Measuring Lifetime Expected Credit Losses Using … Lifetime Expected Credit Losses Using PIT/TTC...

Measuring Lifetime Expected Credit Losses

Using PIT/TTC Dual Ratings to Support IFRS9

IFRS9 Workshop 2nd Edition of Credit Risk Modelling for

IFRS9

Gaurav Chawla - Senior Consultant Models & Methodology, Aguais & Associates Ltd.

[email protected]

November 16, 2015 DRAFT FINAL

2222Aguais & Associates Ltd.

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/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)

Stress Testing

(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

Benchmarking

Across Banks

Internal Ratings

Credit

Benchmarking

Across Banks

Internal Ratings

Full External &

Internal

Benchmarking

Full External &

Internal

Benchmarking

Basel 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

Agency

Replication

Use External

Agency

Ratings &

Default to

Develop Corp

PD Models

Agency

Replication

1990s

Benchmarking

Initiatives to

collect &

compare model

calibration

Benchmarking

Initiatives to

collect &

compare model

calibration

2000s

Market-

Based PD

Models

MKMV

EDFs

Market-

Based PD

Models

MKMV

EDFs

2010 2015 +

PECDC

Loan Loss Data

Collection By

Banks for Banks

Supports LGD

Benchmarking

PECDC

Loan Loss Data

Collection By

Banks for Banks

Supports LGD

Benchmarking

Credit

Derivative

Markets

Pricing

Risk

Neutral PD

Credit

Derivative

Markets

Pricing

Risk

Neutral PD

AIRB

Regulatory

Benchmarking

FSA HPE

EBA/FRB

Etc

AIRB

Regulatory

Benchmarking

FSA HPE

EBA/FRB

Etc

Basel II

Substantial

focus on

collecting &

using Internal

Credit Data for

Internal Model

Calibration

Basel II

Substantial

focus on

collecting &

using Internal

Credit Data for

Internal Model

Calibration

AERB

Just Published: Forest,

Chawla & Aguais, Biased

Benchmarks, Spring 2015

Journal of Risk Model

Validation

&, Forest, Chawla &

Aguais: AERB Developing

AIRB PIT/Tc PD Rating

Models Using External

Ratings, Winter 2015



AgendaAgenda



Measures




Presentation


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 clients 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 Corpora te 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)

Credit

Cycle

Index


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: Moodys KMV, Aguais/Forest research

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


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

Agency

Replication

Use External

Agency

Ratings &

Default to

Develop Corp

PD Models

Agency

Replication

1990s

Benchmarking

Initiatives to

collect &

compare model

calibration

Benchmarking

Initiatives to

collect &

compare model

calibration

2000s

Market-

Based PD

Models

MKMV

EDFs

Market-

Based PD

Models

MKMV

EDFs

2010 2015 +

PECDC

Loan Loss Data

Collection By

Banks for Banks

Supports LGD

Benchmarking

PECDC

Loan Loss Data

Collection By

Banks for Banks

Supports LGD

Benchmarking

Credit

Derivative

Markets

Pricing

Risk

Neutral PD

Credit

Derivative

Markets

Pricing

Risk

Neutral PD

AIRB

Regulatory

Benchmarking

FSA HPE

EBA/FRB

Etc

AIRB

Regulatory

Benchmarking

FSA HPE

EBA/FRB

Etc

Basel II

Substantial

focus on

collecting &

using Internal

Credit Data for

Internal Model

Calibration

Basel II

Substantial

focus on

collecting &

using Internal

Credit Data for

Internal Model

Calibration

AERB

Just Published: Forest,

Chawla & Aguais, Biased

Benchmarks, Spring 2015

Journal of Risk Model

Validation

&, Forest, Chawla &

Aguais: AERB Developing

AIRB PIT/Tc PD Rating

Models Using External

Ratings, Winter 2015


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

Credit

Cycle

Index

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


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 TurnoverRetail/Consumer &

Small Bus up to 1 milCorporates & Banks

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 Agency

Replication

- Mostly TTC

Need Forward

Looking

PD/LGD/EAD Term

Structures

Need PIT Conversion &

Forward Looking

PD/LGD/EAD Term

Structures


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

LGD EADPDBasel TTC

Models

GDP, Equity Indexes & Credit

Spreads

Benchmark PIT PDFs MKMV,

Kamakura, Bloomberg

Industry/

RegionPersonal Income/debt, Unemployment, House

Prices, Consumer Loss Rates, Benchmark DRsCredit Factors

Portfolios/

Obligors

Macro Factors

PIT/TTC Framework for Wholesale Can Be Applied to Retail

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 Pacific 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

Hot els and Leisure M edia Retail & Wholesale Trade Technology Transportation

Regional Credit Cycle Indices Industry Credit Cycle Indices

UNSEC LG CorpsCRERet Morts C Cards SMEAsset Fin Soc Hou SovsNBFIs Banks


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

Stable Credit

Risk

Credit Risk

Improvement

Origination 1 Yr Later + 1 Year

IFRS 1 Year

Credit EL

IFRS Lifetime

Credit EL

Balance of 5-Year Facility/Borrower Term

PIT 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

Agency Ratings

External

Assessment

IFRS Low Risk

vs High Risk

AIRB TTC

Models

Collective IFRS AssessmentBut PIT Risk for a Given TTC Grade Can Move Up to 5X

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

PIT Ratings Changes Driven Only by

Systematic Industry/Region Credit Cycles

PIT Ratings Changes Driven Only by

Systematic Industry/Region Credit Cycles


Assessing IFRS Significant Deterioration Requires Multiple PIT Risk AssessmentsAssessing IFRS Significant Deterioration Requires Multiple PIT Risk Assessments

IFRS 9 Requires Point-in-Time PD Term Structures

Requirements:

An entity shall use the change in the risk of a default

occurring over the expected life of the financial instrument

instead of the change in the amount of expected credit

losses (IFRS 9 5.5.9)

Options:1. Lifetime Point In Time PD term structure

2. 12m PIT PD

3. Other data based triggers (Ref IFRS 9 B5.5.17)

4. Expert Overlay

5. Combined Waterfall / Hierarchy of Application

Methodology to annualize PD Term

Structures:1. Lifetime annualized PD (LAPD)

2. Levelized marginal PD (LMPD)

3. Levelized forward PD (LFPD)

All PIT PD Term structures are same ones used in ECL calc

Ascertain Threshold Levels considering:1. Model Error bands

2. Volatility of Stage Allocation

3. Existing Credit processes

Implementation in Production Solution:


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 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

Huge

Top 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



AgendaAgenda



Measures




Presentation



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(Obligors 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

Retail Commercial - Wholesale

Integrated Stress Test Stress PIT

Batch Analytics


Integrated IFRS9 PIT Batch Analytics


Basel AIRB PD-LGD-EAD Models

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

Date post:	29-Apr-2018
Category:	Documents
Upload:	lyhanh
View:	228 times
Download:	3 times

Measuring Lifetime Expected Credit Losses Using … Lifetime Expected Credit Losses Using PIT/TTC...

Documents