingInitiatives to

collect & compare model

calibration

BenchmarkingInitiatives to

collect & compare model

calibration

2000s

Market-Based PD ModelsMKMV EDFs

Market-Based PD ModelsMKMV EDFs

2010 2015 +

PECDCLoan Loss Data Collection ‘By

Banks for Banks’Supports LGDBenchmarking

PECDCLoan Loss Data Collection ‘By

Banks for Banks’Supports LGDBenchmarking

Credit Derivative MarketsPricing

Risk Neutral PD

Credit Derivative MarketsPricing

Risk Neutral PD

AIRB Regulatory

BenchmarkingFSA HPEEBA/FRB

Etc

AIRB Regulatory

BenchmarkingFSA HPEEBA/FRB

Etc

Basel IISubstantial

focus on collecting &

using Internal Credit Data for ‘Internal’ Model

Calibration

Basel IISubstantial

focus on collecting &

using Internal Credit Data for ‘Internal’ Model

Calibration

AERBAERB

* Just Published: Forest, Chawla & Aguais, ‘Biased

Benchmarks’, Spring 2015

Journal of Risk Model Validation



AgendaAgenda







Systematic Credit Cycles Require PIT/TTC Distinctions

TTC Ratings for B2 are Conditionally Cycle Neutral – PIT Ratings Reflect ‘True’ RiskTTC Ratings for B2 are Conditionally Cycle Neutral – PIT Ratings Reflect ‘True’ Risk

• Credit Conditions deteriorate, client cash flows fall, client PDs rise

• Credit Conditions improve, client cash flows rise, client PDs fall

• Credit Conditions (assessed using equity data) change frequently

• Monthly changes according to client’s primary region and sector (PIT)

• Region and Sector values set to their long run averages (TTC)

Idiosyncratic Risk(Specific to the Client)

Systematic Risk(Specific to the Economy)

• Primarily a function of Leverage & Cash Flow Volatility

• Assessed using financials &qualitative assessments

• Changes measured by internal models are typically infrequent

PIT & TTC PD


Systematic Wholesale Credit Cycles Can Be Measured

The Existence of Credit Cycles Motivates PIT/TTC Modeling ApproachesThe Existence of Credit Cycles Motivates PIT/TTC Modeling Approaches

• Credit cycles are prominent in corporate defaults, losses & MKMV EDFs

• Using credit cycles in PD estimation significantly improves PIT prediction accuracy

• Empty Glass vs 20% Full Glass

ZGAPs: Smoothed 3 Qtr Moving Average - Corporates

-3.00

-2.00

-1.00

0.00

1.00

2.00

3.00

19

90

-Q1

19

91

-Q1

19

92

-Q1

19

93

-Q1

19

94

-Q1

19

95

-Q1

19

96

-Q1

19

97

-Q1

19

98

-Q1

19

99

-Q1

20

00

-Q1

20

01

-Q1

20

02

-Q1

20

03

-Q1

20

04

-Q1

20

05

-Q1

20

06

-Q1

20

07

-Q1

20

08

-Q1

20

09

-Q1

20

10

-Q1

20

11

-Q1

20

12

-Q1

KMV EDFs SP DRs US Bank LRs Moody's DRs

Credit Cycles Indices Derived from Various PD, Rating & Loss Measures: MKMV EDFs, S&P Default

Rates & C&I Loss Rates

Annualised Quarterly default rates: 3 Quarter Moving Average – S&P Corporates

Annualized Quarterly DR: 3-qtr Moving Average - Corporates

0.00%

1.00%

2.00%

3.00%

4.00%

5.00%

6.00%

7.00%

8.00%

19

90

-Q1

19

91

-Q1

19

92

-Q1

19

93

-Q1

19

94

-Q1

19

95

-Q1

19

96

-Q1

19

97

-Q1

19

98

-Q1

19

99

-Q1

20

00

-Q1

20

01

-Q1

20

02

-Q1

20

03

-Q1

20

04

-Q1

20

05

-Q1

20

06

-Q1

20

07

-Q1

20

08

-Q1

20

09

-Q1

20

10

-Q1

20

11

-Q1

20

12

-Q1

S&P Corporate DR Moody's Corporate DR

Default Rate

Predicting historical defaults improves when incorporating credit cycle measures directly in estimating PIT models

Removing the credit cycle in implementation generates correctly calibrated TTC PDs

Long run average historical default rate (TTC) required for capital (25 yrs)

CreditCycleIndex


Almost All Credit Models Are Blind to Credit Cycles

A Component of Credit Cycles is Predictable - Systematic Cycles in Industries & Regions A Component of Credit Cycles is Predictable - Systematic Cycles in Industries & Regions

-3.0

-2.0

-1.0

0.0

1.0

2.0

3.0

Z-G

ap

NA Corp Z Credit

Index

Legacy Credit

Models

Predicted by Credit-

Cycle Model

Legacy Credit

Models


Cycle model

Legacy Credit

Models

1990 20091998 1999 2001 2002 2007 2011-3.0

-2.0

-1.0

0.0

1.0

2.0

3.0

Z-G

ap

NA Corp Z Credit

Index

Legacy Credit

Models


Cycle Model

Legacy Credit

Models


Cycle model

Legacy Credit

Models


Cycle model

Legacy Credit

Models

1990 20091998 1999 2001 2002 2007 20111990 20091998 1999 2001 2002 2007 2011

Source: Moody’s KMV, Aguais/Forest research

Current Credit Models Are Blind to Credit Cycles – 20% Prediction is Therefore PowerfulCurrent Credit Models Are Blind to Credit Cycles – 20% Prediction is Therefore Powerful


Joint Projections of PIT PDs, LGDs, EADs Determine ECLs

Bond & Loan Portfolios – B2 Models & Agency Ratings Converted to PIT for IFRS9 ECLBond & Loan Portfolios – B2 Models & Agency Ratings Converted to PIT for IFRS9 ECL

Projected ‘ECL’ – Utilise Jointly Simulated/Correlated PIT EAD/LGD/EADWholesale Basel PD models & Agency Ratings are mostly TTC -– Therefore Credit

Cycle Indices can be Utilised to Convert Agency Ratings or Internal Scorecard Models to Pure PIT Measures for IFRS9

Time

CreditCycleIndex

Credit-cycle indexes allow banks to convert hybrid indicators to PIT for

accurate outlooks and to TTC for Basel RWA

0

+

-

PIT (100%) TTC (100%)Agency Ratings

BASEL PD or Scorecard

Models

PIT Basel Models & Agency Ratings


Weighted Avg

Aerospace & Defence

Banking

Chemicals & Plastic Products

Construction

Consumer Products

Oil & Gas

Finance, Real Estate & Insurance

Hotels & Leisure

Basic Industries

Machinery & Equipment

Media

Medical

Steel & Metal Products

Mining

Motor Vehicle & Parts

Retail & Wholesale Trade

Business & Consumer Services

Technology

Transportation

Utilities

Commercial Real Estate

Asia

Continental Europe

United Kingdom

Latin America

North America

Pacific

Regional ZR

(Corp/FI)

TTC

Avg PIT PD

Time

PD

Industry Sector ZI Spot Median ZS/R Gap

LR Median ZS/R Gap

t

Industry & Region Systematic Factors are Combined to Develop Credit Cycle IndicesIndustry & Region Systematic Factors are Combined to Develop Credit Cycle Indices

PIT/TTC Approach Models Detailed Industry/Regions


The Core of the Commercial PIT Methodology - The Evolution of Credit CyclesThe Core of the Commercial PIT Methodology - The Evolution of Credit Cycles

PIT/TTC Approach Models Detailed Mean Reversion/Momentum

Z Value

Time

Historical z

Today

Momentum

Mean Reversion

Long Term Mean

Almost All Data Exhibiting Credit Cycles Shows Two Competing Empirical Influences

Credit Cycle Behavior (Z) is Driven by Two Competing Influences – Mean Reversion & Momentum


Industry Examples of Systematic Credit Cycles & Forecasts

Banking, Finance, Insurance & Real EstateBanking, Finance, Insurance & Real Estate

-4.0

-3.0

-2.0

-1.0

-

1.0

2.0

3.0

1-

91

1-

92

1-

93

1-

94

1-

95

1-

96

1-

97

1-

98

1-

99

1-

00

1-

01

1-

02

1-

03

1-

04

1-

05

1-

06

1-

07

1-

08

1-

09

1-

10

1-

11

1-

12

1-

13

1-

14

1-

15

1-

16

1-

17

1-

18

Z G

ap

Banking Finance, Insurance & Real Estate Long Run Average

< Actual Forecast >


Industry Examples of Systematic Credit Cycles & Forecasts

Industries – Basic Industries, Steel & Metal Products, Mining, Utilities, Oil & GasIndustries – Basic Industries, Steel & Metal Products, Mining, Utilities, Oil & Gas

-3.0

-2.0

-1.0

-

1.0

2.0

3.0

1-

91

1-

92

1-

93

1-

94

1-

95

1-

96

1-

97

1-

98

1-

99

1-

00

1-

01

1-

02

1-

03

1-

04

1-

05

1-

06

1-

07

1-

08

1-

09

1-

10

1-

11

1-

12

1-

13

1-

14

1-

15

1-

16

1-

17

1-

18

Z G

ap

Basic Industries Mining Oil & Gas

Steel & Metal Products Utilities Long Run Average

< Actual Forecast >


-3.0

-2.0

-1.0

0.0

1.0

2.0

3.0

1-

91

1-

92

1-

93

1-

94

1-

95

1-

96

1-

97

1-

98

1-

99

1-

00

1-

01

1-

02

1-

03

1-

04

1-

05

1-

06

1-

07

1-

08

1-

09

1-

10

1-

11

1-

12

1-

13

1-

14

1-

15

1-

16

1-

17

1-

18

Z G

ap

Asia Europe Great Britain Latin America

North America Pacific Long Run Average

Forecast >< Actual

Regional Examples of Systematic Credit Cycles & Forecasts

Regional Corporates - Asia, North-America, Europe, Pacific, UK, Latin AmericaRegional Corporates - Asia, North-America, Europe, Pacific, UK, Latin America


MKMV EDFs for S&P BBB Rated Companies (Non-FIs)

-- NA, EU&UK and APAC

0.0%

0.1%

0.2%

0.3%

0.4%

0.5%

0.6%

0.7%

0.8%

Jan-

95

Jan-

96

Jan-

97

Jan-

98

Jan-

99

Jan-

00

Jan-

01

Jan-

02

Jan-

03

Jan-

04

Jan-

05

Jan-

06

Jan-

07

Jan-

08

Jan-

09

Jan-

10

Jan-

11

Asia&Pacific Corps

EU&UK Corps

N.America Corps

Global Average

Source: S& P & Moody’s KMV,

BBB Rated Companies Have Substantial PIT Movements When ConvertedBBB Rated Companies Have Substantial PIT Movements When Converted

Agency Ratings Can Also Be Modelled with PIT Adjustments


Core PD PIT Methodology Needs to Be Adapted for Different Borrowers & ModelsCore PD PIT Methodology Needs to Be Adapted for Different Borrowers & Models

Developing Multi-Year Cycle-Adjusted IFRS9 Credit Estimates

‘Commercial’ - £1 mil to £25-50 mil Turnover

‘Commercial’ - £1 mil to £25-50 mil Turnover

Retail/Consumer & Small Bus up to £1 mil

Retail/Consumer & Small Bus up to £1 mil

‘Corporates & Banks’‘Corporates & Banks’

Dominated by Behavioral Scorecards

- ‘Backward Looking’ PIT

Dominated by Behavioral Scorecards

- ‘Backward Looking’ PIT

Need PIT Conversion & Forward Looking

PD/LGD/EAD Term Structures

Dominated by Commercial Scorecards

- Financials & Qualitatives‘Close to ‘Fully TTC’

Dominated by Commercial Scorecards

- Financials & Qualitatives‘Close to ‘Fully TTC’

Dominated by ‘Agency Replication’

- ‘Mostly TTC’

Dominated by ‘Agency Replication’

- ‘Mostly TTC’

Need Forward Looking


Need PIT Conversion & Forward Looking



PIT & Stress Test Modelling – Model Calibration Utilizes Macro & Credit FactorsPIT & Stress Test Modelling – Model Calibration Utilizes Macro & Credit Factors

LGDLGD EADEADPDPDBasel TTC

Models

GDP, Equity Indexes & Credit Spreads

GDP, Equity Indexes & Credit Spreads

Benchmark PIT PDFs – MKMV, Kamakura, Bloomberg

Benchmark PIT PDFs – MKMV, Kamakura, Bloomberg

Industry/Region

Personal Income/debt, Unemployment, House Prices, Consumer Loss Rates, Benchmark DRs

Personal Income/debt, Unemployment, House Prices, Consumer Loss Rates, Benchmark DRs

Credit Factors

Portfolios/Obligors

Macro Factors

PIT/TTC Framework for Wholesale Can Be Applied to Retail

GDP, Personal Income, Unemployment, House Prices

GDP, Personal Income, Unemployment, House Prices

-5

-4

-3

-2

-1

0

1

2

3

4

Jan-

90

Jan-

91

Jan-

92

Jan-

93

Jan-

94

Jan-

95

Jan-

96

Jan-

97

Jan-

98

Jan-

99

Jan-

00

Jan-

01

Jan-

02

Jan-

03

Jan-

04

Jan-

05

Jan-

06

Jan-

07

Jan-

08

Jan-

09

Jan-

10

Jan-

11

Jan-

12

Jan-

13

Jan-

14

Jan-

15

Jan-

16

Z G

ap

Asia Europe Great Britain Latin America North America Pacif ic South Africa

-5

-4

-3

-2

-1

0

1

2

3

4

Jan-90 Jan-91 Jan-92 Jan-93 Jan-94 Jan-95 Jan-96 Jan-97 Jan-98 Jan-99 Jan-00 Jan-01 Jan-02 Jan-03 Jan-04 Jan-05 Jan-06 Jan-07 Jan-08 Jan-09 Jan-10 Jan-11 Jan-12 Jan-13 Jan-14 Jan-15 Jan-16

Z G

ap

Hotels and Leisure M edia Retail & Wholesale Trade Technology Transportat ion

Regional Credit Cycle Indices Industry Credit Cycle Indices

UNSECUNSEC LG CorpsLG CorpsCRECRERet MortsRet Morts C CardsC Cards SMESMEAsset FinAsset Fin Soc HouSoc Hou SovsSovsNBFIsNBFIs BanksBanks


PIT/TTC Rating – Accurate (1) Assessment of Significant Deterioration & (2) Lifetime ELPIT/TTC Rating – Accurate (1) Assessment of Significant Deterioration & (2) Lifetime EL

IFRS 9 – Requires Point-in-Time Measures

Time

IFRS ‘Significant

Deterioration’

IFRS ‘Significant

Deterioration’

Stable Credit Risk

Stable Credit Risk

Credit Risk Improvement

Credit Risk Improvement

OriginationOrigination 1 Yr Later1 Yr Later + 1 Year+ 1 Year

IFRS 1 Year Credit EL

IFRS 1 Year Credit EL

IFRS Lifetime Credit EL

IFRS Lifetime Credit EL

Balance of 5-Year Facility/Borrower Term

PIT Cycle Adjusted PD Term StructuresPIT Cycle Adjusted PD Term Structures

S&P

AAA

AA

A+/A

A/A-

BBB+

BBB

BBB-

BB+

BB

BB-

B+

B

B-

CCC+

CCC

CCC-

D

S&P

Mapping

AAA

AA

A+/A

A/A-

BBB+

BBB/BBB-

BBB-

BB+

BB

BB-

B+

B

B-

CCC+

CCC

CCC-

D

PD Bins

Mid

0.01%

0.02%

0.04%

0.07%

0.11%

0.22%

0.32%

0.47%

0.71%

1.07%

1.58%

2.27%

3.21%

4.45%

6.05%

8.07%

10.56%

15.26%

23.41%

37.07%

100.00%

PD

Mid-Point

0.01%

0.02%

0.04%

0.07%

0.14%

0.

0.

0.47%

0.71%

1.07%

1.58%

2.27%

3.21%

4.45%

6.05%

8.07%

10.56%

15.26%

23.41%

37.07%

100.00%

InternalRatings

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

D

InternalRatings

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

D


Accurately Assessing IFRS ‘Significant Deterioration’ Requires PIT Risk MeasuresAccurately Assessing IFRS ‘Significant Deterioration’ Requires PIT Risk Measures

IFRS 9 – Requires Point-in-Time Measures

S&P

AAA

AA

A+/A

A/A-

BBB+

BBB

BBB-

BB+

BB

BB-

B+

B

B-

CCC+

CCC

CCC-

D

S&P

Mapping

AAA

AA

A+/A

A/A-

BBB+

BBB/BBB-

BBB-

BB+

BB

BB-

B+

B

B-

CCC+

CCC

CCC-

D

PD Bins

Mid

0.01%

0.02%

0.04%

0.07%

0.11%

0.22%

0.32%

0.47%

0.71%

1.07%

1.58%

2.27%

3.21%

4.45%

6.05%

8.07%

10.56%

15.26%

23.41%

37.07%

100.00%

PD

Mid-Point

0.01%

0.02%

0.04%

0.07%

0.14%

0.

0.

0.47%

0.71%

1.07%

1.58%

2.27%

3.21%

4.45%

6.05%

8.07%

10.56%

15.26%

23.41%

37.07%

100.00%

InternalRatings

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

D

InternalRatings

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

D

Holistic ‘Forward Looking’/Qualitative

Assessments

Holistic ‘Forward Looking’/Qualitative

Assessments

Agency Ratings – ‘External

Assessment’

Agency Ratings – ‘External

Assessment’

IFRS ‘Low Risk vs High Risk’

IFRS ‘Low Risk vs High Risk’

AIRB TTC Models

AIRB TTC Models

‘Collective IFRS Assessment’‘Collective IFRS Assessment’But PIT Risk for a Given TTC Grade Can Move Up to 5XBut PIT Risk for a Given TTC Grade Can Move Up to 5X

Systematic Credit

Conditions

Systematic Credit

Conditions

‘BBB’ Mapping for LR Average Credit Conditions

‘BBB’ Mapping for ‘Good Times’ PIT Credit Conditions

‘BBB’ Mapping for ‘Bad Times’ PIT Credit Conditions

AIRB TTC Models Designed for Capital – IFRS

‘Significant Deterioration’ Assessment Requires PIT Measures to be Accurate

AIRB TTC Models Designed for Capital – IFRS

‘Significant Deterioration’ Assessment Requires PIT Measures to be Accurate

PIT Ratings Changes Driven Only by Systematic Industry/Region Credit Cycles

PIT Ratings Changes Driven Only by Systematic Industry/Region Credit Cycles


Weighted Avg

Depending on the Start Point – BB 5-Yr Loss Predictions Can Vary by a Factor of 5 Times Depending on the Start Point – BB 5-Yr Loss Predictions Can Vary by a Factor of 5 Times

PIT PD/LGD/EAD Conversions - Critical to IFRS9 Accuracy

• See below PD outlook for a TTC BB counterparty starting respectively with the credit cycle at TTC, a trough (Z=-2.5), and a peak (Z=2.5)

• TTC model produces the TTC line under all cyclical circumstances, whereas PIT model produces the more accurate estimates sensitive to initial & prospective credit conditions

• PD term structures are prospective by using a mean reversion momentum model

IFRS9 Accuracy Requires Both PIT Grades & PIT Prospective Credit ConditionsIFRS9 Accuracy Requires Both PIT Grades & PIT Prospective Credit Conditions

0.00%

2.00%

4.00%

6.00%

8.00%

10.00%

12.00%

14.00%

16.00%

Year 1 Year 2 Year 3 Year 4 Year 5

BB Rated Cumulative 5-Year PDs

TTC Cycle Trough Cycle PeakIFRS9 Life Time

Horizon

Portfolio PD

BB PD

AVG

Top-to-Bottom Accuracy Swing is

Substantial


IFRS 9 Loss Inaccuracy Can Be Significant by Not Using PIT

Estimates for a five-year, corporate, revolving facility with a TTC default grade of BB equivalent, expected utilization of 60%, and DT LGD of 40%.

PIT outlook accounts for Z projections and Z sensitivities of PIT PDs, LGDs, and EADs.

• ECL estimates that arise from TTC grades & a fixed, TTC matrix will be accurate only under the special circumstances that the credit cycle is at TTC and expected to remain there

• In other cases, the ECLs will be inaccurate, especially so at credit-cycle troughs and peaks (see below)

• Bottom line: need both PIT grades & a PIT forward-looking outlook

Grade Cycle Z level Z change Year 1 Year 2 Year 3 Year 4 Year 5

TTC TTC TTC BB BB 0.00 0.00 0.83% 2.15% 3.68% 5.47% 7.45% 2.05%

PIT Trough PIT BB B+ -2.50 -0.50 2.37% 4.95% 7.74% 10.78% 13.95% 4.48%

PIT Peak PIT BB BBB+ 2.50 0.10 0.24% 0.61% 1.10% 1.72% 2.50% 0.62%

Outlook

Spot Status

Case

Credit Cycle Status PV

ECL/Limit

TTC

Grade

Spot

Grade

Prospective PDs

HugeTop to

Bottom Accuracy Swing

IFRS9/ECL Accuracy Improvements Using PIT PDs Are Substantial IFRS9/ECL Accuracy Improvements Using PIT PDs Are Substantial


Developing ‘Best Estimates’ of ‘Expected Credit Losses’- ‘ECL’

Calculate ECL Using Counterparty, Facility, Model & Z Credit Factors Calculate ECL Using Counterparty, Facility, Model & Z Credit Factors

• Extract from source systems the relevant, categorical data (region, industry, asset class) & PDs, LGDs, and EADs for all facilities,

• Apply the Z credit factors in converting, where necessary, the PD, LGD, and EAD measures from source systems to spot, PIT ones,

• Run Monte Carlo simulations of the credit factors and apply those credit-factor scenarios together with the spot, PIT measures and the transition, LGD, and EAD models in creating joint, PD, LGD, and EAD scenarios for each facility; and,

• Combine & average those scenarios & thereby produce ECLs over the life of each facility.

Note that to get the completely unbiased estimates anticipated under IFRS 9, will need to eliminate upward biases from regulatory, credit models; 2013 study of Pillar 3 filings found that the regulatory models of large banks over-estimate losses by about 50% on average


Developing ‘Best Estimates’ of ‘Expected Credit Losses’- ‘ECL’

Calculate ECL Using Counterparty, Facility, Model & Z Credit Factors Using SimulationsCalculate ECL Using Counterparty, Facility, Model & Z Credit Factors Using Simulations

• Draw random effects & run Z scenarios:

• Enter z (=Z) into conditional, transition models and Z into LGD and EAD models and produce joint, PD, LGD, EAD, and thereby ECL scenarios; bold = average = unconditional ECL:

-4.00

-3.00

-2.00

-1.00

0.00

1.00

2.00

3.00

4.00

1 2 3 4 5 6 7 8 9 10 11 12

Quarter Number

0.00

0.05

0.10

0.15

0.20

0.25

0.30

0.35

1 2 3 4 5 6 7 8 9 10 11 12

Quarter Number


Extending Credit Cycle Modelling to Retail Portfolios

Preliminary Retail Credit Cycle Modelling: Recurring Cycles Seem Less EvidentPreliminary Retail Credit Cycle Modelling: Recurring Cycles Seem Less Evident

• Strong pattern of cycles in C&I DDGAPs inferred from US bank loss rates

• Less evident in retail, with Great Recession unprecedented; thus, one may apply the legacy, random-walk model or work harder to remove non-stationary features before identifying cycles in the history

• Correlation between retail & wholesale loss rates moderate = diversification

-1.00

-0.80

-0.60

-0.40

-0.20

0.00

0.20

0.40

0.60

0.80

1.00

19

90

Q1

19

91

Q1

19

92

Q1

19

93

Q1

19

94

Q1

19

95

Q1

19

96

Q1

19

97

Q1

19

98

Q1

19

99

Q1

20

00

Q1

20

01

Q1

20

02

Q1

20

03

Q1

20

04

Q1

20

05

Q1

20

06

Q1

20

07

Q1

20

08

Q1

20

09

Q1

20

10

Q1

20

11

Q1

20

12

Q1

20

13

Q1

20

14

Q1

C&I Cards Cons Other RRE

1990-2014: DDGAPs Imputed from Corp & Retail Indicators

Source: US FRB/OCC

Quarterly Net Charge-off

Survey of Banks

C&I LossesCredit Card Losses‘Other’ ConsumerLossesMortgages (‘RRE’)



AgendaAgenda








Stress Test Approach Utilizes Macro Factor & ‘Bridge’ Model on Top of PIT Framework Stress Test Approach Utilizes Macro Factor & ‘Bridge’ Model on Top of PIT Framework

‘Macro-Merton’ Credit Cycle Stress Test Framework

View GDP & equity measures as asset-value proxies

Project macro debt on the basis of trends in asset-value proxies

Treat Debt/GDP & Debt/Equity as leverage measures

Derive macro DDs (‘Default-Distance’) as ratios of leverage to historical, leverage volatility

Convert macro DDs to macro Zs (by normalising mean & variance)

Use bridging relationship to derive industry-region Zs from macro Zs

Enter industry-region Zs into the PD, LGD, and EAD models and derive stress losses

Conditional Stressed PIT PD =

f4(Obligor’s internal assessment,

Stress Industry/Region Z )

Macro DD= L/V =f1 (GDP, Equity)

Macro Z = f2 (Macro DD)

Industry/Region Z = f3 (Macro Z)

Macro Scenarios

(GDP, Equity)


Integrated PIT/TTC & Stress Test Solution

IFRS9 & Stress Testing – Integrated Batch Architecture – Unconditional vs ConditionalIFRS9 & Stress Testing – Integrated Batch Architecture – Unconditional vs Conditional

Integrated Stress Test ‘Stress PIT’ Batch Analytics

Retail – Commercial - Wholesale

Integrated Stress Test ‘Stress PIT’ Batch Analytics

Retail – Commercial - Wholesale

Integrated IFRS9 ‘PIT’ Batch AnalyticsRetail – Commercial - Wholesale

Integrated IFRS9 ‘PIT’ Batch AnalyticsRetail – Commercial - Wholesale

Basel AIRB PD-LGD-EAD ModelsBasel AIRB PD-LGD-EAD Models

Enabling Batch Automation Data/Architecture/Financial Data/CP –Static Data

Enabling Batch Automation Data/Architecture/Financial Data/CP –Static Data

Methodology - A Single Unified Framework Across IFRS9 & Stress Testing

- Batch Analytics- E2E Implementation- E2E Governance &

Regulatory Sign-off- Custom Model Calibration

Batch Processing Leverages All Basel II PD/LGD/EAD Models

‘CONDITIONAL SCENARIO BASED’

‘UNCONDITIONAL SIMULATION BASED’ – BUT ‘CONDITIONAL ON CORRECT CYCLE STARTING POINT


Point-in-Time Methodology & IFRS9 Workshop

Summary - Key Points in this PIT/TTC Dual Ratings IFRS9 WorkshopSummary - Key Points in this PIT/TTC Dual Ratings IFRS9 Workshop

1) Systematic Credit Cycles Exist & Can be Measured – Motivates PIT-TTC

Distinctions

2) Evolving Regulatory Agenda – TTC for Basel II – PIT for IFRS9 & Stress Testing

3) IFRS9 Modelling – Commercial, Corporate & Retail Require Different Approaches

4) Substantial Accuracy Implications Motivate PIT – ECLs can vary across the credit-

cycle between starting at a peak or trough by a factor of 3-5 times !!!

5) IFRS9 & Stress Testing Require One Integrated Batch Solution & Architecture

utilizing dual pit/ttc ratings to support ifrs9 … dual pit/ttc ratings to support ifrs9 expected...

Documents