Measuring Lifetime Expected Credit Losses Using … Lifetime Expected Credit Losses Using PIT/TTC...
Transcript of 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.
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
-
3333Aguais & Associates Ltd.
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
-
4444Aguais & Associates Ltd.
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
-
5555Aguais & Associates Ltd.
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
-
6666Aguais & Associates Ltd.
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
-
7777Aguais & Associates Ltd.
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
-
8888Aguais & 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
-
9999Aguais & Associates Ltd.
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
-
10101010Aguais & Associates Ltd.
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
-
11111111Aguais & Associates Ltd.
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: Moodys KMV, Aguais/Forest research
Current Credit Models Are Blind to Credit Cycles 20% Prediction is Therefore Powerful
-
12121212Aguais & Associates Ltd.
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
-
13131313Aguais & Associates Ltd.
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
-
14141414Aguais & Associates Ltd.
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
-
15151515Aguais & Associates Ltd.
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
-
16161616Aguais & Associates Ltd.
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 >
-
17171717Aguais & Associates Ltd.
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 >
-
18181818Aguais & Associates Ltd.
-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
-
19191919Aguais & Associates Ltd.
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
-
20202020Aguais & Associates Ltd.
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
-
21212121Aguais & Associates Ltd.
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
-
22222222Aguais & Associates Ltd.
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
-
23232323Aguais & Associates Ltd.
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:
-
24242424Aguais & Associates Ltd.
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
-
25252525Aguais & Associates Ltd.
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
-
26262626Aguais & Associates Ltd.
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
-
27272727Aguais & Associates Ltd.
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
-
28282828Aguais & 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
-
29292929Aguais & Associates Ltd.
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(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)
-
30303030Aguais & Associates Ltd.
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
Retail Commercial - Wholesale
Integrated IFRS9 PIT Batch Analytics
Retail Commercial - Wholesale
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
-
31313131Aguais & Associates Ltd.
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