16.30 Giorgio

Cristiano Zazzara, Ph.D.
IFRS 9 and Macroeconomic Managing Director

Risk Services
S&P Global Market Intelligence
Forward-Looking Scenarios Giorgio Baldassarri, Ph.D.

Vice President
Risk Services
S&P Global Market Intelligence
Permission to reprint or distribute any content from

this presentation requires the prior written
approval of S&P Global Market Intelligence. Not for
distribution to the public. Copyright © 2018 by S&P
Global Market Intelligence. All rights reserved.
RISK EMEA 2018
London – April 24, 2018
Permission to reprint or distribute any content from this presentation requires the prior written approval of S&P
Global Market Intelligence. Not for distribution to the public.
Agenda
 Executive summary
I. Overview of IFRS 9
II. An Example Of Expected Credit Loss Calculation Under IFRS 9
III. Conditioning Credit Risk Transitions on Macro-Economic Scenarios
IV. Inclusion of Macroeconomic Forecasts into Multi-year PDs
V. Final Remarks
 Further Reading
 Appendix:
 Wrap-up on our IFRS 9/CECL offering
*The activities of S&P Global Market Intelligence are independent and separate from S&P Global Ratings. S&P Global Ratings does not contribute to or
participate in the creation of credit scores generated by S&P Global Market Intelligence. Lowercase nomenclature is used to differentiate S&P Global Market
Intelligence PD credit model scores from the credit ratings issued by S&P Global Ratings.
Global Market Intelligence. Not for distribution to the public. 2
Executive Summary
 During the financial crisis, the G20 tasked global accounting standard setters to work towards the objective of creating a
single set of high-quality global standards
 In response to this request, the International Accounting Standards Board (IASB) and Financial Accounting Standards Board
(FASB) began to work together on the development of new financial instruments standards, such as IFRS 9 and CECL
(Current Expected Credit Loss Model)
 IFRS 9 has replaced the current IAS 39 starting from January 1, 2018, and is implemented in several jurisdictions (including
Europe and Canada)
 One of the significant changes introduced by IFRS 9 is that entities are required to adjust their historical credit loss
experience to reflect forecasts of future macroeconomic conditions
 In this context, we propose both a multi-state (transition and default rates) and a default mode approach to condition credit
quality changes on macroeconomic factors
 Regression modelling and Markov chain theory are adopted to translate macroeconomic forecasts into lifetime expected credit
losses
Overview of IFRS 9
The Main Requirements Of IFRS9
• Classifications of Financial Instruments (this is an irrevocable election

at recognition)
− At Amortized Cost: loans, trade receivables, etc.,
− At Fair Value through Other Comprehensive Income (FVOCI): This is the old “Available for Sale” asset category
 Debt investments held to capture yields, but with the possibility of selling them to rebalance the portfolio in terms of, for example, risk, duration, liquidity
− At Fair Value through Profit & Loss (FVTPL): This is the so-called Trading Book
 Debt instruments held for trading purposes, where the goal is not to collect cash flows over time, but to maximize advantageous pricing conditions over a
short-time frame
• Impairment: This is the most important change versus IAS 39 (the previous incurred loss model). Expected losses have to
be calculated on performing assets as well, with direct impact on P&L
• Derivatives and Hedge Accounting: Requirements on how to recognize bespoke derivatives hedging. Limitations on macro
hedging policies
Interaction Between IFRS 9 And IFRS 13
Fair value pricing and credit risk impairment differ according to the asset class category
Table 1 – Accounting Treatment of each category of Financial Assets
FAIR VALUE ADJUSTMENT CREDIT IMPAIRMENT

ASSET CATEGORY
(IFRS 13) (IFRS 9)
At Amortised Cost No Yes, recorded in P&L
• Debt Investments: Yes, recorded in

P&L
At FVOCI Yes, recorded in P&L
• Equity investments: No credit
impairment
At FVTPL Yes, recorded in P&L No credit impairment recorded
Source: S&P Global Market Intelligence, elaboration of IASB, “IFRS 9 Financial Instruments”, July 2014. For illustrative purposes only.
Implications For Listed Firms On Specific Asset Classes
Major impact is expected for financial institutions, although large non-financial companies with sizeable investment
portfolios will be affected as well
• Particularly:
− Banks: Loans and other asset classes in the Banking Book + Debt Investments at FVOCI
− Other Financial Institutions (Insurance Companies, Asset Managers): Debt Investments
at FVOCI
− Non-Financial Companies: Trade Receivables + Debt Investments at FVOCI
 However, for trade receivables, non-financial companies could opt for a simplified look-up
table approach that significantly reduces the implementation challenges related to this new accounting standard
• As mentioned before, any assets in the trading books and equities held in the FVOCI categories will be
excluded from the credit impairment test
Key Features Of IFRS 9 Credit Provisions
In July 2014, the International Accounting Standard Board (IASB) published the final text
of the IFRS 9 rules, which also includes the new “Expected Loss” impairment model
STAGE 1 STAGE 2 STAGE 3
Impairment No significant Significant Objective evidence
Criteria deterioration in deterioration in of impairment at
credit risk since credit risk since reporting date
initial recognition, or initial recognition,
low credit risk at but no evidence of
reporting date impairment
Probability of Expected over a Expected up to PD = 100%

Default (PD) 12-month horizon Contractual Maturity (“absorbing state”)
Expected Loss 12-month Expected Life-time Expected Life-time Expected

Allowance Credit Loss Credit Loss Credit Loss,
only Loss Given
Default (LGD) to be
estimated
Change in credit risk since initial recognition
Note: Exposures can move back from stage 2 to stage 1. Since default is an “absorbing” state, it is rare for exposures in
stage 3 to move back to stage 2 or 1.
Source: IASB, July 2014.
How To Calculate IFRS 9 Credit Loss Provisions
STAGE 1
STAGE 2
No Indicators of a significant increase in credit risk
• A downgrade of a borrower by a recognised credit rating
No Significant agency, or within a bank’s internal credit risk system
>30 days
increase in • An increase larger than a specified threshold in the average
arrears?
credit risk? lifetime Probability of Default (PD) over the remaining life of the
financial instrument
Yes • Credit measures such as warning signals and watch lists result in a
Yes reassessment of the credit rating
• For retail, delinquency on obligations with the bank or on bureau
profiles will trigger stage transition
STAGE 2
…With 30 days past due rebuttable presumption
No
STAGE 3
No Events Events indicating default
>90 days
indicating
arrears? • Bankruptcy of financial reorganisation
default
• Breach of contract (past due / default)
Yes • Borrower in significant financial difficulty
Yes • Disappearance of active market for financial asset
• Purchase of financial asset at deep discount reflecting incurred
credit losses
STAGE 3
…With 90 days past due rebuttable presumption
Source: S&P Global Market Intelligence elaboration of IASB (2014). For illustrative purposes only.
Focus On Credit Impairment: A Three-Stage Approach
IFRS 9 proposes a three-stage approach for the recognition of impairment losses for
financial instruments:
• Performing: Low credit risk at initial recognition (first time in the book). Usually meant as
asset at the “Investment Grade” rating level
− Accounting recognition: 12-month expected credit losses
New credit loss provision – Material Impact
• Underperforming: Significant deterioration of credit risk versus the initial recognition phase,
such as 30-day past due for loans and trade receivables, or asset at “speculative grade”
level, or asset moving from the “investment” to the “speculative grade” area
− Accounting recognition: Lifetime expected credit losses
New credit loss provision – Material Impact
• Non-performing: assets with objective evidence of declared credit impairment at the

reporting date (official default by a rating agency, bankruptcy, 90-day past due Basel default
definition, etc., )
− Accounting recognition: Lifetime expected credit losses
Not materially different from IAS 39

An Example Of Expected Credit Loss
Calculation Under IFRS 9
Estimation of PDs under IFRS 9
Banks need to adjust their current estimation of PDs:
– Internal ratings are usually calibrated on a Through-The-Cycle (TTC) basis, therefore the
resulting PDs don’t reflect the current and future conditions of the credit cycle
– The risk horizon for PDs should extend beyond 1 year in order to estimate lifetime credit risk
– Macroeconomic forward looking scenarios have to be explicitly included
1 2 3
Assess Credit Quality Calibration of Point-In- Derivation of PD Term
of Counterparty Time PDs Structure
• Internal Rating Model • Adjustment of Through-The- • Historical default experience

• External rating if available Cycle PDs (including vintage analysis)
• External vendor model • Recent Default experience • Benchmarks from Industry or
• Market-based indicators Ratings default studies
Macroeconomic forward looking forecast overlay

• Use of experienced credit judgement
• Macro-econometric model (albeit the Basel Committee recognises it may not always be possible to demonstrate a
strong link in formal statistical terms between macroeconomic factors and credit risk of some exposures)
Expected Loss Calculation Under IFRS 9 :
What’s the Impact of 12-month vs. Lifetime?
An example focused on the Probability of Default (PD)
Methodology: PD/LGD approach, under a discounted cash-flow model. Internal or vendor-based

methodologies to assess credit quality of counterparty’s exposures, or external ratings if available
• Stage 1: At initial recognition, calculate 12-month expected credit loss

(EAD * PD * LGD) / (1+ Effective Interest Rate)
• Stage 2: If significant increase in credit risk (30-day past due, transition to speculative grade, overlay
by management based on idiosyncratic and macroeconomic conditions), calculate lifetime
expected credit loss - Need to estimate the full term structure of PDs until maturity
( 𝑡 EADt ∗ Marginal PDt ∗ LGDt) / (1+ Effective Interest Rate)t
• Stage 3: Defaulted exposures (90-day past due, official default, overlay by management), calculate
lifetime expected credit loss – No need to estimate PD, since this is equal to 100%
EAD * PD=100% * LGD
Expected Loss From 12-Month To Lifetime: An Example
Provisioning from 12-month to lifetime expected credit loss
A bank originates a loan of $1M. (EAD) with a 5-year maturity. Risk parameters have been assessed as follows:
Internal Rating = equivalent to an S&P Global Rating of B; LGD = 45%; Term structure of PDs = derived from the
previous recalibration based on TTC Default Rates and CDS proxy spreads. No transaction costs, no optionalities.
Time (Years) 1 2 3 4 5
EAD 1,000,000 1,000,000 1,000,000 1,000,000 1,000,000
Effective Interest Rate 5% 5% 5% 5% 5%
Discounting Factor (DF) 0.95 0.91 0.86 0.82 0.78
Cumulative Probability of Default
3.89% 8.80% 12.82% 15.74% 17.93%
(PD cum)
Marginal Probability of Default (PD) 3.89% 4.91% 4.02% 2.92% 2.19%
LGD 45% 45% 45% 45% 45%
Expected Loss (EAD* PD * LGD) 17,498 22,089 18,104 13,123 9,875
Discounted Expected Loss
16,665 20,035 15,639 10,797 7,737
(EAD* PD * LGD*DF)
12-month Expected Loss 16,665 - IFRS 9 (Stage 1)

Lifetime Expected Loss 70,873 4.3x IFRS 9 (Stage 2) and CECL (Performing)
Lifetime Expected Loss 450,000 IFRS 9 (Stage 3) and CECL (Defaulted)
Source: S&P Global Market Intelligence, For illustrative purposes only.
Expected Loss From 12-Month To Lifetime: An Example
The impact of moving from 12-month to lifetime expected credit loss depends not only
on Maturity, but also on Credit Quality
• The previous example has been extended to include the following aspects:
 Two additional exposures rated “A” and “CCC”: underlying Default Rates always based on S&P Global
Ratings
 Maturity extension up to 10-year: EAD, Effective Interest Rate, LGD assumed to be constant beyond the
5-year horizon
• As expected, Lifetime Expected Losses increase

with the Maturity of the Exposures:
Longer-term exposure = Higher Expected
Credit Losses
• However, the increase is a function of the

Credit quality of the exposures (in this case,
represented by external ratings):
– The worse the credit quality, the lower the
impact of the Lifetime approach (see the
impact on the “CCC” rated exposure)
– And vice versa (see the impact on the “A”
rated exposure)
– For lower credit quality exposures a maturity
extension has a lower impact than for higher
credit quality exposures
• Interesting Results: Higher credit quality

exposures more impacted by the Lifetime
approach
Sources: S&P Global Market Intelligence and S&P CreditPro®. For illustrative purposes only.
Conditioning Credit Risk Transitions on
Macro-Economic Scenarios
How to include Macroeconomic forecasts
A range of possible scenarios should be evaluated
 IFRS 9 requirements:
 At each reporting date, an entity assesses whether the credit risk on a financial
instrument has increased significantly since initial recognition. When making the
assessment, 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
 However, in assessing significant increases in credit risk, the official text of IFRS 9 allows for
an operational simplification: “[…] changes in the risk of a default occurring over the next 12
months may be a reasonable approximation, unless circumstances indicate that a lifetime
assessment is necessary.” [IFRS9 – B5.5.13]
 An entity shall measure expected credit losses of a financial instrument in a way that
reflects:
(a) an unbiased and probability-weighted amount that is determined by evaluating a
range of possible outcomes;
(b) the time value of money; and
(c) reasonable and supportable information about past events, current conditions and
forecasts of future economic conditions
How to include Macroeconomic forecasts
Simple macroeconomic scenario modelling could be enough
 IFRS 9 requirements (cont’d):
 When measuring expected credit losses, an entity need not necessarily identify every
possible scenario
 In some cases, relatively simple modelling without the need for a large number of
detailed simulations of scenarios could be enough; in others, the identification of
scenarios and their estimated probability will probably be needed
 In general, scenario analysis could be relevant when there is a non-linear relationship
between scenarios and EL or scenarios and changes in the credit risk
 For example, the empirical relationship between Real Estate price growth and related
credit losses is highly non-linear: real estate price declines have a much larger effect on
credit losses than price increases
 The maximum period to consider when measuring expected credit losses is the
maximum contractual period (including extension options) over which the entity is
exposed to credit risk
Examples of EL Calculation under Stages 1 and 2
One Macroeconomic factor and three scenarios considered
 Probability-weighted staging assessment

 At initial recognition, only 12-month Expected Loss (EL) is recognized
 Over time, if a significant credit risk deterioration occurs, Lifetime EL is recognized
Table A - Expected Loss calculation under multiple macroeconomic scenarios : Stage 1, At Initial Recognition
Unemployment rate Scenario Probaility 12-month PD Lifetime LGD EAD 12-m EL Lifetime EL
Upside 4% 30% 4% 7% 55% 1000 22 39
Base case 5% 55% 8% 11% 65% 1000 52 72
Downside 6% 15% 16% 20% 85% 1000 136 170
Weighted 56
Base Case EL ≠ Probability-weighted EL

Table B - Expected Loss calculation under multiple macroeconomic scenarios : Stage 2, Significant Credit Risk Deterioration
Unemployment rate Scenario Probaility 12-month PD Lifetime LGD EAD 12-m EL Lifetime EL
Upside 4% 30% 4% 7% 55% 1000 22 39
Base case 6% 60% 12% 15% 65% 1000 78 98
Downside 8% 10% 25% 30% 85% 1000 213 255
Weighted 77
Source: IFRS (2016), “IFRS 9 Forward-looking information and multiple scenarios”, July.
Macro-Model Concept
Objective: A global model that estimates how the current credit-worthiness of a
corporation will change under future macro-economic scenarios.
Immediate Application: Calculate PD component for IFRS9 and CECL purposes.
INPUTS OUTPUTS
Company
Credit-Worthiness
- today
Credit-worthiness –
for next year(s)
GDP, Inflation rate,
Interest rate, etc
- next year(s)
scenario
Source: S&P Global Market Intelligence. For illustrative purposes only.
Analytic Framework Comparison S&P Global Market Intelligence
Risk Transitions Matrix Approach (RTM) Exponential Density Functions

Z
Generalization of logistic regression to 22
TTC states: 1,2, etc., default
Q1 2000
Q2 2000
Q3 2000
Q2 2016
• Granular approach, optimized on each

rating category.
time
1. Zt+1 = F (Zt, macrost+1)
2. Scoret+1 = G (Zt+1) Scoret+1 = F’ (Scoret , macrost+1)
• Industry standard • Calibrated on individual Ratings transitions

•
PROS • Simple to implement and intuitive
•
Granular methodology
Outperforms the RTM approach
• Requires large dataset • Requires large dataset, but expectation
•
CONS May be hard to calibrate properly [single-factor
model, with 2-step optimization]
value helps compensating for
incomplete/noisy dataset.
S&P Global Ratings does not contribute to or participate in the creation of credit scores generated by S&P Global Market Intelligence.
21
Macro-Model Design
“Regions” Macro-variables^ Industry Segmentation**
United States • Real GDP growth 24 separate clusters
• Unemployment rate change IG/SG indicator
• DowJones Index change
• BBB Corp - 5 yrs Treasury yield change’
• House price Index change
• Oil Price’
• Mortgage Rates – Prime Rates change
Canada Same as above (for Canada) 24 clusters (indicators)
IG/SG indicator
Europe • EU STOXX Index change 24 clusters (indicators)
(10 sub-regions) • FTSE100 Index change* IG/SG indicator
• EU 10yrs-3mnths Interest rate spread
• EU28 real GDP growth
RoW Out of scope for the moment
^ Subset of CCAR variables, or ECB variables. * To account for Brexit, we treat UK as a separate sub-region and include a UK-specific variable for UK. ** We will also include
industry clusters not yet covered in CreditModel, such as NBFI and RE, so we will be ready, once CM_NBFI will be trained. ‘1 year lag ‘ For Energy industry only.
Source: S&P Global Market Intelligence, as of 30.11.2017.For illustrative purposes only.
Modelling choices and features

• Due to characteristics of Large Corporates (few defaults), we used S&P Global Ratings’ issuer rating
transitions.
• Depends on data availability, need to account for different industry/regional economic cycles.
• Recession Indicator: Users can switch on/off, to simulate scenarios with “recession-like behaviour”.
S&P Global Ratings does not contribute to or participate in the creation of credit scores generated by S&P Global Market Intelligence.
22
Model Performance - Overall
- We compare observed average rating vs modelled (weighted average) score.
- The weighted score represents the expected value (average) based on the 22 model
transition probabilities (weights) and possible scores (states) between 1 and 22 (see
sketch below).
Weighted Scores vs Next Rating Weighted Scores vs Next Rating

12 (US) 4000 10 (EU) 1500
9 1450
3000 1400
10 9 1350
8 1300
2000 1250
8 1200
8 1150
1000 7
1100
7 1050
6 0 6 1000
Obs Obs
Source: S&P Global Market Intelligence. Data as of April 2017.For illustrative purposes only.
Expected
Value
S&P Global Ratings does not contribute to or participate in the creation of credit scores generated by S&P Global Market Intelligence..
23
Model Performance – Downgrades (EU)
Recession Recession
Indicator 1 Indicator 2
• EU Downgrades by 1 category 20.0% 1350

1300
(eg: aaa to aa+,aa or aa-) 15.0% 1250
1200
10.0%
1150
1100 Obs
5.0%
• How does it look if we remove 1050 Actual
0.0% 1000 Model
the recession indicators?
20021
20031
20041
20051
20061
20071
20081
20091
20101
20111
20121
20131
20141
20%
15% 25.0%
IG 1200
10% 1000
20.0%
5% 800
0% 15.0%
600
20021
20031
20041
20051
20061
20071
20081
20091
20101
20111
20121
20131
20141
10.0%
400 Obs
5.0% 200 Actual
0.0% 0 Model
20021
20031
20041
20051
20061
20071
20081
20091
20101
20111
20121
20131
20141
24
Application to IFRS9/CECL: PD Calculation
Macro Economic Scenario
EuroStoxx 50 Index Change Real GDP Growth 10Y - 3M Spread Stress
-5% -1% 0.01 NO
INPUTS Company Specific Inputs

OUTPUTS
Country Industry Current Score
Germany Energy bbb+
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Current
Score
MODEL ENGINE Score within 1 year from now Expected
Numerical Score
Expected 1 yr
PD*
Expected
Lifetime PD**
aaa aa+ aa aa- a+ a a- bbb+ bbb bbb- bb+ bb bb- b+ b b- ccc+ ccc ccc- cc c d
aaa 79% 11% 3% 1% 0% 0% 0% 0% 0% 1% 0% 1% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 1.7 0.01% 0.01%
aa+ 0% 74% 14% 10% 3% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 2.4 0.01% 0.01%
aa 0% 1% 75% 12% 8% 3% 0% 1% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 3.4 0.02% 0.03%
aa- 0% 0% 6% 75% 16% 2% 1% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 4.2 0.03% 0.05%
a+ 0% 0% 0% 4% 81% 11% 2% 1% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 5.2 0.05% 0.07%
a 0% 0% 0% 0% 4% 82% 9% 2% 2% 1% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 6.2 0.07% 0.10%
a- 0% 0% 0% 0% 0% 5% 83% 8% 2% 1% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 7.1 0.11% 0.25%
bbb+ 0% 0% 0% 0% 0% 1% 0% 85% 12% 1% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 8.1 0.18% 0.43%
bbb 0% 0% 0% 0% 1% 1% 2% 3% 84% 6% 1% 2% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 9.1 0.28% 0.69%
bbb- 0% 0% 1% 0% 0% 1% 0% 2% 10% 78% 4% 1% 2% 0% 0% 0% 0% 0% 0% 0% 0% 0% 10.1 0.43% 1.10%
bb+ 0% 0% 0% 0% 0% 0% 0% 0% 0% 16% 70% 8% 2% 3% 0% 0% 0% 0% 0% 0% 0% 0% 11.1 0.72% 1.76%
bb 0% 0% 0% 0% 0% 0% 0% 1% 7% 3% 7% 71% 7% 2% 1% 0% 0% 0% 0% 0% 0% 0% 12.2 1.02% 2.82%
bb- 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 1% 16% 64% 11% 5% 2% 0% 0% 0% 0% 0% 0% 13.1 1.85% 4.51%
b+ 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 4% 9% 64% 16% 5% 1% 1% 0% 0% 0% 0% 14.2 3.10% 7.21%
b 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 2% 3% 6% 71% 12% 3% 0% 0% 1% 0% 1% 15.1 4.93% 11.54%
b- 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 1% 16% 62% 10% 6% 0% 1% 0% 3% 16.2 8.28% 18.47%
ccc+ 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 34% 56% 0% 0% 9% 0% 0% 17.1 11.40% 25.00%
ccc 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 22% 0% 0% 44% 11% 11% 11% 0% 0% 0% 0% 18.1 20.69% 25.12%
ccc- 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 100% 0% 0% 0% 19.1 29.99% 42.13%
cc 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 67% 0% 33% 20.6 65.52% 69.36%
c 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 100% 0% 21.0 76.61% 78.02%
• Missing ratings: filled with 99.9% (e.g.: values in red circles)
*Expected PD is adjusted for Credit Cycle and
• Transitions to “d” state can be very limited: to calculate the PD, take
made monotonic
expected score and map it to the historical default rates, then include
** With optional adjustment by market 25
Credit Cycle Adjustment and market expectations expectations for country/industry
Model Scope
Coverage Region
Financial and non-Financial Large Corporates* US, CA, Europe
Small and Medium Enterprises** US, CA, Europe
Sovereigns, Commercial Real Estate, Municipalities, Project Finance, Retail 
* Includes REITS and Real Estate Developers.

** Despite trained on Large Corporates, the rating transition behaviour is assumed to be similar to that of Large Corporates
Use Case Comment
IFRS9 Expected Credit Loss 1 yr PD and Lifetime Calculation
US CECL Lifetime calculation
Stress Testing (benchmark model) May suffer from granular approach (see A&L slide)
Asset Class Model

Bonds CreditModel™
(Corporate) Loans (intercompany financing) CreditModel / PD Model Fundamentals
Trade Receivables*** 
*** Simplified approach with “lookup table” is accepted. Source: S&P Global Market Intelligence. For illustrative purposes only.
Inclusion of Macroeconomic Forecasts
into Multi-year PDs
How to derive multi-year PDs adjusted for macro forecasts
 Incorporation of macroeconomic forecasts for future years
 The 1-year PD is the key ingredient for exposures at initial recognition (Stage 1)
- As shown in the previous example, multi-year PDs conditioned on macroeconomic
forecasts don’t enter into the calculation of EL
 However, for stage 2 exposures, it is required to include lifetime PDs conditioned on

macroeconomic forecasts
 A model that conditions PDs over a 1-year horizon can be used to derive PDs in
subsequent periods, taking into account macroeconomic forecasts available at the
reporting date
 Forecasts of macroeconomic factors (such as GDP, Unemployment Rate, etc…) are available
from public (ECB, Eurostat and others) and private (economics departments of investment
banks and vendors) sources
Estimation of Multi-year PDs via a Mortality Rate Approach
Analysis of PDs and Probabilities of Survival over time
 Under the hypothesis of risk independence between periods, multi-year PDs can be derived
using a Mortality Rate approach (this has been pioneered by Rating Agencies – and also
by Prof. Ed Altman)
 Example: if the 1-year PD is equal to 3%, the Probability of Survival is equal to 97% (that is,
100%-3%)
• Assuming the exposure has survived in the previous year, the 2-year Probability of
Survival is equal to: 97%*97%= 94.09%. Then, the 2-year Probability of Default is equal
to: 100%-94.09% = 5.91%
• Similarly, the derived 3-year PD will be equal to 8.73%
 In formula, this approach can be generalized as follows:

𝑛
𝑃𝐷 𝑡 + 𝑛 = 1 − 1 − 𝑃𝐷(𝑡 + 1)
where n= year
 Unfortunately, this approach doesn’t take into account the economic forecasts in
subsequent periods at the time of provisioning. For IFRS 9 purposes, an adjustment is
therefore needed
Adjustment of marginal PDs based on macroeconomic forecasts for future years
 Under a Mortality Rate approach, a simple and intuitive way to estimate PDs beyond the 1-
year horizon is to condition the marginal PDs on the economic forecasts for future years
 As shown before, a statistical model can be estimated to assess the historical relationship
between changes in default and migration rates and changes in macro factors
 Focusing on default rates only (in line with the IFRS 9 requirements), we use a simple
relationship between historical changes in default rates and changes in GDP to adjust the
1-year PD (this approach can be easily generalized to several macro factors)
 Let’s assume that our (simple) regression model takes the following form:
∆𝐷𝑒𝑓𝑎𝑢𝑙𝑡 𝑅𝑎𝑡𝑒 = 𝛾 ∗ ∆𝐺𝐷𝑃
where the estimated coefficient 𝛾 = −0.5
 With this coefficient of sensitivity for GDP growth, we adjust the previous mortality rate
formula incorporating the GDP forecasts for future years into the marginal 1-year PDs
 Here is the “adjusted” formula for multi-year PDs (for each scenario):
𝑃𝐷 𝑡 + 𝑛 = 1 − 1 − 𝑃𝐷 𝑡 + 1 + ∆𝑡+𝑘 ∗ 𝛾 ∗ 𝑤
𝑘=1
where:
 ∆𝑡+𝑘 = forecast change in GDP
 𝛾 = the coefficient of the previous regression (that is, -0.5)
 𝑤 = weighting for the macroeconomic scenario (adjustable by the user)
 In the next slide, we provide an example of this calculation for an asset with a lifetime of 3
years, under both a base and an adverse scenario
Example of calculation over a 3-year risk horizon
 At the reporting date, at time t, the realized GDP Growth is 2%
 According to various economic forecast sources, we create two scenarios of GDP growth
over the next 3 years: a Base and an Adverse case. The changes (D) versus the current
macro factor level are reported below:
GDP Growth, t 2%
Time period Baseline D t+k Adverse D t+k

t+1 2.20% 0.20% 1.00% -1.00%
t+2 2.30% 0.30% 0.00% -2.00%
t+3 2.50% 0.50% -0.40% -2.40%
Source: S&P Global Market Intelligence. For illustrative purposes only (2017).
 The implementation of the previous “adjusted” formula yields the following results (with w
= 1)
PD with
PD with
Time period macro forecast adjust.
no adjustment
Baseline Adverse
t+1 3.0% 2.9% 3.5%
t+2 5.9% 5.7% 7.4%
t+3 8.7% 8.3% 11.3%
Source: S&P Global Market Intelligence. For illustrative purposes only (2017).
Final Remarks: Challenges Ahead For Firms
 Under IFRS 9, entities are required to estimate forward-looking Credit Losses,
explicitly accounting for future macroeconomic conditions
 Expected Credit Losses should be computed over a range of probability-weighted
economic scenarios
 In this context, one of the main measurement challenges is the estimation of multi-
year (lifetime) PDs that take into account macroeconomic forecasts
 However, entities are not required to incorporate detailed forecasts of future economic
conditions over the entire lifetime of the exposure [IFRS 9. B5.5.50]: it can be assumed
that, beyond a “forecastable” risk horizon, economic conditions revert to their long-term
average
 Entities should make use of experienced credit judgment when selecting future
macroeconomic scenarios and then assigning related probabilities of occurrence
 To support this judgmental process, we propose a quantitative approach to condition
credit risk losses on future macroeconomic factors. Such a framework can also facilitate
disclosure requirements and communication to stakeholders
The IFRS 9 framework poses substantial modelling and operational challenges to
entities adopting this upcoming accounting standard
- Good governance and control is required to manage this new credit loss estimation
process
Further Reading
 Basel Committee on Banking Supervision (2015), “Guidance on Credit Risk and Accounting for Expected Credit
Losses”, December (www.bis.org)
 Deloitte (2016), “Sixth Global IFRS Banking Survey”, May (www.deloitte.com)
 Ernst & Young (2014), “Applying IFRS – Impairment of financial instruments under IFRS 9”, December (www.ey.com)
 Financial Accounting Standards Board- FASB (2016), “Financial Instruments – Credit Losses (Topic 326)”, June
(www.ifrs.org)
 International Accounting Standards Board- IASB (2014), “IFRS 9 Financial Instruments”, July (www.ifrs.org)
 International Accounting Standards Board- IASB (2016), “IFRS 9 Forward-looking information and multiple
scenarios”, IFRS Foundation, July (www.ifrs.org)
 S&P Global Ratings (2017), “Default, Transition, and Recovery: 2016 Annual Global Corporate Default Study and
Rating Transitions”, April (www.spglobal.com)
 Vanek T, D. Hampel (2017), “The Probability of Default under IFRS 9: multi-period estimation and macroeconomic
forecast”, Acta Univ. Agric. Silvic. Mendelianae Brunensis, Issue 2, April. (https://acta.mendelu.cz)
APPENDIX
Wrap-Up On S&P Global Market Intelligence’s
IFRS 9 and CECL Bottom-Up Offering
IFRS 9 STAGE 1 STAGE 2 STAGE 3
PERFORMING DEFAULTED
CECL (STAGE 1) (STAGE 2)
• External Ratings (at issuer and issue level)
• Credit Scorecards (at counterparty and facility level)
• CreditModel™ (at counterparty level)
No estimation required (PD =
Probability of Default • PD Fundamental (at counterparty level)
100%)
(PD) • CDS Proxy spreads (at counterparty level)
• CreditPro® Database (actual default rates statistics)
• Macroeconometric model for Europe
under development
• Recovery Rate Scorecards (at counterparty and facility level)

Loss Given Default • CreditPro® Database (actual recovery rates statistics)
(LGD) • Top-down statistical model for Europe under development (LossStats Model for US assets available
– Econometric forecasts of LGD Values)
Data Warehouse • SNL Banker (a reporting system that securely integrates data form several internal sources, such as
and bank’s core processors, general ledger, and other systems for credit risk and other business
Reporting system analysis)
MODEL COVERAGE
• Scorecards: Sovereigns, banks, insurance firms, other financial institutions, corporates, specialized lending, and commercial real
estate
• Quantitative Models: Banks, insurance firms, corporates, and SME-corporates
Assumptions & Limitations of Macro-Scenario Model
Assumption Limitation Mitigation
Rating transitions are modelled year-on-year. May use Markov assumption, and take 4th
root of rating transition matrix generated by
model to obtain quarter-on-quarter change:
this matrix may not exist!
The rating transition happens any time within the 1 Associated default rate is based on historical PD value will incorporate forward looking
year time horizon, not at the end of the period data, and is not modelled separately. market expectations (suitable for
CECL/IFRS9)
Correlation between transitions of different rating Modelling framework ensures to capture
categories is not elicited explicitly (like, instead, in rating transition behaviour in detail
RTM approach, despite with low final R2)
Missing rating transitions are assumed to be Rating transitions may lead to counterintuitive PD outputs for IFRS9/CECL are corrected
stable (99.9% probability) behaviour (eg: all rating categories deteriorate for monotonicity:
during a recession, except for some rating • In a negative (positive) scenario, PD
categories) mapped to each rating categories
increases (decreases)
• PD decreases with improving rating
PD values are assumed monotonic vs rating

category and vs macroscenario
Credit Score transitions of SMEs are similar to Default rates of SMEs are larger than for Adjust PDs of SMEs (by the user)
those of large Corporates Corporates
The lifetime PD can be calculated by a rigid shift May not reflect actual evolution of the economy Run model iteratively, using multiple YoY
of the long-term average term-structure economic scenarios, and use output from
each step, as input for the next step. (to be
tested)
Thank You
Cristiano Zazzara, Ph.D. Giorgio Baldassarri, Ph.D.

Managing Director, Risk Services Vice President, Risk Services S&P
S&P Global Market Intelligence Global Market Intelligence
T: +44 (0)20 7176 7454 T: +44 (0)20 7176 3947
cristiano.zazzara@spglobal.com giorgio.baldassarri@spglobal.com
Copyright © 2018 by S&P Global Market Intelligence, a division of S&P Global Inc. All rights reserved.
These materials have been prepared solely for information purposes based upon information generally available to the public and from sources believed to be reliable. No
content (including index data, ratings, credit-related analyses and data, research, model, software or other application or output therefrom) or any part thereof (Content) may
be modified, reverse engineered, reproduced or distributed in any form by any means, or stored in a database or retrieval system, without the prior written permission of S&P
Global Market Intelligence or its affiliates (collectively, S&P Global). The Content shall not be used for any unlawful or unauthorized purposes. S&P Global and any third-
party providers, (collectively S&P Global Parties) do not guarantee the accuracy, completeness, timeliness or availability of the Content. S&P Global Parties are not
responsible for any errors or omissions, regardless of the cause, for the results obtained from the use of the Content. THE CONTENT IS PROVIDED ON “AS IS” BASIS.
S&P GLOBAL PARTIES DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, ANY WARRANTIES OF
MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE OR USE, FREEDOM FROM BUGS, SOFTWARE ERRORS OR DEFECTS, THAT THE CONTENT’S
FUNCTIONING WILL BE UNINTERRUPTED OR THAT THE CONTENT WILL OPERATE WITH ANY SOFTWARE OR HARDWARE CONFIGURATION. In no event shall
S&P Global Parties be liable to any party for any direct, indirect, incidental, exemplary, compensatory, punitive, special or consequential damages, costs, expenses, legal
fees, or losses (including, without limitation, lost income or lost profits and opportunity costs or losses caused by negligence) in connection with any use of the Content even
if advised of the possibility of such damages.
S&P Global Market Intelligence’s opinions, quotes and credit-related and other analyses are statements of opinion as of the date they are expressed and not statements of
fact or recommendations to purchase, hold, or sell any securities or to make any investment decisions, and do not address the suitability of any security. S&P Global Market
Intelligence assumes no obligation to update the Content following publication in any form or format. The Content should not be relied on and is not a substitute for the skill,
judgment and experience of the user, its management, employees, advisors and/or clients when making investment and other business decisions. S&P Global Market
Intelligence does not act as a fiduciary or an investment advisor except where registered as such. S&P Global keeps certain activities of its divisions separate from each
other in order to preserve the independence and objectivity of their respective activities. As a result, certain divisions of S&P Global may have information that is not
available to other S&P Global divisions. S&P Global has established policies and procedures to maintain the confidentiality of certain non-public information received in
connection with each analytical process.
S&P Global Ratings does not contribute to or participate in the creation of credit scores generated by S&P Global Market Intelligence. Lowercase nomenclature is used to
differentiate S&P Global Market Intelligence PD credit model scores from the credit ratings issued by S&P Global Ratings.
S&P Global may receive compensation for its ratings and certain analyses, normally from issuers or underwriters of securities or from obligors. S&P Global reserves the right
to disseminate its opinions and analyses. S&P Global's public ratings and analyses are made available on its Web sites, www.standardandpoors.com (free of charge), and
www.ratingsdirect.com and www.globalcreditportal.com (subscription), and may be distributed through other means, including via S&P Global publications and third-party
redistributors. Additional information about our ratings fees is available at www.standardandpoors.com/usratingsfees.

16.30 Giorgio

Uploaded by

Copyright:

Available Formats

16.30 Giorgio

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

16.30 Giorgio

Uploaded by

Copyright:

Available Formats

Cristiano Zazzara, Ph.D.

IFRS 9 and Macroeconomic Managing Director

Forward-Looking Scenarios Giorgio Baldassarri, Ph.D.

Permission to reprint or distribute any content from

RISK EMEA 2018

London – April 24, 2018

II. An Example Of Expected Credit Loss Calculation Under IFRS 9

III. Conditioning Credit Risk Transitions on Macro-Economic Scenarios

IV. Inclusion of Macroeconomic Forecasts into Multi-year PDs

• Classifications of Financial Instruments (this is an irrevocable election

Table 1 – Accounting Treatment of each category of Financial Assets

FAIR VALUE ADJUSTMENT CREDIT IMPAIRMENT

At Amortised Cost No Yes, recorded in P&L

• Debt Investments: Yes, recorded in

At FVTPL Yes, recorded in P&L No credit impairment recorded

Probability of Expected over a Expected up to PD = 100%

Expected Loss 12-month Expected Life-time Expected Life-time Expected

Change in credit risk since initial recognition

New credit loss provision – Material Impact

New credit loss provision – Material Impact

• Non-performing: assets with objective evidence of declared credit impairment at the

Not materially different from IAS 39

• Internal Rating Model • Adjustment of Through-The- • Historical default experience

Macroeconomic forward looking forecast overlay

Methodology: PD/LGD approach, under a discounted cash-flow model. Internal or vendor-based

• Stage 1: At initial recognition, calculate 12-month expected credit loss

12-month Expected Loss 16,665 - IFRS 9 (Stage 1)

Source: S&P Global Market Intelligence, For illustrative purposes only.

• As expected, Lifetime Expected Losses increase

• However, the increase is a function of the

• Interesting Results: Higher credit quality

 IFRS 9 requirements (cont’d):

 Probability-weighted staging assessment

Base Case EL ≠ Probability-weighted EL

Immediate Application: Calculate PD component for IFRS9 and CECL purposes.

Source: S&P Global Market Intelligence. For illustrative purposes only.

Risk Transitions Matrix Approach (RTM) Exponential Density Functions

• Granular approach, optimized on each

• Industry standard • Calibrated on individual Ratings transitions

Modelling choices and features

Weighted Scores vs Next Rating Weighted Scores vs Next Rating

• EU Downgrades by 1 category 20.0% 1350

INPUTS Company Specific Inputs

* Includes REITS and Real Estate Developers.

Use Case Comment

IFRS9 Expected Credit Loss 1 yr PD and Lifetime Calculation

US CECL Lifetime calculation

Asset Class Model

 Incorporation of macroeconomic forecasts for future years

 However, for stage 2 exposures, it is required to include lifetime PDs conditioned on

 In formula, this approach can be generalized as follows:

 At the reporting date, at time t, the realized GDP Growth is 2%

Time period Baseline D t+k Adverse D t+k

 Deloitte (2016), “Sixth Global IFRS Banking Survey”, May (www.deloitte.com)

• Recovery Rate Scorecards (at counterparty and facility level)

PD values are assumed monotonic vs rating

Cristiano Zazzara, Ph.D. Giorgio Baldassarri, Ph.D.

You might also like