Post on 20-Mar-2018
transcript
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.
saguais@aguaisandassociates.co.uk
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
3Aguais & Associates Ltd. CRC - IFRS9
Point-in-Time Methodology & IFRS9 Challenges
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
4Aguais & Associates Ltd. CRC - IFRS9
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
5Aguais & Associates Ltd. CRC - IFRS9
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
6Aguais & Associates Ltd. CRC - IFRS9
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
7Aguais & Associates Ltd. CRC - IFRS9
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
8Aguais & 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
9Aguais & Associates Ltd. CRC - IFRS9
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
10Aguais & Associates Ltd. CRC - IFRS9
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
11Aguais & Associates Ltd. CRC - IFRS9
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
Predicted by Credit-
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
Predicted by Credit-
Cycle Model
Legacy Credit
Models
Predicted by Credit-
Cycle model
Legacy Credit
Models
Predicted by Credit-
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
12Aguais & Associates Ltd. CRC - IFRS9
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
13Aguais & Associates Ltd. CRC - IFRS9
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
14Aguais & Associates Ltd. CRC - IFRS9
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
15Aguais & Associates Ltd. CRC - IFRS9
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 >
16Aguais & Associates Ltd. CRC - IFRS9
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 >
17Aguais & Associates Ltd. CRC - IFRS9
-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
18Aguais & Associates Ltd. CRC - IFRS9
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
19Aguais & Associates Ltd. CRC - IFRS9
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
PD/LGD/EAD Term Structures
Need PIT Conversion & Forward Looking
PD/LGD/EAD Term Structures
20Aguais & Associates Ltd. CRC - IFRS9
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
21Aguais & Associates Ltd. CRC - IFRS9
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
22Aguais & Associates Ltd. CRC - IFRS9
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
23Aguais & Associates Ltd. CRC - IFRS9
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
24Aguais & Associates Ltd. CRC - IFRS9
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
25Aguais & Associates Ltd. CRC - IFRS9
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
26Aguais & Associates Ltd. CRC - IFRS9
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
27Aguais & Associates Ltd. CRC - IFRS9
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’)
28Aguais & 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
29Aguais & Associates Ltd. CRC - IFRS9
Point-in-Time Methodology & IFRS9 Challenges
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)
30Aguais & Associates Ltd. CRC - IFRS9
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
31Aguais & Associates Ltd. CRC - IFRS9
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