Moodys Analytics Risk Perspectives Risk Finance Accounting

THE CONVERGENCE OF RISK, FINANCE, AND ACCOUNTING VOL 7 | JUNE 2016
THE CONVERGENCE OF RISK, FINANCE, AND ACCOUNTING VOL 7

FEATURED ARTICLES:
ONLINE
IMPLEMENTING AN IFRS 9 SOLUTION 8

Challenges faced by financial institutions
Find additional risk, finance, and accounting articles, interviews, IFRS 9 SURVEY RESULTS 28
and multimedia content at MoodysAnalytics.com/RiskPerspectives Preparedness for and opinions on IFRS 9 guidelines
SOLVING THE COUNTERPARTY DEFAULT

SCENARIO PROBLEM 74
JUNE 2016
A 2016 CCAR case study
MARP_RFAC_COVER_FINAL_Print_edit_0.25spine.indd 1 6/14/2016 4:54:38 PM

FROM THE EDITOR
Welcome to the seventh edition of Risk Perspectives™, a Moody’s to IFRS 9 Impairment and the next issue to its US counterpart, Current
Analytics publication created by risk professionals for risk professionals. Expected Credit Loss standard, finalization of which is expected in the
summer of 2016. Convergence of risk and finance functions is also driven
One of the Seven Sages of Ancient Greece, Periander, has said, “Be by regulatory-mandated improvements to capital and liquidity planning,
moderate in prosperity, prudent in adversity.” This is the mantra that as well as resolution and recovery planning. This is why this theme is also
underlies the actions of global accounting standard setters as they shared by our annual Risk Practitioner Conference, which will take place in
implement some of the biggest changes to bank October this year.
accounting in recent history. The changes to the
impairment standards are a direct response to The structure of this edition is a bit different from
the global financial crisis, when many financial prior ones. You will see that the first section,
institutions built insufficient loan loss reserves Spotlight, is focused squarely on implementation
based only on historical loss experience. The new of the new accounting standards, IFRS 9 and CECL.
accounting standards are aiming to make loan The rest of the magazine is split in three sections.
loss provisions counter-cyclical by incorporating Principles and Practices is a section dedicated to
forecasts of economic conditions into provision case studies and implementation approaches. Here,
calculations. Nancy Michael details the changing landscape of
small business lending and how banks can better
Almost two years into IFRS 9 implementation, with compete with alternative lenders. In Innovation
a looming deadline in 2018, full impacts of the Zone, our “looking ahead” section, Sam Malone
transition are still difficult to measure. Of the three describes an innovative methodology using
elements of the new standard, Hedge Accounting, compound scenarios to assess counterparty risk.
Classification and Measurement, and Impairment, And finally, in Regulatory Review, we look at
the last appears to be most challenging. Our survey guidance and standards, already issued and on the
of regional banks, on page 28, found that 80% of docket, that will affect risk management practices
respondents were still in early stages of planning for in the coming year.
IFRS 9 compliance. Another recent industry survey
conducted by Deloitte found that while some banks We hope you enjoy the articles and the new format.
expect increases in provisions of up to 25%, over half of respondents could As always, I encourage you to take part in this conversation and help us
not quantify impact yet. One thing is clear: complexity of calculations shape future issues of Risk Perspectives by sharing your feedback and
required by the new accounting standard is forcing institutions to rethink comments.
their processes and systems and often reorganize their resources. A
process that was historically managed by an Accounting team alone with Anna Krayn
input from Risk functions, will now be a joint effort between Risk, Finance, Editor-In-Chief
and Accounting. Hence the theme of this edition. We know this is top Senior Director and Team Lead, Capital Planning and Stress Testing
of mind for industry participants and, as a result, we dedicate this issue [email protected]
EDITORIAL Web Content Managers David Little Dr. Anthony Hughes Michael van Steen
Editor-in-Chief Mary Sitzenstatter Katherine Pilnick Anna Krayn Dr. Yashan Wang
Anna Krayn Meghann Solosy Katerina Soumilova Glenn Levine Ed Young
Ed Young Dr. Juan Licari
Managing Editors ADVISORY BOARD Jing Zhang David Little SPECIAL THANKS
Clemens Frischenschlager Michael Adler Emil Lopez Gus Harris
Katherine Pilnick Michelle Adler CONTRIBUTORS Dr. Samuel W. Malone Mary Hunt
Celia Chen Barnaby Black Michael McDonald Robert King
DESIGN AND LAYOUT Dr. Shirish Chinchalkar María C. Cañamero Travis Nacey
Nancy Michael
Art Directors Danielle Ferry Dr. Shirish Chinchalkar Nihil Patel Salli Schwartz
Chun-Yu Huang Clemens Frischenschlager Christopher Crossen Jeffrey Reiser Stephen Tulenko
Yomi Ikotun Jacob Grotta Cayetano Gea-Carrasco Yagmur Uenal
Anna Krayn Christian Henkel
While we encourage everyone to share Risk PerspectivesTM and its articles, please contact [email protected]
to request permission to use the articles and related content for any type of reproduction.
CONTENTS
FROM THE EDITOR 1
Anna Krayn, Editor-in-Chief, introduces this Risk Perspectives edition and its focus on the convergence of Risk,
Finance, and Accounting.
SPOTLIGHT: IFRS 9
Implementing an IFRS 9 Solution: Challenges Faced by Financial Institutions 8
Cayetano Gea-Carrasco and Nihil Patel
Preparing for the New Impairment Requirements: Practitioner's View 11

Christian Henkel and Emil Lopez
IFRS 9 Survey Results 28

María Cañamero, Michael McDonald, and Yagmur Uenal
IFRS 9 Impairment Regulations: Implementation Challenges 34

and Potential Solutions
Christopher Crossen and Dr. Yashan Wang
Probability-Weighted Outcomes Under IFRS 9: A Macroeconomic Approach 42

Barnaby Black, Glenn Levine, and Dr. Juan M. Licari
Complying with IFRS 9 Impairment Calculations for Retail Portfolios 48

Barnaby Black, Dr. Shirish Chinchalkar, and Dr. Juan M. Licari
PRINCIPLES AND PRACTICES

How Banks Can Raise Their Game in Small Business Lending 58
Nancy Michael
INNOVATION ZONE
Are Deposits Safe Under Negative Interest Rates? 66
Dr. Anthony Hughes
Solving the Counterparty Default Scenario Problem: A 2016 CCAR Case Study 74
Dr. Samuel W. Malone
REGULATORY REVIEW
Reading the Tea Leaves of Recent Regulatory Guidance 84
Anna Krayn, Ed Young, and David Little
AnaCredit: A New Approach to Banking Regulatory Compliance 92

Michael van Steen
BY THE NUMBERS 4
SUBJECT MATTER EXPERTS 98
CONNECT WITH US 102
GLOSSARY 103
BY THE NUMBERS
2018
Financial entities must adopt IFRS 9 by January
60%
Despite rapid provisioning, coverage (measured
96%
By late 2015, 96% of surveyed regional
1, 2018. as the ratio of reserves to noncurrent loans) fell banks expected challenges in the IFRS 9
Implementing an IFRS 9 Solution: Challenges below 60% in the fourth quarter of 2009, down Impairment phase.
Faced by Financial Institutions. from a typical 100%. IFRS 9 Survey Results.
Page 8 Preparing for the New Impairment Page 28
Requirements: Practitioner's View.
Page 11
3
Factors influencing vintage-segment
24
Small business borrowers spend an average of
0%
Crossing the threshold from positive to negative
performance can be conceptually divided into 24 hours on paperwork for bank loans. interest rates may stimulate the economy, as
three classifications. How Banks Can Raise Their Game in Small Central Banks are hoping, but it can also cause
Complying with IFRS 9 Impairment Calculations Business Lending. economic challenges.
for Retail Portfolios. Page 58 Are Deposits Safe Under Negative Interest
Page 48 Rates?
Page 66
4 MOODY’S ANALYTICS RISK PERSPECTIVES

80%
By late 2015, 80% of regional banks surveyed
12
In some cases, firms can use changes in
2
For most firms, IFRS 9 guidelines suggest using
had not initiated work or were still focusing on 12-month probability of default as an at least two economic scenarios for impairment
gap analysis to comply with IFRS 9 guidelines. approximation of the lifetime default risk calculations.
IFRS 9 Survey Results. change. Probability-Weighted Outcomes Under IFRS 9: A
Page 28 IFRS 9 Impairment Regulations: Implementation Macroeconomic Approach.
Challenges and Potential Solutions. Page 42
Page 34
8
For eight bank holding companies, the Fed now
3
When implementing recent regulatory guidance,
95
The new AnaCredit regulation proposes that
requires layering a counterparty default scenario there are three common themes to consider. banks report at least 95 data elements regarding
onto standard CCAR exercises, and next- Reading the Tea Leaves of Recent Regulatory exposures’ credit risk.
generation stress tests could feature shocks to Guidance. AnaCredit: A New Approach to Banking
bank interconnectedness. Page 84 Regulatory Compliance.
Solving the Counterparty Default Scenario Page 92
Problem: A 2016 CCAR Case Study.
Page 74
THE CONVERGENCE OF RISK, FINANCE, AND ACCOUNTING | JUNE 2016 5

SPOTLIGHT: IFRS 9
IMPLEMENTING AN IFRS 9 SOLUTION:
CHALLENGES FACED BY FINANCIAL
INSTITUTIONS
By Cayetano Gea-Carrasco and Nihil Patel
IFRS 9 is a new international standard set forth to address the

Cayetano Gea-Carrasco
Managing Director – Advisory weaknesses of IAS 39. This article provides an overview of the new
Services standard and analyzes the major challenges financial institutions will
face in ensuring compliance. While there is still uncertainty in terms
Cayetano Gea-Carrasco works with financial of implementation approaches, we believe IFRS 9 adoption will lead
institutions to address their technology and to a more efficient and lower-risk financial system.
enterprise risk management needs. Previously, he
held leadership positions at various institutions and
global banks. He is a regular speaker at international As part of the response to the last financial principle-based approach replaces existing
conferences and has published several articles in crisis, the International Accounting Standards rule-based requirements which are complex
the areas of risk management, financial technology,
Board (IASB) recently issued IFRS 9 to resolve and often difficult to apply.
and derivatives pricing. Cayetano holds a BSc. and
MSc. in Telecommunication Engineering, a master’s the weakness of IAS 39. Under IAS 39, incurred
»» Impairment: Under IFRS 9, the expected
in Economics and Finance, and an MSc. in Financial loss resulted in credit loss recognition that was
credit loss (ECL) model will require more
Mathematics, with distinction, from King’s College “too little, too late.” Improvements under IFRS
timely recognition of credit losses compared
London.
9 include a logical model for the classification
with the incurred loss model of IAS 39. The
and measurement of financial instruments, a
new standard requires entities to account for
forward-looking expected credit loss impairment
expected credit losses using forward-looking
model, and a substantially reformed approach to
information and lowers the threshold for
hedge accounting.
Nihil Patel recognition of full lifetime expected losses.
Senior Director – Product The new standard has a wide reach; it is required
Management
»» Hedge Accounting: IFRS 9 represents a
in more than 100 countries across Europe, the
substantial overhaul of hedge accounting
Middle East, Asia, Africa, the Americas, and
that aligns the accounting treatment with risk
Nihil Patel is a Senior Director within the Enterprise Oceania. While all financial entities must adopt
management activities, enabling entities to
Risk Services division at Moody's Analytics. He serves IFRS 9 by January 1, 2018, many organizations
better reflect these activities in their financial
as the business lead driving our product strategy are targeting parallel runs and impact analyses
statements.
related to credit portfolio analytics. Nihil has broad
of the end-to-end process (including staging
experience in research, modeling, service delivery,
and classification, impairment calculation, and IFRS 9 will drive profit and loss, which will affect
and customer engagement. Prior to his current role,
Nihil spent nine years in the Research organization reporting) by mid-2017. Quantitative Impact earnings. In addition, the standard will materially
leading the Portfolio Modeling Services team as well Studies (QiS), such as European Banking influence financial institutions’ financial
as the Correlation Research team. Nihil holds a MSE in Authority's QiS in Europe, are also accelerating statements, with impairment calculations most
Operations Research and Financial Engineering from
timelines of infrastructure and tactical short- affected. IFRS 9 will lead to changes including
Princeton University and a BS in Industrial Engineering
and Operations Research from UC Berkeley. term solutions for an early assessment of the the following:
impacts on provision levels. »» It will no longer be necessary for a credit
IFRS 9 covers three areas with profound event to occur before credit losses are
implications for financial institutions: recognized.
»» Classification and Measurement: IFRS »» The measurement of allowance of credit loss

9 introduces a logical approach for the will depend on the instrument’s impairment
classification of financial assets driven by cash stages.
flow characteristics and the organization's »» An entity will be required to base its
business model in which an asset is held. This assessment and measurement of expected

SPOTLIGHT: IFRS 9
credit losses on historical, current, and required by IFRS 9 in a cost-effective, scalable

forecast information that is available without way.
undue cost or effort.
»» ECL calculation engine: The calculation
»» Measurement of financial assets will be engine will need to be robust and flexible.
aligned with a bank’s business model, It will need to incorporate facility level and
contractual cash flow of instruments, and be adjusted by credit events. The ECL engine
future economic scenarios. will need to support granular calculations
»» The forward-looking provision framework and expected modeling challenges. It must
will make financial institutions evaluate have built-in data quality checks and reports,
how economic and credit changes alter and must be able to define or choose ad hoc
their capital and provision levels at each economic forecast and scenarios. It must be
subsequent reporting date. capable of modeling or importing PD, LGD,

and EAD term structures and behavioral
An expected credit loss impairment model will metrics affecting cash flows. It must be able
also bring significant challenges for auditors to allocate, optimize, and value collateral and
given the move from a factual credit event as credit risk mitigants.
IFRS 9 provision calculation will require integration of multiple processes

across different areas, including risk, finance, and accounting.
a driver of provision and toward quantitative

»» Risk, finance, and accounting integration:
credit forecasting approaches and staging
Previously separate processes will need to
classification. In turn, this will create significant
integrate, especially from a data and process
risks due to the effect on profitability, capital
perspective.
ratios, fair value measures, and tax rates.
Primarily for these reasons, auditors are actively »» General ledgers reconciliation: Ledgers will
monitoring the development of ECL models need to reflect IFRS 9 calculations and new
and the implementation of IFRS 9 solutions at impairment metrics. Financial institutions
financial institutions. usually have several general ledgers within a

single legal entity.
From a solution design perspective, ability to
track data and manage overrides (for example, »» Computational and performance
due to effect on earnings) will be critical. In requirements: The IFRS 9 forward-looking
addition, multiple processes including those impairment calculation will require higher
in risk, finance, and accounting groups will volumes of data than the current IAS incurred
need to be integrated for the IFRS 9 provision loss model, Basel guidelines, or stress
calculation. In terms of architecture design, testing. Institutions will want to do facility-
an IFRS 9 solution requires multiple layers, level analyses, and calculations leveraging
including: risk and finance data aggregation scalable architecture, such as grid computing
layer, model risk management and workflow processes, will be imperative.
layer, ECL calculation engine, general ledger (GL) »» Tax treatment: IFRS 9 may affect effective
reconciliation layer, and reporting and variance tax rates, as some institutions may leverage
analysis layer. IFRS 9 as a tax optimization tool.
For financial institutions transitioning to IFRS 9, »» Underwriting, risk-adjusted pricing, and

the main architecture design questions involve limits systems: Financial institutions will have
the business, systems, and processes. Main to estimate and book an upfront, forward-
challenges include the following: looking expected loss (either 12-month
»» Systems, processes, and automation: or lifetime) and monitor for ongoing
Systems will need to change significantly deterioration of credit quality.
in order to calculate and record changes »» Risk-adjusted pricing metrics: Pricing

and performance metrics will need to be such as short-term cash facilities and/or
redesigned and/or expanded (e.g., IFRS 9 covenant-like facilities.
based risk-return metrics) in order to be
»» Flexibility of implementation: Exact
aligned to IFRS 9 dimensions and capital
implementation procedures must be able
impacts.
to change depending on data according to
»» Impairment calculation: Institutions must the asset classes and model availability. For
have the ability to calculate a probability- example, if a granular approach should be
weighted impairment that incorporates past applied to a certain part of the portfolio (e.g.,
events, current conditions, and forecasts corporate) or if it should be aggregated (e.g.,
of future economic conditions. In addition, retail).
valuation analysis needs to consider scenario-
»» Gather and store data: Very granular data
specific cash flows.
must be gathered and stored for any new
»» Collateral allocation and valuation: transactions.
Institutions will need to determine how
Given the IFRS 9 requirements in terms of
to incorporate collateral effects on the
classification, measurement, impairment
valuation and computation of cash flows for
calculation, and reporting, financial institutions
impairment calculation purposes.
should expect a need for significant changes to
»» Hedge accounting: IFRS 9 will affect existing
the way they do business, allocate capital, and
documentation, hedging models, and
manage the quality of loans and provisions at
software systems.
origination.
»» Reporting and financial statements: It will be
Financial Institutions will face modeling, data,
necessary to reconcile with other regulatory
reporting, and infrastructure challenges in terms
rules, including Basel 3, the Dodd-Frank Act,
of reassessing the granularity (e.g., facility
and the Foreign Accounting Tax Compliance
level provisioning analysis) and/or credit loss
Act (FATCA). Institutions will need to
impairment modeling approach, and maintaining
reconcile risk and finance data where risk data
consistency in the definition of risk metrics
will be used down to the legal reporting entity
between Basel and IFRS 9 models. Institutions
level. Additionally, impairment values and
will also face challenges in enhancing their
variance changes over reporting horizons will
coordination across finance, risk, and business
need to be included in FINREP reporting by
units. Furthermore, considerable uncertainty
European institutions.
remains regarding the interpretation of the IFRS
»» Operational risk: This type of risk will increase 9 standard and modeling approaches. These will
as a result of changes in systems, models, likely be fine-tuned after QiS and parallel runs
processes, and data. are performed by institutions and regulatory
bodies.
Financial institutions will also face additional
data requirements to meet IFRS 9-related Effectively addressing these challenges will
calculations and ongoing monitoring. These enable boards and senior management to
requirements will lead to related challenges, make better-informed decisions, proactively
including: manage provisions and effects on capital plans,

make forward-looking strategic decisions for
»» Retrieval of old portfolio data: It will be
risk mitigation in the event of actual stressed
necessary to save old data, which will be
conditions, and help in understanding the
especially difficult for transactions originated
evolving nature of risk. In the end, a thoughtful,
many years ago.
repeatable, and consistent capital planning
»» Classification of transactions at origination: and impairment analysis should lead to a more
There is the need to map products if they can sound, lower-risk financial system with more
be categorized prior to the calculation. An efficient institutions and better allocation of
additional effort would be required to identify capital, thus enhancing returns for shareholders.
products that can be considered out of scope,

SPOTLIGHT: IFRS 9
PREPARING FOR THE NEW IMPAIRMENT

REQUIREMENTS: PRACTITIONER’S VIEW
By Christian Henkel and Emil Lopez
This article describes the new standards set forth by the FASB. It
Christian Henkel
covers the history of the ALLL and explains how the recent financial Senior Director – Risk
crisis highlighted the need for new standards. It also suggests how Management
banks should align with the new CECL impairment standards,

including early preparation and core capabilities. Christian Henkel is a Senior Director in the Moody’s
Analytics Risk Measurement Services group. He leads
a national team of risk consultants who work closely
Overview Understanding the Existing Guidance with banks, insurers, and other financial institutions.
An appropriate allowance for loan and lease Setting aside reserves for future bad debts is a Having spent most of his career in commercial banking,
losses (“ALLL”) covers estimated credit losses concept with a long history, as shown in Figure Christian has a unique blend of business and academic
inherent in an institution’s loan and lease 1. The reserve for bad debts became a legitimate experience across the financial services industry
– including expertise in areas such as commercial
portfolio. The ALLL represents management’s tax accounting method with the Revenue Act of
credit and financial analysis, portfolio management,
best estimate of likely net charge-offs that are 1921. Nearly a century later, regulators continue and asset quality. He has a BA and an MBA from
to be realized for a loan or group of loans, given to fine-tune processes for estimating losses and the University of Texas at Dallas in Finance and
facts and circumstances as of the evaluation adequate provisions. Accounting, and he graduated valedictorian from the
Southwestern Graduate School of Banking at SMU.
date.1
Most recently, in 2006, the banking supervisory
On April 27, 2016, the Financial Accounting agencies issued a policy statement on the
Standards Board (FASB) voted to move forward ALLL which remains in place today. Its primary
with a new credit impairment model, known as objectives were to incorporate allowance-related
Emil Lopez
Director – Risk Measurement
In many cases, the ALLL does little to show the true extent of the credit risk
inherent in a bank’s loan portfolio. That is among the most commonly cited
criticisms of the existing rules. Emil Lopez is a Director in the Moody’s Analytics Risk
Measurement Group, where he leads risk modeling
advisory engagements and manages the team's data
quality, risk reporting, and IFRS 9 research. Prior to
the Current Expected Credit Loss model (CECL), developments since earlier policy statements
joining the group, he oversaw operations for Moody's
for the recognition and measurement of credit and to ensure consistency with Generally Analytics Credit Research Database, one of the world's
losses for loans and debt securities. The final Accepted Accounting Principles (GAAP). It also largest private firm credit risk data repositories. Emil
standard is expected to be released in June 2016 expanded the scope of coverage to credit unions. has extensive experience in credit risk modeling and
with implementation beginning in 2018. This reporting, data sourcing, and quality control. Emil has
While the 2006 policy statement is the most an MBA from New York University and a BS in Finance
new standard is far more than an exercise in
comprehensive guidance to date – helping to and Business Administration from the University of
financial accounting and bank regulation. It will Vermont.
establish rules and governance and to bring
replace the current incurred loss model with an
together supervisory entities – it was left with
expected loss model, one of the most significant
significant deficiencies that the Great Recession
changes in the history of bank accounting.
would soon reveal.
1 Interagency Policy Statement on the Allowance for Loan and Lease Losses, Federal Reserve, December 2006.

Figure 1 History of the loan loss reserves in the US
Federal Reserve Standards
Aim to improve financial

statements but are
ineffective Securities Exchange Act
1917 Establishes the Securities and Exchange Commission (SEC)

Moody’s Ratings Requires companies to publicize information pertinent to investment decisions
Cover essentially all of Additional Laws
the US bond market
Company financial statements must undergo an independent audit before the
company is listed on the exchange
1924
1934
1921 1929 1965
Revenue Act Fraud Exposed Loan Loss Estimation
Recognizes reserves Major corporate accounting Process established

for future bad debts frauds and inconsistencies are
as a legitimate tax exposed
accounting method
Banks allowed to
deduct loan loss
provisions from tax
returns
Source: Moody's Analytics It is important to first understand how the To help illustrate the point, Figures 2, 3, and
existing guidance is applied in practice. There are 4 show excerpts from an annual report of a
approximately 6,000 banks in the US (far fewer $25 billion commercial bank.
if you consider that more than three-fourths are
For commercial banks, loans and leases
part of a bank holding company, or BHC), and
comprise the majority of their assets, and the
they are all required to report their allowance in
ALLL is the most significant estimate on their
the same way and under the same rules. As we’ll
balance sheets. Commercial banks make loans
discuss later, how they derive at the allowance
to businesses and individuals with the money
estimate will differ considerably.

SPOTLIGHT: IFRS 9
Joint Working Group
Banking supervisory agencies

establish a joint working
group to study ALLL concerns
Generally Accepted
Accounting Principles 1999
ALLL Reporting (GAAP)
All banks with total Interagency policy CECL

assets exceeding statement provides New Policy Statement
$25 million are guidance on ALLL, in Expected
accordance with GAAP Banking agencies provide
required to report release of final
joint guidance on ALLL
ALLL standard
1993
2001
1976 2016
See Figure 6 for 2008-2016 details.
1973 1986 1994 2006
FASB Public Statement General Accounting Office Policy Update

(GAO) Study
Establishes standards SEC cites inadequate Banking agencies issue
for financial report documentation of Existing practices are shown to the current policy
preparation procedures for loan obscure changes in credit quality statement, which
portfolio reviews incorporates new
and allowance Highlights a disconnect in the allowance-related
and provision approaches of regulators and developments and
determination auditors ensures consistency with
GAAP
they raise through issuing deposits and other In this example, the bank set aside $175 million
borrowings, with the objective of getting fully in reserves to account for management’s best
repaid on both the principal and interest on estimate of the NCOs that are likely to be
the loan. However, some borrowers inevitably realized from its $15 billion in loans outstanding,
default on their loans, which often results in given the facts and circumstances as of the
the bank having to charge off all or a portion of evaluation date (December 31, 2015). At the
the debt. On average, the net charge-off (NCO, end of the prior reporting period, the bank held
which is gross charge-offs less recoveries) rate is $159 million in reserves in anticipation of future
about 1%.2 charge-offs. The amount of the allowance is
2 Average annualized quarterly NCO rate for all loans and leases 1984–2015 is 0.91%; FDIC.

Figure 2 Balance sheet (assets only)
December 31,
Assets 2015 2014
Cash and due from banks $ 251,258 $ 261,544
Interest-bearing deposits 155,907 132,695
Securities available-for-sale, at fair value 2,984,631 2,793,873
Securities held-to-maturity 3,923,052 3,872,955

(fair value of $3,961,534 and $3,948,706)
Federal Home Loan Bank and Federal Reserve Bank stock 188,347 193,290
Loans held for sale 37,091 67,952
Loans and leases The ALLL is a contra- 15,671,735 13,900,025

asset that reduces the
Allowance for loan and lease losses (174,990) (159,264)
carrying value of total
Loans and leases, net loans 15,496,745 13,740,761
Deferred tax asset, net 101,578 73,873
Premises and equipment, net 129,426 121,933
Goodwill 538,373 529,887
Other intangible assets, net 39,326 2,666
Cash surrender value of life insurance policies 503,093 440,073
Accrued interest receivable and other assets 328,993 301,670
Total assets $ 24,677,820 $ 22,533,172
Source: Webster Financial Corporation
Figure 3 Changes in the ALLL during fiscal year 2015
(In thousands)
Allowance for loan and lease losses: Total

Balance at January 1, 2015 $ 159,264
Provision (benefit) charged to expense 49,300
Losses charged off 43,560
Recoveries 9,986
Balance at December 31, 2015 $ 174,990

SPOTLIGHT: IFRS 9
Figure 4 Income statement
(In thousands, except per share data) Years ended December 31,
Interest Income: 2015 2014
Interest and fees on loans and leases $ 552,441 $ 511,612
Taxable interest and dividends on securities 190,061 189,408
Non-taxable interest on securities 15,948 17,064
Loans held for sale 1,590 857
Total interest income 760,040 718,941
Interest Expense:
Deposits 46,031 44,162
Securities sold under agreements to repurchase and 16,861 19,388

other borrowings
Federal Home Loan Bank advances 22,858 16,909
Long-term debt 9,665 10,041
Total interest expense 95,415 90,500
Net interest income 664,625 628,441
Provision for loan and lease losses 49,300 37,250
Net interest income after provision for loan and 615,325 591,191
lease losses
Non-interest Income:
Deposit service fees 136,578 103,431
Loan and lease related fees 25,594 23,212
Wealth and investment services 32,486 34,946
Mortgage banking activities 7,795 4,070
Increase in cash surrender value of life insurance policies 13,020 13,178
Gain on sale of investment securities, net 609 5,499
Impairment loss on securities recognized in earnings (110) (1,145)
Other income 23,573 18,917
Total non-interest income 239,545 202,108
Non-interest Expense:
Compensation and benefits 297,517 270,151
Occupancy 48,836 47,325
Technology and equipment 80,026 61,993
Intangible assets amortization 6,340 2,685
Marketing 16,053 15,379
Professional and outside services 11,156 8,296
Deposit insurance 24,042 22,670
Other expense 70,584 73,101
Total non-interest expense 554,554 501,600
Income before income tax expense 300,316 291,699
Income tax expense 93,976 91,973
Net income 206,340 199,726

increased or decreased through the combination In simple terms, ASC 450-20 (FAS 5) is the
of NCOs and the provision expense through the reserve that institutions calculate for performing
operating income. loans. Since these borrowers have not defaulted,
In this case, the bank reported NCOs during the amount of potential loss is unknown, so
fiscal year 2015 of $33.6 million ($43.6 million it is usually estimated on a “pool” basis rather
gross charge-offs and $10.0 million in than an individual basis. That is, the assets are
recoveries) and a provision expense of $49.3 grouped into relatively homogenous groups of
million to arrive at the $175 million allowance. risk characteristics. This segmentation approach
Said another way, the NCOs during 2015 is similar to the approach bank management
reduced the bank’s ALLL by $33.6 million, but might take when determining the appropriate
the bank had to expense through the income risk rating methodology or model for a specific
statement another $49.3 million in order to portfolio. To perform this grouping, the portfolio
ensure the amount of the allowance remained of borrowers is stratified by characteristics such
adequate (i.e., $175 million) for future charge- as sector, size, geography, and loan type before
offs based upon the facts and circumstances at determining the best metrics for estimating
the end of 2015 (Figure 3). future credit risk. Under the existing guidance,
A critical component of the existing guidance is the distinction between

accrual versus disclosure. As subtle as it may seem, this is perhaps the most
significant justification for a new impairment model.
As depicted in the income statement in Figure

a widely used method is applying a historical
4, the bank reported net interest income during
NCO rate to each group, adjusted for the effects
2015 of $665 million but the $49 million
of qualitative or environmental factors.
provision to increase the ALLL reduced operating
income and, in turn, capital. ASC 310-10 (FAS 114) is the reserve that
institutions calculate for non-performing or
The concept of the ALLL and its presentation on
“impaired” loans. Although the impairment
an institution’s balance sheet is straightforward,
designation is institution-specific, a general rule
but in many cases it does little to inform
is that a loan is impaired when the institution
investors and other interested parties about
believes repayment of the loan will not be
the true extent of the credit risk inherent in a
realized. According to the current guidance,
bank’s loan portfolio. This is among the most
the allowance is estimated using one of the
commonly cited criticisms of the existing rules.
following three impairment measurement
Let’s take a closer look. methods:
The principal sources of guidance on accounting 1. The present value of expected future cash
for impairment in a loan portfolio under US flows
GAAP are as follows:
2. The loan’s observable market price
»» ASC 450-20, Loss Contingencies (formerly
3. The fair value of the collateral if the loan
known as FAS 5)
is collateral-dependent (repayment solely
»» ASC 310-10, Receivables (formerly known as based on collateral)
FAS 114)

SPOTLIGHT: IFRS 9
Figure 5 Ratio of reserves to total loans and reserves to noncurrent loans (1985–2015)
Reserves/NCL Reserves/TL
4.0 200%
The Great Recession
180%
Reserves as a Percentage of Outstanding Loans
Reserves as a Percentage of Noncurrent Loans

3.5
160%
3.0
140%
2.5
120%
2.0 100%
80%
1.5
60%
1.0
40%
0.5
20%
0 0%
05
06
84
85
86
87
88
89
90
91
94
93
94
95
96
97
97
99
00
01
02
02
03
04
07
08
09
10
11
12
13
14
15
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
1Q
Source: FDIC
An institution may choose the appropriate “When the [down]turn finally did come, and
impairment measurement method on a pool or the tidal wave of losses began hitting shore,
loan-by-loan basis for an individually impaired banks have had to recognize losses through a
loan, except for a collateral-dependent loan.3 sudden series of increased provisions to the loan
Aside from the fact that these rules are loss reserve, which in turn has more than offset
inherently complex, with several impairment earnings and eaten into precious capital. Stated
models, another critical component of the differently, rather than being counter-cyclical,
existing guidance is the distinction between loan loss provisioning has become decidedly
accrual versus disclosure. As subtle as it may pro-cyclical, magnifying the impact of the
seem, this is perhaps the most significant downturn.”4
justification for a new impairment model. Post-Crisis Era and Why the Requirements Are
According to the rule, an allowance should Changing
be recorded in the financial statements if it is While many institutions had been interpreting
“probable” that a loss will incur and the amount the existing guidance more broadly and
can be reasonably estimated. Otherwise it increasing reserves proactively as problems
should be disclosed in the notes, or omitted arose, it is hard to argue that the incurred loss
altogether. In practice, this incurred loss model model was working as intended.
delays recognition of loss by only considering Figure 5 shows a 30-year time series of two
past events and current conditions. important financial ratios used when analyzing
In the words of Comptroller of the Currency John asset quality. The bars in the chart, whose values
Dugan in 2009: are associated with the left vertical axis, show
the trend in the amount of reserves held by the
3 Interagency Policy Statement on the Allowance for Loan and Lease Losses, Federal Reserve, December 2006.
4 Comptroller Dugan Urges Less Pro-Cyclical Approach to Reserves, Office of the Comptroller of the Currency, March 2009.

industry relative to the amount of outstanding crisis, the FASB and International Accounting
loans. The quarterly average is 1.84%, which Standards Board (IASB) began a joint effort to
means that for every $100 in loans, the industry address reporting issues arising from the global
was setting aside just under $2 in reserves. financial crisis. As part of that commitment,
During the financial crisis and in its wake, the the Financial Crisis Advisory Group (FCAG) was
industry began rapidly setting aside reserves in formed to advise the accounting boards of the
anticipation of greater future loan losses. accounting issues emerging from the crisis,
Now look at the line in the chart, whose values along with recommendations for potential
are associated with the right vertical axis. changes to the global regulatory environment.
On average, the average ratio of reserves to The July 2009 report issued by the FCAG
noncurrent loans (defined as loans that are 90 contained several recommendations, including
days or more past due or placed on nonaccrual the need to explore alternatives to the incurred
status by the bank) has been 100.16%, loss model for loan loss provisioning that use
indicating that for every $100 of problem loans more forward-looking information. These
the industry has set aside $100 in reserves. alternatives include an expected loss model and
While it is debatable whether banks should be a fair value model.
reserving for the full amount of defaulted loans
While an objective of the joint advisory group
(loss given default is generally less than 100%),
was convergence in accounting standards,
the trend in the two ratios leading into and
the FASB and IASB decided to go in different
through the Great Recession highlights a major
directions. In December 2012, FASB introduced
problem.
its proposed accounting standards update,6
In the fourth quarter of 2009, the ratio of known as the Current Expected Credit Loss
reserves to total loans rose to more than 3%, model (CECL). In July 2014, the IASB released
nearly tripling in just two years. In order for the its final impairment rules, known as IFRS 9. The
industry to maintain an allowance that was FASB is expected to release its final standard in
believed to be adequate for future loan losses, June 2016. Figure 6 shows the timeline of key
banks had to record $583 billion in provision events.
expenses from 2008 to 2010. As a direct5
While the two boards did achieve convergence
impact on earnings and capital, this put many
on a number of issues raised by the FCAG, there
institutions in capital preservation mode, which
are two significant distinctions worth noting:
made financing less available for businesses and
»» Impairment under IFRS 9 begins with a
individuals at a time when they needed it most,
classification stage to determine how
exacerbating the downturn.
financial assets and liabilities are measured.
Although the industry was increasing the
The classification is driven by the cash flow
allowance at record pace, it could not keep
characteristics and business model in which
up with the pace of rising problem loans. The
an asset is held, but measurement ultimately
coverage ratio fell to a level not seen since
ends up in a single impairment model being
the savings and loan crisis in the late 1980s
applied to all financial instruments. While
and early 1990s. Despite the costly rapid
FASB’s proposal includes a single impairment
provisioning to boost the ALLL, the ratio of
model, it does not include a classification
reserves to noncurrent loans fell below 60%,
stage.
underlining one of the primary limitations of the
»» Under IFRS 9, full lifetime expected losses
incurred loss model.
are to be measured only if credit risk
A New Impairment Model Is Born has increased significantly since initial
In October 2008, in the midst of the financial recognition. Otherwise, the impairment
5 FDIC Quarterly Banking Profile (all insured institutions).

6 Financial Instruments – Credit Losses (Subtopic 825-15), Accounting Standards Update, FASB, December 2015.

SPOTLIGHT: IFRS 9
Figure 6 Timeline of key events leading to a new impairment model
FASB published
the Exposure Draft
“Proposed Accounting
Standards Update,
Financial Instruments
– Credit Losses.”
Introduced CECL.
IASB published
Exposure Draft,
adding further
support for a
FASB and IASB
forward-looking
published a
measure of ECL 2Q16
Joint effort supplementary
between FASB document FASB expected
and IASB introducing to release final
to address “Good Book” standards for
reporting issues and “Bad Book” CECL
arising distinction
from the global
financial crisis
July May July July

2009 2010 2012 2014
October November January December

2008 2009 2011 2012
IASB issued IFRS 9
Financial Crisis FASB published

Advisory Group (FCAG) a proposed ASU
published report on to ECL
Comment periods, deliberations, refinements
delayed recognition of »» Remaining life
(2013 – Present)
losses and complexity »» Cashflow-
with different based
impairment approaches »» Economic
that included forward- conditions
looking information. remain FASB and IASB jointly
unchanged released the “three-
bucket” impairment
model whereby credit
instruments would
have had different
measurement approaches
and migration criteria
across buckets
Source: FASB

measurement period is limited to twelve flow, loss rates, probability of default (PD), or
months from initial recognition. FASB’s a provision matrix using loss factors.
proposed model requires a life-of-loan
»» FASB is expected to issue CECL in June 2016,
forecast of credit losses to be recorded at
which will be effective for SEC registrants’
origination, regardless of credit quality.
2020 financial statements and in 2021
Since the release of the FASB’s accounting for banks that are not SEC registrants.
standards update more than three years ago, Early adoption will be permitted for all
there have been comment periods, deliberations organizations for fiscal years beginning after
and re-deliberations, and strong industry December 15, 2018.
feedback. The FASB has met with countless

Under CECL, an institution will be required to
stakeholders – bankers, regulators, auditors,
impair (reflected as an allowance for expected
solution providers, the SEC, the Public Company
credit losses) its existing financial assets based
Accounting Oversight Board (PCAOB), and
on an estimate of the present value of the
members of the investing community. As we
contractual cash flows not expected to be
The goal of CECL is to improve the process by which institutions measure

credit risk, to the benefit of third parties and the institutions themselves.
draw closer to the soon-to-be-final standard, collected at the reporting date. Not only will
the FASB has released several key decisions to this remove the “probable” threshold in the
date, including the following:7 current approach, but it will also broaden the
»» An entity should apply the CECL model range of information to be considered when
for financial assets measured at amortized estimating the allowance.
cost, such as loans, debt securities, trade The following paragraphs illustrate some of the
receivables, lease receivables, and any other key changes CECL may bring. There will not be
receivables that represent the contractual a “one size fits all” approach when it comes to
right to receive cash. implementation, a common misconception.
»» An entity should consider available The rules to comply and the presentation of an
information relevant to assessing the entity’s financial statements will largely be the
collectability of contractual cash flows, same from one institution to the next, but how
including information about past events, they arrive at an estimate of expected credit
current conditions, and reasonable and losses will depend on factors unique to the size
supportable forecasts. and complexity of the institution’s portfolio.
»» An entity should consider all contractual cash Measuring Expected Credit Loss
flows over the life of the related financial It is clear that the goal of CECL is to improve
assets (life of loan). the process by which institutions measure
credit risk, to the benefit of third parties and the
»» An entity’s estimate of expected credit losses
institutions themselves.
should always reflect the risk of loss, even
when that risk is remote. The measurement of expected credit loss often
starts with a look to the past as a predictor of
»» Methods to estimate expected credit losses
future performance. By grouping financial assets
may include the following: discounted cash
into pools of similar risk characteristics, an
7 FASB (as of March 23, 2016).

SPOTLIGHT: IFRS 9
institution can look to its historical experience its risk measurement process and capabilities.
or the experience of a suitable benchmark for The same recommendation would apply if
those assets. Although no two credit cycles the output of the ratings is not calibrated to a
are the same, reasonable inferences about the specific risk measure such as a PD or expected
future can be made from information from the loss (EL).
past. In fact, that’s the fundamental assumption
Over the last decade, many regional banks and
in the current allowance process and in most
larger community banks have sought to improve
credit risk rating models.
their risk rating practices by making their
CECL will not prescribe a specific methodology internal ratings much less subjective. A common
to be used for measuring expected credit losses, approach has been a bifurcation of credit risk
but a logical approach toward compliance would whereby borrowers are rated on their likelihood
be one that starts with an institution’s current of default (i.e., PD), and credit facilities are rated
risk rating practices, to the extent they are on the severity of loss should default occur
effective at both differentiating the credit risk (i.e., LGD). Through accurate risk measures, an
Figure 7 Quarterly (annualized) charge-off rates: C&I and CRE loans (1985–2015)
CRE C&I
3.5
3.0
2.5
Annualized NCO Rate (%)
2.0
1.5
1.0
0.5
0
1991Q1
1992Q2
1993Q3
1994Q4
1996Q1
1998Q3
1998Q3
1999Q4
2001Q1
2002Q2
2003Q3
2004Q4
2006Q1
2007Q2
2008Q3
2009Q4
2011Q1
2012Q2
2013Q3
2014Q4
Source: Federal Reserve
of borrowers within a portfolio and producing institution can derive an estimate of EL that
a reliable financial measure of credit risk. That could be used not only for managing risk, but
is a limitation with which many institutions also as a foundation for CECL compliance.
struggle. For example, if 80% of the loans in
Incorporating Current Conditions
a relatively diverse commercial loan portfolio
Using historical averages as a basis for deriving
share a similar rating, then it may be necessary
forecasts of credit quality is an approach widely
for management to revisit the effectiveness of
accepted in the banking industry. While such a

through-the-cycle view has its merits, it loses impact on an institution’s credit losses. As
effectiveness when historical experience differs depicted in Figure 7, the average annualized
from prevailing and near-term conditions. quarterly NCO rate for Commercial and
Take the energy sector as a very recent and Industrial (C&I) loans over the last 25 years is
relevant example. With oil prices continuing to 0.77%, but it climbed to approximately 2.50%
hover around $40 per barrel as of March 2016 during the recession of 2001 and during the
(compared to roughly $100 per barrel only recent financial crisis. The impact of the credit
two years ago),8 many energy companies are environment is even more pronounced with
defaulting or nearing default on their loans. At Commercial Real Estate (CRE) loans. While the
the same time, to hedge future credit losses, average NCO rate for the same period is 0.62%,
lenders are curtailing lending and seeking to the median is only 0.14%, indicating that loans
reduce exposure as they ramp up reserves. If secured by CRE are usually a safe and low-risk
oil prices continue to remain at a level not seen investment. That is, until the cycle shifts.
since the height of the financial crisis, the $3 During the nine-quarter period between Q3
trillion sector could soon face a funding crisis 2007 and Q4 2009, the industry’s NCO rate
with rippling effects cascading throughout the for CRE loans rose exponentially, from 0.16%
broader economy. to 3.26%. At the end of 2015, the rate of NCOs
Taking a longer view, we can see how current on CRE loans had returned to near zero – below
conditions in a given cycle can have a profound pre-crisis levels.
Figure 8 Term structure of default risk for a low-risk firm and a high-risk firm
Low-Risk Firm High-Risk Firm
Linear Prediction Non-Linear Prediction Linear Prediction Non-Linear Prediction
4.00% 30.00%
3.93%
3.50% 25.35%
25.00%
2.97%
3.00%
20.28%
2.60% 20.00%
2.50%
Cumulative PD
Cumulative PD
2.07% 15.21% 18.31%

2.00% 15.00%
2.08% 15.43%
1.50% 1.56% 10.14% 12.32%

1.24% 10.00%
1.04% 8.92%
1.00%
5.07%
0.52% 5.00%
0.50% 5.07%
0.52%
0 0
1 2 3 4 5 1 2 3 4 5
Year Year
Source: Moody's Analytics
8 Crude Oil WTI (NYMEX); NASDAQ.

SPOTLIGHT: IFRS 9
Figure 9 Suggested action items for bank management
Manage Establish Program Perform Financial Impact Perform Gap Develop an Initial
Expectations Governance Analysis Analysis Roadmap
It is quite a different story when compared to properly, CECL should enable institutions to add
the energy sector. to reserves when times are good, in anticipation
By incorporating information about current of a shift in the cycle, and to begin to release
conditions, perhaps as a factor within a model reserves when it appears the worst is behind
or as a qualitative adjustment to a cycle-neutral them, to help facilitate growth.
rating, an institution will be better positioned to A forward-looking view requires an ability

understand the impact of the prevailing credit to predict the future. The proposed update
cycle on its loan portfolio in order to improve its will require an entity to consider available
estimate of expected credit losses. information relevant to assessing the
collectability of contractual cash flows,
Predicting the Future
including information about past events, current
In parallel with the actions of the accounting
conditions, and reasonable and supportable
boards to rectify problems that arose during
forecasts. Once the forecast and its impact
the financial crisis, the Federal Reserve and
on the portfolio can no longer be reliably
other banking agencies were making waves of
estimated, CECL will allow an entity to revert
their own. The accounting boards and banking
to historical credit loss experience for future
agencies sought to incorporate a more forward-
periods.
looking view of credit risk, but they focused on
different measures. Whereas bank supervisors It is worth reiterating that the practical
focused on low-probability, high-impact events interpretation will be different across
that could strain a firm's capital adequacy, institutions. What will be expected of a
the FASB and IASB dedicated their efforts on community bank is not the same as what will
accounting for “reasonable and supportable” be expected of an institution that is subject
forecasts under more probable scenarios. One to the Dodd-Frank Act Stress Tests (DFAST).
way to look at it would be that the allowance Institutions that are already translating
serves as a cushion for “expected” credit losses, macroeconomic scenarios into a granular
and capital serves to absorb tail events, or forecast of credit losses are well-positioned
“unexpected” credit losses. to incorporate reasonable and supportable
forecasts into the allowance. Community banks
Without a forward-looking component in
will likely apply a broader and more judgmental
the estimation of loss forecasting, reserves
approach to deriving forecasts.
are inherently pro-cyclical. Banks add to the
allowance during periods of stress, usually when Extending Measurement Across the Life of the
access to financing is needed most, and they Loan
release reserves during periods of expansion, Most loans issued by banks do not mature
when many businesses and individuals can within a year of origination. However, many
meet their financing needs with operating cash institutions set aside an allowance for a year’s
flow or discretionary income. If implemented worth of charge-offs. Under CECL, a life-of-the-

Figure 10 Core capabilities required for new impairment calculations
Economic
Scenarios
Analysis Credit
& Reporting Data
Credit
Impairment
Analysis
Workflow Credit Modeling

& Overlay &
Management EL Calculation
Sensitivity
Analysis
loan forecast of credit losses will be recorded exponential (0.52% x 5)), the cumulative
at origination, thereby mandating reserves be default risk is 2.60%, which is considerably
set aside when a loan is made and maintained less than 3.93%. The opposite is the case with
throughout its contractual life. What’s more, the the high-risk firm, which has a 5.07% one-year
new standard suggests that it is inappropriate PD. On a linear basis, the five-year cumulative
to simply “gross-up” annual measures. Figure default risk is more than 25%; however, on
8 illustrates the difference between a term an empirical basis, the probability of default
structure of default risk calculated on a over five years is only about 18%. These two
purely linear basis and one calculated on an examples highlight the reality that the term
empirically-derived non-linear basis. structure for low-risk firms is increasing
The low-risk firm has a one-year PD of 0.52%. (mathematically, the intercept coefficient is
Assuming a five-year maturity, the cumulative positive), whereas for high-risk firms, the term
default risk is nearly 4% when modeled structure of PD is decreasing. This behavior
empirically.9 That equates to nearly an eightfold follows a somewhat mean-reverting pattern.
increase over a five-year period. When the To summarize, with many institutions
term structure of default risk is calculated by establishing an allowance equal to an estimate
simply multiplying the one-year measure by of NCOs over a one-year horizon, the life-of-
the number of years (i.e., linear rather than loan requirement introduces complexity beyond
9 Moody’s Analytics RiskCalc v4.0 Corporate Model.

SPOTLIGHT: IFRS 9
the capabilities most institutions currently »» Perform Financial Impact Analysis:

possess. In our meetings with FASB officials, Management will not know the exact
we were able to confirm that CECL has no impact of the new standards on the
mandate for modeling and leaves questions of organization’s financial statements until the
methodology to individual banks. Nonetheless, new framework has been implemented at
many of the objections to date are focused on an enterprise level. However, pilot tests on
implementation. Bankers fear that regulators segments of the portfolio using simplifying
and auditors will expect them to use the same assumptions (i.e., flat LGD term structure)
tools and methodologies used by larger banks, can help management identify a range
which is something they cannot afford to do. of possible outcomes. These results can
be socialized with peer institutions and
Early Preparation
compared to public impact studies.
For jurisdictions under IFRS 9, the
implementation deadline is set firmly for »» Perform Gap Analysis: Management must
January 2018. FASB’s CECL standard is slated identify what in its “toolbox” will help the
for release by June 2016, with implementation organization meet the requirements. What
required by January 2020 for SEC filers approaches meet basic requirements? What
and January 2021 for all others. Given the data, models, and technology can the bank
significance of the changes, financial institutions repurpose, and who owns these within the
would benefit from a proactive approach organization?
to develop the organizational capabilities »» Develop an Initial Roadmap: The gap
necessary to satisfy the new impairment analysis will identify relevant existing tools,
requirements. Figure 9 highlights the key actions as well as areas where the organization must
management should consider to get started. develop new capabilities. These findings
»» Manage Expectations: Organizations will will inform the organization’s roadmap
face questions from a variety of stakeholders, for implementation. The roadmap should
including employees, auditors, regulators, identify the key objectives, major milestones,
and investors. It is critical for management and broad timelines spanning preparation
to understand the new guidance and be able through implementation. The roadmap sets
to clearly communicate to stakeholders how the development priorities for the program
the organization may be affected financially and serves as the foundation for a more
and non-financially. Communicating early detailed project plan. Major milestones
and often regarding the potential impact, the typically include methodology design,
firm’s implementation plan, and progress in software implementation, and impact
the firm’s preparations will be essential to analysis (i.e., parallel run). As part of the
managing expectations. roadmap, the organization should determine
whether to seek an accelerated path to
»» Establish Program Governance:
achieve early adoption. Even the most
Implementation of the new impairment
advanced institutions may need at least 18
accounting rules will require resources and
months to go live with a new impairment
coordination from across the organization,
framework.
including lending, risk, finance, and
IT. A steering committee should be
Core Capabilities Required
developed with ultimate responsibility for
Figure 10 summarizes the core set of capabilities
implementation of the new framework. The
institutions will need to estimate credit
committee can form task groups to focus on
impairment under the new standard. The
specific workstreams such as modeling, data
analytical rigor demanded in each category may
infrastructure, and reporting.

vary widely across institutions and portfolio Credit modeling represents the analytical tools
segments. required to estimate probability of default, loss
Expected credit losses must represent an severity, exposure at default, and/or expected
unbiased estimate using reasonable and losses for the various segments of the portfolio.
supportable information about past events and Some institutions have developed sophisticated
current conditions, as well as forecasts of future model development and validation functions to
economic conditions. To account for forecasts support internal model development. Others
of future economic conditions, institutions will have outsourced some of these capabilities,
need to source economic scenarios internally leveraging the data or expertise of third
or by third parties. They must determine which parties for specific asset classes. Regardless
variables to forecast, the number of possible of the source of these tools, the models must
outcomes to consider, the likelihood of the be powerful, forward-looking estimates of
possible outcomes, and the source(s) of the credit risk throughout the life of the exposure.
economic forecast. Large financial institutions Some institutions have developed credit risk
have developed economic forecasting models for Basel and internal risk management
capabilities for stress testing purposes,10 but purposes. In most cases these will need to be
economic forecasting is likely to represent a modified to extend the forecasting horizon
capability gap for most financial institutions. (most current models estimate credit risk over
a one-year horizon) and to reflect current and
Credit data encompasses the current
forward-looking information.
information required to estimate credit losses
for each of the exposures in the portfolio Institutions must also develop the capability
(balances, commitment, PD/LGD profile, to perform sensitivity analysis. In the context
cash flow profile, etc.). In addition, it includes of impairment calculation, this refers to the
the credit research data required to develop ability to test the sensitivity of the impairment
loss estimation models that are trained using estimates to model assumptions. Sensitivity
historical data. Some institutions will need to analysis could take place in various forms,
develop the capability to integrate all the loan including changing the scenarios or the
accounting and risk profile data into a single probability assigned to each scenario, or using
system for impairment calculations. In addition, alternative credit risk models to estimate credit
firms will need to aggregate historical credit losses. This could be a very manual process, or it
risk data from internal and external sources to could be carried out in a controlled environment
facilitate credit risk model development. with auditability, reporting, and archiving
10 For example, CCAR banks in the US are required to generate firm-specific stressed loss forecasts in addition to the regulatory
scenarios.

SPOTLIGHT: IFRS 9
features. Ultimately, the idea is to better inform disseminated throughout the organization.
management of the uncertainty around the Accordingly, institutions may need to enhance
impairment estimates. reporting capabilities to address new and
Workflow and overlay management and evolving reporting requirements.
analysis and reporting focus on the operating Large institutions have developed robust data
environment used for impairment calculations. repositories and reporting infrastructures to
Because impairment values are used directly address Basel and stress testing requirements,
in an organization’s financial statements, they but enhanced reporting will pose a particular
require strong governance and controls. Unlike challenge to small and mid-size institutions.
risk parameter estimates used for regulatory Technical footprint, performance, flexibility, and
reporting, risk estimates used for impairment compatibility with existing systems should be
calculation will fall directly under the purview of carefully considered when investing in a new
auditors. The calculation environment will ideally infrastructure solution.
support workflow and overlay management to
Conclusion
define user roles and track overrides to model
While we await the release of the final standard,
estimates. The system will need to integrate
we interpret CECL to be consistent across
the scenarios, data, models, and provision
institutions. However, implementation of the
calculations in a way that facilitates user
rules will be unique to the size, complexity, and
interactivity and auditability.
geographical footprint of the institution. One
In many jurisdictions, reporting requirements size certainly will not fit all. The capabilities
for regulated financial institutions are being required to be compliant will differ throughout
adapted to reflect changes in the impairment the industry, but the mandate to provide
framework. Institutions would be required to stakeholders with actionable information about
explain the drivers of the changes in provisions an institution’s credit risk will not.
between reporting periods. For example, banks
For all its shortcomings, CECL should bring about
may need to separate changes due to new
a more comprehensive view and a disciplined
originations, asset disposal, change in the risk
approach for quantifying the expected credit
of existing loans, and changes due to updates
losses inherent in an institution’s financial
in the estimation methodology. Furthermore,
instruments.
management may have its own preferences
regarding the analysis and reports that will be

IFRS 9 SURVEY RESULTS
By María Cañamero, Michael McDonald, and Yagmur Uenal
This article is a summary of the views expressed by regional banking

institutions in a recent survey about IFRS 9 regulation. The survey
was conducted to assess progress, potential challenges, and plans
of banks with regards to IFRS 9 compliance. The article describes
the survey methodology and reviews current preparedness levels
of banks. It shows expected impact on loan provisions, origination
policies, and anticipated increases in personnel. It also reports
on banks’ main challenges in the areas of data, models, and
infrastructure, and summarizes institutions’ overall opinions of
IFRS 9.
Introduction impairment and the expected impact on

Moody’s Analytics conducted a survey in Q3 provisioning. Data and modeling demands
and Q4 2015 to gauge the state of IFRS 9 were perceived as the most challenging
implementation and related challenges. The aspects of the impairment calculations.
survey was conducted via one-on-one interviews Almost all participants expected their banks’
with 25 regional banking institutions in North provisions to increase.
America and Europe to understand their views
»» Nearly 75% of all interviewed parties
regarding the IFRS 9 guidelines. The group was
had defined IFRS 9 budgets. Two in five
defined as banks with under $30 billion in assets.
respondents planned to invest more than $1
The purpose of this survey was to assess million in Austria, Canada, the Netherlands,
progress, central areas of concern, and future and the UK, with credit modeling as the
investment plans of regional banks with regards primary focus.
to IFRS 9 compliance.
Methodology
The survey revealed these key points:
Moody’s Analytics conducted 25 one-on-one
»» 80% of interviewed institutions were still at market research discussions with regional banks
early stages of IFRS 9 preparedness and would between September and December 2015. All
be considering external help for their IFRS 9 participating banks had under $30 billion in total
projects. assets, with retail being the main asset class of
»» In Canada, the Netherlands, and the UK, most banks’ portfolios. Figure 1 summarizes the
most banks anticipated investment decisions complete breakdown of participants by country,
in 2016, and the largest institutions were total assets, main asset class, and job function.
furthest along. Spanish and Italian banks Low Preparedness Levels
generally had not begun their IFRS 9 The survey found that banks were largely
preparations. unprepared for IFRS 9 at the time of the
»» The main challenges cited were IFRS 9 interviews, and most were still in early

SPOTLIGHT: IFRS 9
Figure 1 Demographics of banks surveyed
7
4
8
2
1
Market Research Sample by Total Assets Main Asset Class of Portfolio Participants by Job Function
C&I Finance 15
13% SME CRE
13% 7% Risk 8
5 9 11
Other Accounting 8
6%
0 5 10 15 20 25 Res. Financial Reporting 3
Retail
<$5b 58% Real
$5-15b $15-30b Estate Other 2
3% Treasury 1
Source: Moody’s Analytics
preparation steps. Nearly a third of respondents of their efforts. More than a quarter of those not
had not initiated any work, and almost half currently following an internal ratings-based
were still focusing on their gap analysis. That (IRB) approach were planning on leveraging
leaves a scant 20 percent that had started design the IFRS 9 enhancements to move to an IRB
or building work. Figure 2 shows the status of approach for regulatory capital.
banks’ IFRS 9 implementation progress at the
Anticipated Challenges
time of this survey.
New IFRS 9 guidelines are structured in three
Both country and bank size were found to main phases:
factor into preparedness levels. In Canada, the
1. Classification and Measurement
Netherlands, and the UK, most banks anticipated
investment decisions in 2016, and the largest 2. Impairment
institutions were furthest along. Spanish and 3. Hedge Accounting
Italian banks generally had not begun their IFRS
9 preparations. Banks anticipated that Phase 2, Impairment,
would pose the most challenges. Nearly all
For most banks surveyed, IFRS 9 compliance
respondents expected challenges in this phase,
requires more resources than they have readily
while relatively few foresaw issues in Phases 1
available. Nearly 80 percent of respondents
and 3, as shown in Figure 3.
stated that they were or would be considering
external help for their IFRS 9 compliance Banks expected IFRS 9 to significantly impact
projects. loan origination policies and bank provisioning,
with 80 percent expecting increases in
Banks were, however, planning to make the most
provisions. More than a quarter expected

Figure 2 Progress of IFRS 9 implementation Figure 3 IFRS 9 phases where challenges are expected
More than 2/3 of banks 8%

expect to begin parallel
2/3 of respondents
runs before 2018 16%
have involved external
contractors 16%
56% 96% 84%
Compliant
Gap analysis
initiated
Build
Design
32 4%
28%
16%
48%
4% 8%
32% Phase 1: Phase 2: Phase 3:
Classification & Impairment Hedge Accounting
Measurement
- Progress of IFRS 9 implementation + No Response No Yes
Source: Moody’s Analytics Source: Moody’s Analytics
changes to loan origination policies, with The banks interviewed also noted infrastructure
anticipated impacts on data capturing, pricing, challenges of low or medium criticality,
and credit decisioning (Figure 4). including:
This survey found that when addressing »» Issues related to handling larger volume of
impairment, most banks were challenged calculations
with data and modeling demands, with some
»» Data quality due to legacy system issues
infrastructure challenges, as well.
»» Need for improved systems to gain
In terms of data challenges, the most critical
automation and auditability in IFRS 9
challenge for many banks was the lack of
calculations
historical data for some portfolios. Banks also
faced a lack of PD data at origination, and some IFRS 9 compliance requires sizable investments
data characteristics which were needed for IFRS by banks, with a primary focus on credit
9 were not previously gathered or stored. modeling to overcome stated challenges. Nearly
75 percent of respondents stated they had
Modeling challenges, the most critical for banks,
defined an IFRS 9 budget to be used for staff
included the following:
cost, fees of external advisors, and tool upgrades.
»» Lack of PD / LGD models for some portfolios
Banks in Austria, Canada, the Netherlands,
»» Lack of robustness in existing PD / LGD and the UK were the most likely to spend a
models significant amount, with almost 40 percent
»» Issues converting Through-the-Cycle (TTC) to reporting that they planned to invest over $1
Point-in-Time (PIT) and estimating lifetime million.
expected credit losses Personnel increases were a major area of planned
»» Need for model enhancements to address investment among survey respondents, with
forward-looking requirement 60 percent reporting they planned to increase

SPOTLIGHT: IFRS 9
Figure 4 Impact on loan origination policies Figure 5 IFRS 9 impact on resources and business areas
3 in 5 regional banks are planning to increase

head count to manage the IFRS 9 challenges
AREAS OF IMPACT
»» More granular data needs to Department with increase of internal resources planned
(# of respondents)
be captured at origination
YES »» Pricing as a result of
28% increased provisions 13
»» Credit decisioning
N/A NO 6
52% 20% 2
1
Credit & Risk Finance Data IT

Management &
Analytics
Source: Moody’s Analytics Source: Moody’s Analytics
staffing to handle IFRS 9 challenges (Figure 5). representative. “There are inconsistencies in
Other main areas of planned investment how financial instruments are accounted for and
included equipment upgrades and third party the balance sheet still will not reflect the real
vendor support. situation.”
But despite its shortcomings, respondents

Banks’ Opinions and Reactions
expressed a general sentiment that IFRS 9 is a
While most participants agreed that IFRS 9
step in the right direction.
would bring a degree of uniformity to the
international banking industry, they also voiced According to a UK respondent:
significant concerns about the guidelines’ “Even though IFRS 9 will involve a lot of effort
limitations and weaknesses. and high costs, it will help preserve quality across
“IFRS 9 brings ‘positive concepts’ to address the entire industry by creating a more solid
some of the weaknesses of IAS 39,” according governance structure across all banks.”
to one respondent from a UK bank. “However, As indicated in this survey, IFRS 9 poses
it raises issues from a practical point of view, substantial challenges to the banks interviewed.
particularly in terms of the modeling, since the To prepare for this new accounting standard
new standard will not enhance comparability banks will have to adapt their operations and the
among institutions.” way they classify and manage the quality of their
Some believed the guidelines would not effect loans and provisions at origination. Moreover,
change or improve transparency on a large scale. banks will need to transform their processes and
practices in terms of expected loss estimation,
“Conceptually, the framework makes sense, but
data management, capital calculation, and
in practice it’s unlikely that many banks have
reporting.
been estimating provisions based on incurred
losses only,” according to a Canadian bank

Moody’s Analytics
Risk Practitioner
Conference 2016
Our annual Risk Practitioner Conference brings
together industry experts and leading risk
practitioners from across the globe.
For more information visit

MoodysAnalytics.com/RPC2016
Interview with Mark Almeida
PRESIDENT, MOODY’S ANALYTICS
The theme of this year’s Risk Practitioner Conference is The Convergence of Risk, Finance,
and Accounting. Why?
The Risk Practitioner Conference has evolved since its inception 11 years ago, and this evolution reflects
the considerable change that has impacted banking and financial markets in this century. What started
as a rather unique, regional seminar focused on innovations in credit risk has grown into an event
that attracts risk, finance, and technology professionals from around the world. In response to the
financial crisis, new regulatory forces and governance practices are driving dramatic change in financial
institutions’ management of risk. Stress testing programs implemented by regional banking regulators,
new accounting standards, and more rigorous capital adequacy and liquidity risk requirements are
bringing risk, treasury, and finance functions closer together. Because this is top of mind for financial
institutions, we have organized our program accordingly.
Why is this theme important to Moody’s Analytics?

We have long believed that financial institutions would inevitably leverage modern technology
capabilities to undertake better, more precise, and more efficient risk management. This vision is at the
core of what we do at Moody’s Analytics. We believe that advanced quantitative analytics can have
greater impact if they are integrated and made accessible across an enterprise – from a risk modeling
professional to a front-line banker and to the CFO. Although responses to the spate of recent regulatory
imperatives have been difficult and costly to implement, we believe that the investment made by banks
and insurers to comply with such regulation will ultimately provide the means for better decision-
making and enhanced operational oversight of financial institutions. Increasingly, we see that executives
of financial institutions are recognizing this. While we’re not there yet, the industry is clearly evaluating
© 2016 Moody's Analytics, Inc. and/or its licensors and afﬁliates. All rights reserved.
this convergence of risk and finance functions, and Moody’s Analytics is eager to contribute to this
process by facilitating dialogue among industry participants on these important topics.
What should participants expect at this year’s conference?

This year’s conference builds on what is now a well-established tradition, with sessions designed jointly
by industry practitioners and Moody’s Analytics subject matter experts. This results in a selection of
topics that are of high priority for the industry. One enhancement that we’re introducing this year is
the organization of seminar streams by business function. We are planning four concurrent streams
oriented toward technologists, finance and treasury professionals, credit risk managers, and specialists
in quantitative risk. This structure will allow us to zero in on function-specific issues that are part of the
broader industry themes. For example, we will discuss upcoming changes in impairment standards from
multiple perspectives: methodology design and organizational design within the credit risk management
stream, and technical architecture design within the technology stream. In a separate session, we will
also solicit feedback from auditors and banking supervisors, who of course are the ultimate reviewers of
any accounting standard implementation. Through this framework, we look forward to hosting a series
of productive sessions that build on the ongoing conversations we have with market participants, all
within a forum that brings these important issues to a larger and broader audience.
IFRS 9 IMPAIRMENT REGULATIONS:
IMPLEMENTATION CHALLENGES AND
POTENTIAL SOLUTIONS
By Christopher Crossen and Dr. Yashan Wang
IFRS 9 aims to streamline and strengthen risk measurement and

Christopher Crossen
Associate Director – Research reporting of financial instruments in an efficient, forward-looking
manner. This new accounting standard will have far-reaching impacts
on accounting practices and performance results. This article
Christopher Crossen is the Research Writer for Moody’s focuses specifically on the IFRS 9 impairment model and challenges
Analytics Quantitative Research Group. He has worked in interpreting the IFRS 9 requirements. We suggest solutions for
for Moody’s Analytics since 2008. Prior to Moody’s, he
worked as an analyst in equity research and portfolio
meeting requirements in areas such as portfolio segmentation,
management. He earned advanced degrees from UC thresholds for transitions among impairment stages, and calculating
Berkeley’s Haas School of Business and the University expected credit losses, leveraging Moody’s Analytics expertise in
of Notre Dame.
credit risk modeling.
Introduction: The Revised Impairment Model is increasing the timeliness of loss recognition
and Implementation Challenges and addressing the over-complexity of the
Dr. Yashan Wang IASB published the final version of IFRS 9 in July multiple impairment approaches required under
Senior Director – Research
2014, which marked the completion of replacing the IAS 39 “incurred loss” model.
IAS 39. The revised impairment model aims to
IFRS 9 requires recognition of loss allowance
provide users with more transparent and useful
for expected credit losses at all times. It further
Dr. Yashan Wang is a Senior Director at Moody’s information regarding expected credit losses.
requires that this amount be updated at each
Analytics where he leads the research and quantitative
One of the key differences between the two reporting date to reflect changes in the credit
modeling team for portfolio valuation and balance
sheet analytics. He has led several research initiatives standards, with large implications, is the risk of financial instruments in scope. IFRS
in asset valuation, credit migration, joint credit and clarification and methodology for recognizing 9 provides three approaches for recognizing
interest rate dynamics, and balance sheet analytics. impairment. Under the old IAS 39 “incurred impairment loss:
Yashan has also worked with global clients and
loss” model, impairment depended upon how
provided training and advice on enterprise risk »» A general “three-bucket” approach for regular
a financial instrument was classified. Under the
management, asset and liability management, PPNR, financial instruments
and stress testing. Prior to joining Moody’s Analytics, new IFRS 9 model, impairment measurement
Yashan was an Assistant Professor at the MIT Sloan is the same regardless of instrument type and »» A simplified approach for lease receivables,
School of Management. He has a PhD in Management classification. The new impairment model uses trade receivables, and contract assets without
Science from Columbia University. a significant financing component
a single, forward-looking expected credit loss
model that applies to all types of financial »» A special, “credit-adjusted Effective
instruments within the scope of impairment Interest Rate (EIR)” method for purchased
accounting. The new model requires recognizing or originated credit-impaired financial
expected losses since origination or acquisition instruments
date. The major advantage of the new approach

SPOTLIGHT: IFRS 9
The new impairment standard applies to all »» Enhancements required for PD/LGD/EAD and
firms reporting under IFRS 9. In particular, loss rate models, in order to achieve IFRS
requirements affect firms holding financial 9-compliant expected credit loss calculation.
instruments such as loans, investments in debt, 1
We discuss these specific challenges in further

and trade and lease receivables. The revised IFRS
detail next.
9 model will impact banks and insurance firms
most, due to their large financial instrument Portfolio Segmentation
holdings. Non-financial firms with portfolios Firms typically segment portfolios along business
including trade and lease receivables, debt lines, product types, and risk characteristics for
IFRS 9 requires a more granular and dynamic approach for portfolio

segmentation.
securities, and intragroup loans must also revise impairment calculation. IFRS 9 requires a more
current impairment loss calculations. granular and dynamic approach for portfolio
Firms must capture and collect historical segmentation. Firms must group financial assets
data and other trend information required for based on shared credit characteristics that
building a forward-looking impairment model typically react in a similar way to the current
and for tracking credit risk migration since the environment and macroeconomic factors. These
origination and recognition of the financial characteristics include instrument type, credit
instrument. Data will include the historical risk ratings, industry, geographical location,
probability of defaults, ratings, loss amount, date of initial recognition, remaining term to
product features, and economic scenario maturity, and underlying collateral. Groupings
variables. Firms may also need to develop are reevaluated and re-segmented whenever
new models and processes or upgrade existing new, relevant information arises, such as a
models in order to identify an increase in change in economic conditions, or when credit
credit risk and calculate one-year or lifetime risk expectations change.
expected losses. Gathering this granular data Determining Significant Changes in Credit
has been ranked the number one challenge by Quality
banks responding to a recent Moody’s Analytics
A true economic loss occurs when current
survey.2
expected losses exceed initial expectations.
Key Challenges to Implementing IFRS 9 Recognizing lifetime expected credit losses after
Impairment Requirements
a significant risk increase reflects economic loss
The primary methodological and analytical
more accurately in the financial statements. To
challenges that firms may encounter while
determine significant credit deterioration, a firm
implementing an IFRS 9 impairment model will
should consider reasonable and supportable
arise in the following areas:
information available without undue cost or
»» Portfolio segmentation techniques for credit effort, and then compare the following:
risk modeling and expected credit losses
»» Risk of default at the reporting date
calculation
»» Risk of default at the date of initial
»» Application of different thresholds for
recognition
assessing significant increases in the credit
risk of financial instruments A significant increase in credit risk assessment
1 Investments in equity instruments are outside the scope of the IFRS 9 impairment requirements, because they are accounted
for either at Fair Value through Profit or Loss (FVTPL) or at Fair Value through Other Comprehensive Income (FVOCI), with no
reclassification of any fair value gains or losses to profit or loss (i.e. the FVOCI election for equity instruments).
2 Gea-Carrasco (2015).

may be done on a collective basis (for example, sophisticated “expected loss approach” for most
on a group or subgroup of financial instruments), portfolios.
if evidence is not yet available at the individual
Many banks may leverage their existing internal
level. While IFRS 9 does not prescribe any
credit risk management systems and expected
specific approach for assessing changes in
loss calculation processes used for Basel
credit risk, it allows the following operational
regulatory requirements, but they will need to
simplifications for assigning the instrument into
modify them to comply with IFRS 9 impairment
different stages:
requirements. Modifications include adjustments
»» A rebuttable presumption of a significant for through-the-cycle vs. point-in-time estimates
increase in credit risk when the borrower is 30 and extending the Basel one-year PD/LGD/EAD
days past-due. This indicator is not absolute, to full term structures to capture the expected
but it is presumed to be the latest point. lifetime of financial instruments.
»» For instruments with low credit risk, firms can Other institutions may use in-house models
continue to recognize a 12-month allowance. and processes for stress testing and adjust
the forecast for the forward-looking scenario
The low credit risk exemption is often viewed as
rather than the stressed scenarios. Estimating
a suitable approach for wholesale and corporate
“forward-looking,” future economic conditions is
exposures because firms can often map internal
only the first step of the adjustment process, for
grades to external rating agencies. Likewise, the
which institutions may need to develop single
30 days past-due criterion is often applied to
or multiple economic scenarios to calculate
retail portfolios because firms usually cannot
expected credit losses. The most challenging
map the portfolio to external ratings.
aspect of the change may be incorporating the
While IFRS 9 does not explicitly require it, Moody’s Analytics recommends
that banks and insurers consider a more robust and sophisticated “expected
loss approach” for most portfolios.
However, the Basel committee maintains higher macroeconomic factors forecast (interest rates,
expectations for banks implementing IFRS 9. The unemployment, GDP growth, etc.) into the PD/
committee considers both the low credit risk LGD/EAD modeling and, thus, into the expected
exemption and the 30 days past-due criterion credit loss calculation. Adjusted models must
to be a “very low-quality implementation” of an reflect how such changes in factors affected
expected credit loss model. The committee has defaults and losses in the past. However, it is
strong expectations that a bank will not fall back possible that the combination of forecast factors
on the 30 days past-due assumption, unless all may never have been seen historically.
forward-looking information has no substantive
Even if all the IFRS 9-compliant models for
relationship with credit risk. The appropriate
loss rate and the different components in the
approach will vary by the institution’s level of
expected loss approach are readily available,
sophistication, the financial instrument, and data
additional issues will arise when determining the
availability.
expected credit loss. Rules require discounting
Expected Credit Loss Calculation the expected cash shortfalls in order to obtain
The IASB acknowledges firms may measure the current value at the reporting date. Current
expected credit losses (ECL) using various regulatory calculations do not discount at all
techniques. While IFRS 9 does not explicitly or discount only from the date of the expected
require it, Moody’s Analytics recommends that default point. Firms will need to modify existing
banks and insurers consider a more robust and systems to better capture the expected timing of

SPOTLIGHT: IFRS 9
credit losses and to discount future amounts to risk changes after initial recognition affect only
the reporting date. IFRS 9 requires the use of the some exposures within a group, those exposures
effective interest rate at initial recognition when should be segmented out into appropriate
discounting the cash flows. Firms must also subgroups.
backfill the effective interest rate for financial
Determining Significant Changes in Credit
instruments if this information is missing in the
Quality
current accounting system. In addition, firms
IFRS 9 requires assessing financial instruments
may need to enhance or replace a current loan
for significant credit risk increases since initial
loss calculation engine to accommodate the
recognition. Firms must use change in lifetime
demanding computational loads of exposure
default risk (considering quantitative and/
level, cash flow-based, lifetime expected credit
or qualitative information), a low credit risk
loss calculations.
exemption, and a rebuttable presumption of 30
Potential Solutions for Implementing the IFRS 9 days past-due. For instruments whose default
Impairment Model
occurrences are not concentrated at a specific
Given these challenges, we next discuss
point in time during the expected life, firms
potential solutions for each of the previous areas
can use changes in one-year in default risk to
of discussion.
approximate changes in lifetime default risk.
Portfolio Segmentation
When using a loss rate approach to measure
Implementing the IFRS 9 impairment model credit risk increases, firms should use changes
results in a granular and dynamic portfolio in credit risk isolated from other expected loss
segmentation scheme. Financial instruments drivers, such as collateral. Also, the loss rates
should be segmented based on shared credit should be applied to groups defined in a similar
One challenge in calculating credit risk changes is the backfilling of credit

risk assessment at origination. For this purpose, institutions must consider
credit risk characteristics at initial recognition.
risk characteristics. Instruments grouped way to the groups for which the historical credit
together should respond to historical and loss rates are calculated. Since loss rates should
current environments, as well as to forward- incorporate information regarding current and
looking information and macroeconomic factors forward-looking economic conditions, firms
in a similar way, with respect to changes in should apply historical loss rates consistent with
credit risk level. The grouping method should the current and expected economic conditions.
be granular enough to assess changes in credit If the historic economic conditions differ, an
quality leading to migration to a different credit adjustment is needed. A possible approach for
risk rating, thus impacting the estimation of calculating loss rates dependent upon economic
expected credit losses. Segmentation should conditions is to develop a model linking loss
be reevaluated and exposures re-segmented rates with economic variables.
whenever there is relevant new information
PDs can also be used to identify significant credit
or whenever credit risk expectations change.
risk increases. If using PD changes, Moody’s
Most importantly, exposures should not be
Analytics recommends assessing the logarithmic
grouped in such a way that the performance of
change instead of raw changes,3 as the
the segment as a whole masks an increase in
significance of a specific change in PD depends
a particular exposure’s credit risk. When credit
on the starting point.
3 Logarithmic changes are similar to percentage changes for small fluctuations. However, logarithmic changes have more
desirable properties, as they are symmetric and additive.

IFRS 9 states that firms cannot simply compare Expected Credit Loss Calculation
the change in absolute risk over time. Instead, To overcome the expected loss calculation
they should incorporate the relationship challenges, firms can implement different
between expected life and default risk. One solutions to comply with IFRS 9 including
possible approach to doing this is to use existing internal models or new tools. The
annualized PD values instead of cumulative PD targeted IFRS 9 solution should possess the
values. For instruments whose default patterns following characteristics:
are not concentrated at a specific point in
»» Applies a default definition consistent with
time, one can use changes in 12-month PD as
internal credit risk practices
an approximation of the lifetime default risk
»» Reflects an unbiased and probability-
change. This approach may not be suitable
weighted amount of expected credit losses
for instruments with only significant payment
obligations after the next 12 months, or for »» Is able to calculate expected losses for both
which changes in macroeconomic or other one year and expected life
credit-related factors are not adequately »» Incorporates information regarding past
reflected in the default risk during the next 12 events, current conditions, and forecasts of
months. future economic conditions
In addition to using PD changes, changes in PD- »» Discounts expected credit losses to the
implied rating, expressed as notch differences, reporting date, using the effective interest
can also determine significant increases in credit rate as the discounting rate
risk. Ratings are sometimes preferred over PD
»» Reflects cash flows expected from collateral
measures, as many institutions are more familiar
and other credit enhancements as part of the
with internal or agency ratings. However,
contractual terms
implied ratings have the disadvantage of being
non-continuous (like PD measures). Additionally, »» Considers all contractual terms of the
if using an internal rating system, it must be financial instrument
well-designed, incorporating a reasonable »» Estimates the portion of the commitment
number of rating categories and avoiding too to be drawn down for financial instruments
many credits classified into specific categories. that include both a loan and an undrawn
For IFRS 9 purposes, an internal rating system commitment component
should also incorporate the relationship between
expected life and default risk. The internal rating Specific Models for IFRS 9 Impairment Model
mappings, therefore, should depend on the Implementation
instrument’s maturity. Firms can leverage Basel and stress testing
models for IFRS 9 purposes. They can also
One challenge in calculating credit risk changes
utilize vendor models to help comply with
is the backfilling of credit risk assessment at
IFRS 9 requirements. We recommend specific
origination. For this purpose, institutions must
adjustments in order to comply with IFRS 9.
consider credit risk characteristics at initial
recognition. This requires historical information PD Models
such as internal ratings, external ratings, Most banks are subject to the Basel Capital
financial statements, and economic conditions Standards, which state three possible
statistics. approaches for calculating capital requirements
We next discuss incorporating forward-looking for credit risk: the Standardized Approach,
information into credit risk measures, as well as the Foundation Internal Ratings-Based (FIRB)
other challenges in calculating expected credit Approach, and the Advanced Internal Ratings-
losses. Based (AIRB) Approach. The Standardized
Approach uses predefined risk weight values set

SPOTLIGHT: IFRS 9
by the regulator, which are not suitable for IFRS parameter that represents the credit
9 requirements. However, banks estimate PD cycle.
under both FIRB and AIRB, which can be used as »» Another option for incorporating
a starting point for calculating IFRS 9-compliant forward-looking information into an
PDs. existing PD is to use a stress testing
approach, where the projected PD
In order to use the Basel framework to obtain
depends upon particular economic
PDs for the IFRS 9 calculation, firms should
scenarios.
consider the following adjustments:
»» A third option is to develop a PD
I. Align the Basel definition of default and the
model that incorporates the current
institution’s risk management practice.
explanatory variables as well as
IFRS 9 states that firms shall apply a
forward-looking variables, such as
definition of default consistent with the
forecasts of macroeconomic variables
definition used for internal credit risk
and/or signals from the equity market.
management purposes. However, there is a
rebuttable presumption that a default does In addition to the aforementioned three
not occur later than when the instrument approaches used for developing the PD
is 90 days past-due. The firm may rebut model, firms can also consider simulating
the presumption if it has reasonable and individual loan and collateral performance,
supportable information to determine that a as well as corresponding market conditions,
more lagging criterion is more appropriate. based on historical probability distributions.

With a sufficiently large number of
II. Apply adjustment for economic cycle and
simulation paths, the final PD becomes an
incorporate forward-looking information.
unconditional risk measure, which reflects a
The desire for stable capital requirement
probability-weighted outcome as required by
estimates leads many banks to adopt
IFRS 9.
through-the-cycle (TTC)4 PDs. Since IFRS
9 requires firms to incorporate information III. Calculate lifetime PDs.
regarding current conditions and forecasts To calculate lifetime expected loss, users
of future conditions, TTC PDs require a cycle must construct a term structure of PDs
adjustment incorporating forward-looking beyond one year. Different modeling
information. techniques include:

»» Develop separate models for different
In particular, firms can leverage TTC PDs and
time horizons and interpolate
apply a cyclical adjustment. The adjustment
probabilities of default for intermediate
can be based on credit cycle signals from
maturities; developers must ensure that
macroeconomic variables or information from
PDs of long horizons are higher than
the equity or debt markets, which incorporate
PDs of short horizons.
market participants’ expectations and
»» Develop a model that uses the most
therefore reflect forward-looking information.
up-to-date information at each point
Since the credit cycle affects industries
in time. The resulting PD is not time-
in different ways, adjustments should be
dependent, but requires forecasting the
industry-specific. If the credit signals show
risk factors for each loan’s lifetime.
an increase in risk level, PD levels should be
»» Use transition matrices, which measure
adjusted upward. If the risk level falls, PD
the probability of moving between
levels should be adjusted downward.
credit categories.
»» One possible implementation
»» Develop a model in which the PD is
adjustment is via a Z factor, as
time-dependent.
illustrated in Aguais, et al.,5 a single
4 While there is no universally agreed-upon definition, the conventional view is that a rating system or a PD model with outputs
that remain relatively stable across different macroeconomic conditions is a TTC system.
5 See Belkin, Suchower, & Forest (1998a, 1998b); Aguais, et al. (2004, 2006).

LGD Models
Some firms also develop internal LGD models Alternatively, firms can develop an LGD
for Basel and risk management purposes, which model that uses the most up-to-date
they can leverage for IFRS 9. To use the Basel information at each point in time. The
framework to obtain IFRS 9-compatible LGDs, resulting LGD requires that each explanatory
firms should make the following adjustments: variable is forecast for the entire lifetime
of each loan, but it does not require an
I. Remove the downturn component.
assumption on the LGD term structure.
IFRS 9 states expected loss estimations
should reflect current and forward-looking V. Other adjustments.
expected losses, not downturn economic Basel LGD estimations may include indirect
conditions. This method disregards the costs related to collecting on the exposure
conservative approach suggested by the and credit derivatives used as risk-mitigating
Banking Supervision Committee. Therefore, instruments. Since IFRS 9 requirements do
for IFRS 9 purposes, the downturn component not include these components, they must be
should be removed. removed.
II. Adjust the discount rate. EAD Models

Basel does not specify which discount
For financial instruments with predetermined
rate to use for estimating the LGD. IFRS 9
draw and amortization terms (e.g., term
requires using the effective interest rate or
loans and bonds), EAD in future periods can
an approximation thereof. Therefore, in order
be calculated from known contractual terms
to use Basel models, firms should align the
during the cash flow generation process, taking
interest rate used or apply an adjustment.
into account probability of prepayment for
Under IFRS 9, firms must also discount
prepayable loans and the probability of the call
expected losses to the reporting date, while
(or similar) options being exercised for bonds
Basel states discounting to the default date.
with contingencies.
III. Incorporate forward-looking information.
For irrevocable loan commitment and line of
Since IFRS 9 requires expected loss to be
credit with a loan, and an undrawn commitment
forward-looking, firms should consider
component, firms may need an EAD model to
building a cyclical adjustment into the LGD
estimate the instrument’s exposures to credit
model. Similar to PD, the stress testing
losses. One option is to leverage the Basel EAD
approach can also be used to produce a
model, used under the AIRB approach.6 Basel
forward-looking LGD. defines EAD as “the expected gross exposure of
To implement the stress testing approach, the facility upon default of the obligor.”7
firms may choose to calculate a probability-
In order to adjust the Basel EAD modeling for
weighted, average LGD across multiple
IFRS 9 purposes, the following modifications are
scenarios or simply use one scenario that
needed:
represents the best future estimate to
i. Remove the downturn component.
produce a single LGD. The implementation
If the EAD Basel estimation includes an
method should be consistent with PD and
economic downturn component, it should be
LGD.
removed.
IV. Extend the term structure.
ii. Extend the term structure.
As Basel models typically have a one-year
For IFRS 9 purposes, the Basel EAD models
horizon, they should be extended to provide a
should be extended beyond a one-year
term structure for LGDs.
6 Under the Standardized and the Foundation Internal Ratings-based Approaches, firms have less flexibility with EAD calculation.
7 Basel Committee on Banking Supervision (2015).

SPOTLIGHT: IFRS 9
horizon in order to cover the expected life of for calculating the 12-month or lifetime expected
the financial instrument. losses, including the expected loss approach
based on PD/LGD/EAD modeling or loss rate
Loss Rate Method approach. These models can be developed
Unlike the PD/LGD/EAD modeling approach internally or provided by vendors.
discussed above, loss rate models estimate credit
Credit risk models developed for Basel capital
losses by aggregating PD, LGD, and EAD. These
requirement calculation or stress testing
models are often used for short-term portfolios
purposes can be leveraged for IFRS 9 expected
such as credit cards, trade and lease receivables,
credit loss calculation as well. The forward-
and some non-material exposures. In addition,
looking information required by IFRS 9 can be
medium- or small-sized firms often rely on these
incorporated into credit risk models based on
simple modeling approaches for loss allowance
signals from macroeconomic variables or from
calculations.
the equity or debt markets. Possible approaches
Commonly used loss rate models include: for incorporating forward-looking information
»» Net charge-off rate model include transition matrices, scenario-dependent
estimations, and simulation approaches. Firms
»» Roll-rate model
must extend the one-year PD, LGD, and EAD
»» Vintage loss curve model estimations to the instrument’s lifetime, for
which different statistical techniques can be
Summary
used. Possible techniques include transition
To address the new IFRS 9 impairment model
matrices, time-dependent models, separate
requirements, we recommend firms use a more
models for different time horizons, and models
granular and dynamic approach for portfolio
that use the most up-to-date information at
segmentation by grouping financial assets
each point in time. Further, institutions will need
based on shared credit characteristics that
to incorporate specific adjustments to models
typically react in a similar way to the current
developed for Basel requirements.
environment and forward-looking information.
Firms can implement different credit risk models
Aguais, Scott, et al., M. Ong (ed.), Designing and Implementing a Basel II Compliant PIT-TTC Ratings Framework, The Basel
Handbook: A Guide for Financial Practitioners, London: Risk Books, 2006.
Aguais, Scott, et al., M. Ong (ed.), Point-in-Time versus Through-the-Cycle Ratings, The Basel Handbook: A Guide for Financial
Practitioners, London: Risk Books, 2004.
Bao, Eric, Maria Buitrago, Jun Chen, Yanping Pan, Yashan Wang, Jing Zhang, and Janet Zhao, IFRS 9 Impairment Regulations:
Implementation Challenges and Potential Solutions, Moody’s Analytics white paper, December 2015.
Basel Committee on Banking Supervision, Guidance on Accounting for Expected Credit Losses, February 2015.
Belkin, Barry, Stephan Suchower, and Lawrence Forest, The Effect of Systematic Credit Risk on Loan Portfolios and Loan Pricing,
Credit-Metrics Monitor, pp 17-28, 1998a.
Belkin, Barry, Stephan Suchower, and Lawrence Forest, A One-Parameter Representation of Credit Risk and Transition Matrices,
Credit-Metrics Monitor, pp 17-28, 1998b.
Deloitte, Fifth Global IFRS Banking Survey: Finding Your Way, September 2015.
Ernst and Young, Facing the Challenges: Business implications of IFRS 4, 9 and Solvency II for insurers, 2015.
Gea-Carrasco, Cayetano, IFRS 9 Will Significantly Impact Banks’ Provisions and Financial Statements, Risk Perspectives Magazine,
June 2015, Moody’s Analytics.
International Accounting Standards Board, International Financial Reporting Standard: IFRS 9 Financial Instruments, July 2014.
International Accounting Standards Board, Implementation Guidance: IFRS 9 Financial Instruments, July 2014.
International Accounting Standards Board, Basis for Conclusions: IFRS 9 Financial Instruments, July 2014.
International Accounting Standards Board, IFRS 9 Project Summary, July 2014.
McPhail, Joseph and Lihong McPhail, Forecasting Lifetime Credit Losses: Modeling Considerations for Complying with the New FASB
and IASB Current Expected Credit Loss Models, Working paper, 2014.

PROBABILITY-WEIGHTED OUTCOMES UNDER
IFRS 9: A MACROECONOMIC APPROACH
By Barnaby Black, Glenn Levine, and Dr. Juan M. Licari
In this article, we discuss development of a framework that addresses

Barnaby Black
Assistant Director, Economist the forward-looking and probability-weighted aspects of IFRS 9
impairment calculation using macroeconomic forecasts. In it, we
address questions around the practical use of alternative scenarios
Barnaby Black is an Assistant Director in the Moody's and their probabilities. We also include a case study to illustrate
Analytics London office. His responsibilities involve these concepts in practice.
leading client projects regarding IFRS 9 impairment,
capital and impairment forecasting, and stress testing,
particularly for retail and SME portfolios. He has The incoming IFRS 9 regulation provides for an unbiased estimate of lifetime expected
an MS in Operational Research and Statistics from the use of macroeconomic forecasts and losses. Given the accounting standard’s goal
Southampton University and a BS in the same field
probability-weighted outcomes, particularly of consistency, however, the scenario weights
from the University of Kent.
when accounting for the impairment of financial derived from Option 1 may also be used in
assets. Indeed, the spirit of IFRS 9 suggests Option 2.1
that finance officers should be more forward- Like the IFRS 9 standard itself, this article
looking in their recognition of credit losses on a does not prescribe a specific plan of action
Glenn Levine firm’s balance sheet, with the macroeconomy or a one-size-fits-all approach to the use of
Associate Director, Senior often taking a central place in any impairment macroeconomic forecasts and probability
Research Analyst
forecast. Paragraph B5.5.42, for example, weights. Rather, it is designed to help
“requires the estimate of expected credit losses institutions build a framework that addresses the
Glenn Levine is an Associate Director in the Moody’s to reflect an unbiased and probability-weighted “forward-looking” and “probability-weighted”
Analytics Capital Markets Research Group. He provides amount that is determined by evaluating a range aspects of IFRS 9 impairment calculation
support for the EDF product suite and is the lead of possible outcomes.” More specifically, two using macroeconomic forecasts. Moreover, we
researcher for Stressed EDF. Prior to his current role,
key areas of IFRS 9 suggest that macroeconomic provide a purely quantitative approach to the
he was a Senior Economist in the Economics and
scenario forecasts may be utilized: problem. The use of qualitative overlays, which
Consumer Credit division based in Sydney, Australia.
He holds an MS from the London School of Economics 1. Section 5.5.3, which outlines the method for are allowable within the framework of IFRS 9, is
and a bachelor’s degree from the University of New
calculating lifetime expected credit losses beyond the scope of this article.
South Wales.
once an instrument has passed from Stage 1 This report outlines three areas of discussion for
to Stage 2. banks to consider:
2. Section 5.5.9, which describes the procedure 1. The number of macroeconomic scenarios to
for assessing whether an instrument has utilize
undergone a significant deterioration in
2. How to ensure an unbiased probability-
credit risk.
weighted outcome
This report focuses primarily on Option 1 above, 3. Where in the impairment calculation
and how probability weights can be derived to incorporate the macroeconomy and
from macroeconomic forecasts to produce probability weights
1 See paragraph 54 of IFRS staff paper, Transition Resource Group for Impairment of Financial Instruments, December 11, 2015

SPOTLIGHT: IFRS 9
The report concludes with an example from the the third approach uses an unconditional PD
wholesale lending space, which illustrates three that does not require a specific macro scenario
Dr. Juan M. Licari
different approaches to IFRS 9 compliance. or probability weighting. Similarly, there is an Managing Director, Chief
upper limit to the number of scenarios that may International Economist
How Many Macroeconomic Scenarios?
be appropriate. Section BC5.265 suggests,2 “The
The IFRS 9 standard does not explicitly define
calculation of an expected value need not be
the number of macroeconomic scenarios that Dr. Juan M. Licari is a managing director at Moody's
a rigorous mathematical exercise whereby an
should be used for impairment calculations. Item Analytics and the head of the Economic and Consumer
entity identifies every single possible outcome Credit Analytics team in EMEA. Dr. Licari’s team
B5.5.42 is again instructive:
and its probability,” so the requirement of a provides consulting support to major industry players,
builds econometric tools to model credit phenomena,
and has implemented several stress testing platforms
to quantify portfolio risk exposure. He has a PhD
The language used by IFRS 9 is intentionally vague, and the interpretation
and an MA in economics from the University of
of the number and type of economic scenarios will differ by firm, portfolio Pennsylvania and graduated summa cum laude from
complexity, geographical spread, and local regulator. the National University of Cordoba in Argentina.
"In practice, this may not need to be a complex simulation-based approach over thousands of
analysis. In some cases, relatively simple scenarios can be disregarded.
modelling may be sufficient, without the need
The language used by the standard is
for a large number of detailed simulations of
intentionally vague (“at least two”), and the
scenarios. For example, the average credit losses
interpretation of the number and type of
of a large group of financial instruments with
economic scenarios will differ by firm and
shared risk characteristics may be a reasonable
portfolio complexity. In this article, we outline
estimate of the probability-weighted amount. In
three approaches, two of which use multiple
other situations, the identification of scenarios
economic scenarios covering both upside and
that specify the amount and timing of the cash
downside possibilities. This seems appropriate for
flows for particular outcomes and the estimated
most firms and most portfolios as the standard
probability of those outcomes will probably be
is designed for firms to consider a representative
needed. In those situations, the expected credit
sample of the complete distribution.3 The
losses shall reflect at least two outcomes in
framework can be extended to incorporate more
accordance with paragraph 5.5.18.” (Emphasis
scenarios or greater complexity.
added.)
How to Ensure an Unbiased Probability-
In some limited cases, then, the use of one or
Weighted Outcome Using Alternative
even zero economic scenarios may be sufficient. Macroeconomic Scenarios
The illustrative example below, from the Moody’s Analytics economics division produces
wholesale sector, outlines three approaches monthly off-the-shelf macroeconomic forecasts
to the problem. The first two methods utilize under a baseline scenario and several alternative
macro scenarios and probability weights, while economic scenarios, known as S1 through S4.
2 See Page 5, Section 10, of Incorporation of forward-looking scenarios by the Transition Resource Group (IFRS staff paper,
December 11, 2015).
3 See paragraph 46(b) of IFRS staff paper.

These forecasts cover 54 countries and over guidelines require expected losses to be
90% of the world’s GDP. Each scenario has a calculated on the probability-weighted mean
probability attached to it based on its historical of the distribution, not the median, so even if
distribution. a single scenario were to be used, the baseline
The baseline is a 50% scenario, implying a 50% may not be appropriate. (Other economists may
probability that the actual outcome is worse forecast the mode – the most likely outcome –
than the baseline forecast, broadly speaking, and which is also inappropriate, without overlays,
a 50% probability that the outcome is better. within IFRS 9.) These scenario probabilities
Similarly, the S1 upside scenario has a 10% describe a cumulative distribution function
probability attached to it (10% probability that (CDF) showing probabilities for the economy to
the outcome is better; 90% probability that perform better or worse than a given forecast
the outcome is worse); S2 is a 25% downside (Figure 1).
scenario; S3 is a 10% downside scenario; and S4 An expected value can be derived from a CDF in
is a 4% downside scenario. Moody’s Analytics two ways. First, we could “integrate” the CDF to
Figure 1 Scenario probabilities – cumulative distribution function
1.0
0.8
0.6
0.4
0.2
0.0
LO S4 S3 S2 BL S1 H1
also internally produces two “bookend” calculate the area under the curve. This would
scenarios, which are 1-in-10,000 probability give a single mean economic outcome that could
events that describe the upper and lower bounds be conditioned on in expected loss calculations.
of possible economic outcomes. These bookend However, as will be discussed later, it may be
scenarios help to illustrate the theoretical preferable to use several economic scenarios,
approach, but were excluded from the following push these scenarios across credit expected
wholesale example as the guidance recommends loss inputs (PDs, EADs, LGDs), and then weight
that firms “should not estimate a worst-case these scenario-conditional risk parameters by
scenario nor the best-case scenario.”4 the scenario probabilities. A second option is
The baseline scenario is therefore the median to “differentiate” the CDF, or take its slope at
outcome, and not the mean. The IFRS 9 each point, to produce a probability distribution
4 See paragraph 46(d) of IFRS staff paper

SPOTLIGHT: IFRS 9
function (PDF). Figure 2 describes the PDF using scenario may not be appropriate if the
US GDP (in billions of 2009 USD). relationship between credit losses and the
We can calculate an expected value by using macroeconomy is nonlinear. This will often be
the probability masses from this PDF to weight the case in a properly specified credit model.
either the economic data or the credit outcomes Moreover, even if the credit estimate is unbiased,
conditioned on that economic data, depending a single weighted scenario may be undesirable
on which stage of the process the weights are as the standard emphasizes evaluating a range
applied to. of outcomes, not a range of scenarios. This is

because firms may gain additional insight into
How and Where to Incorporate Macroeconomic the exposure of their portfolio by assessing a
Scenarios and Probability Weights
distribution of credit outcomes.
IFRS 9 provides no explicit guidance on how the
probability-weighted outcomes should be used, This can be illustrated through a simple example.
although we can glean some insight from the Imagine it is 2006 and there are two firms that
standard itself and follow-up discussions. For both have a large subprime mortgage exposure.
Figure 2 Scenario probabilities – probability density function
0.01
0.008
0.006
0.004
0.002
0
LO-S4 S4-S3 S3-S2 S2-BL BL-SL S1-H1
example, using the above approach, should the Firm A models its expected credit losses under a
probability weights be applied to the economic single, probability-weighted economic scenario,
data to produce a single, probability-weighted showing only mild credit losses under this
economic scenario which is then put through scenario. Firm B, however, uses several economic
the credit model? Or should the user put all scenarios and notices that while its expected
relevant scenarios through the credit model probability-weighted credit losses are modest,
and then apply the scenario weights to obtain a its losses under a sharp recession (such as S4)
probability-weighted credit outcome? are severe enough to put it out of business.
Public discussion at the Transition Resource From an accounting perspective, both Firm A and
Group for Impairment of Financial Instruments Firm B may recognize similar expected losses
emphasized that using a single macroeconomic under IFRS 9. Yet from a statistical perspective,

the measure of expected credit losses recognized deteriorated by, say, three or more notches,
by Firm A may be biased because of the non- would determine stage allocation.
linear relationship between credit losses and
Once an instrument has passed into Stage 2,
the macroeconomy. And from a risk manager’s
lifetime EL must be calculated and accounted.
perspective, the information available from Firm
There are three options for performing this
B’s accounting of expected losses provides a
calculation, based on the discussion in previous
far richer information set and the possibility to
sections:6
take mitigating action if the risk of an S4-type
1. Apply the economic scenario probability
scenario is considered material.
weights to the Stressed EDF forecasts
Figure 3 JP Morgan, scenario conditional 1-year PD curves & weighted average (expected value)
2.5 History (EDF)

Baseline
2.0 S1
S2
1.5 S3
EDF (%)
S4
1 Weighted Average
0.5
0
Jan-15 Jan-15 Jan-16 July-15 Jan-15 Jan-18 Jan-18
Case Study: Wholesale Portfolio Example

The Moody’s Analytics CreditEdge model produced by conditioning on those economic
provides a suitable framework for IFRS 9 scenarios. This is our recommended approach
compliance with C&I exposures. This example 5 for the reasons outlined previously. Figure 3
uses the EDF metric, which provides an illustrates this example using JP Morgan. The
unconditional firm-level PD with a tenure of probability-weighted PD is above the baseline
one to 10 years, and the Stressed EDF satellite through the forecast period.
model, which uses the core EDF metric to 2. Combine the economic scenarios into a
provide a firm-level PD forecast conditioned on single probability-weighted scenario. This
any economic scenario. Stressed EDF already will, however, produce a biased measure
produces monthly forecasts conditioned on the of lifetime EL if the relationship between
Moody’s Analytics scenarios described above, the macroeconomy and PD is non-linear.
and so is well-suited to this purpose. This is the case, by design, with the Stressed
In this example, we may decide IFRS 9 stage EDF model. Moreover, it glosses over the
allocation by comparing the unconditional EDF potential distribution of credit losses. Figure 4
at the reporting date with EDF at origination to illustrates the approach.
determine whether a significant deterioration 3. A third option is to use the unconditional
in credit risk has occurred. A suitable criterion, EDF to calculate lifetime EL. That is, if an
such as if a firm’s implied credit rating has instrument has eight years left until maturity,
5 A modeling approach for retail portfolios is detailed in Black, Chinchalkar, & Licari, Complying with IFRS 9 Impairment
Calculations for Retail Portfolios, Risk Perspectives Magazine, June 2016, Moody’s Analytics
6 The IFRS staff paper outlines three approaches that broadly mirror the three options, plus a fourth possibility which uses the
modal or most likely economic scenario in combination with a qualitative overlay. Our recommendations borrow heavily from
this directive.

SPOTLIGHT: IFRS 9
simply use the eight-year EDF as a measure Figure 5 summarizes the results using the two-
of lifetime PD. The resulting EL calculation year cumulative PD from each approach.
can be considered a weighted distribution
Concluding Remarks
of economic scenarios. That is, the EDF
In this article we have analyzed the use of
metric combines data from a firm’s balance
macroeconomic scenarios as part of the
sheet with the firm’s stock price, which is
forward-looking, probability-weighted IFRS 9
also the market’s expectation of discounted
framework. Some of the key questions around
future profits, with every possible profit path
the practical use of alternative scenarios and
weighted by the probability of that path
their probabilities have been answered, and a
Figure 4 JP Morgan, 1-year PD conditioned on a single probability-weighted average economic scenario
0.6
0.5
0.4
History (EDF)
EDF (%)
0.3 Stressed EDF
0.2
0.1
0
Jan-15 Jan-15 Jan-16 July-16 Jan-17 Jan-17 Jan-18 Jan-18
Figure 5 JP Morgan, two-year cumulative PD under the three approaches (%)
Two-Year Cumulative PD
1. Probability weighted on the SEDFs 1.19
2. Probability weighted on the economic scenarios 0.86
3. Unconditional EDF 0.91
occurring. The measure is also unbiased by case study illustrates these concepts in practice.
the construction of the EDF model, which is We argue in favor of leveraging a handful of
calibrated to physical default probabilities alternative forecasts in order to comply with
using Moody’s Analytics default database. recent regulation. The shape and severity of the
scenarios can vary over industries and firms,
All three options may be suitable in different but the regulatory language is fairly clear when
situations, depending on the relationship requesting the need to account for alternative
between credit risk and the macroeconomy and outcomes under a probability-weighted
the desired objective of the reporting process. framework.

COMPLYING WITH IFRS 9 IMPAIRMENT
CALCULATIONS FOR RETAIL PORTFOLIOS
By Barnaby Black, Dr. Shirish Chinchalkar, and Dr. Juan M. Licari
This article discusses how to address the specific challenges

Barnaby Black
Assistant Director, Economist that IFRS 9 poses for retail portfolios, including incorporating
forward-looking information into impairment models, recognizing
significant increases in credit risks, and determining the length of an
Barnaby Black is an Assistant Director in the Moody's instrument’s lifetime. We describe two approaches to analyzing retail
Analytics London office. His responsibilities involve portfolios, suggest practical interpretations of IFRS 9 guidelines, and
leading client projects regarding IFRS 9 impairment,
capital and impairment forecasting, and stress testing,
answer common questions pertaining to retail portfolios.
particularly for retail and SME portfolios. He has
an MS in Operational Research and Statistics from
Introduction »» Unbiased point-in-time estimates of the
Southampton University and a BS in the same field
from the University of Kent. When building and implementing econometric expected credit losses have to be computed
models for different asset classes, the modeler by using a probability-weighted amount
needs to carefully examine the requirements that is determined by evaluating a range of
from the perspective of the final users of the possible outcomes.
models. A trader of whole loans may be more
»» Depending on the stage into which an
Dr. Shirish Chinchalkar interested in the accurate modeling of loan-
asset is classified, either a 12-month or a
Managing Director – level cash flows and exploiting any statistical
Quantitative Finance lifetime expected credit loss may have to be
arbitrage. A servicer is likely to be concerned
determined.
about delinquency transitions and time to
Dr. Shirish Chinchalkar heads the group responsible for liquidation. Regulatory stress testing requires This paper addresses these considerations for
building credit models for different retail asset classes that the models demonstrate sensitivity to retail portfolios. Retail portfolios can be analyzed
such as US, UK, and Dutch residential mortgages, macroeconomic conditions. Risk management with either a bottom-up approach or a top-down
US auto loans, and credit cards, as well as for several
requires that the models correctly capture the one. We discuss each of these approaches in
different US and EMEA asset-backed securities. His
group also implements the models in the Portfolio correlation between different assets in the more detail below.
Analyzer platform which is used for stress testing and portfolio.
Bottom-Up Approach
risk management of retail and structured portfolios.
The IFRS 9 guidelines pose some interesting A bottom-up approach involves constructing
He has a PhD from Cornell University and a BTech from
IIT Bombay. challenges, including the following: loan-level models for each loan in the portfolio.
»» An important consideration in the Results can be aggregated over all the loans
impairment model in IFRS 9 is the use of in different cohorts or segments to arrive at
forward-looking information in the models. segment-level or portfolio-level results.
Decisions around classification of assets into Loan-level models are usually hazard rate
different stages and the calculation of the models and can be constructed in a competing
expected credit losses require consideration risk framework. The data is naturally organized as
of forward-looking macroeconomic panel data; each loan has multiple observations
information. through time. Defaults and prepayments
»» A classification of assets into different compete with each other in a multi-period
stages requires determination of a significant setting. Survival models in this framework can be
increase in the credit risk. built using a panel logit model.

SPOTLIGHT: IFRS 9
A bottom-up approach has the advantage that Segment-level models are easier to build
the results are naturally available at the highest because they typically require less data. The
Dr. Juan M. Licari
level of granularity. The explanatory variables, performance variables of interest can be Managing Director, Chief
such as loan and borrower characteristics and directly modeled as a function of segment-level International Economist
macroeconomic variables, are used at the loan characteristics. Models can be implemented
level. Likewise, the performance variables, such faster.
Dr. Juan M. Licari is a managing director at Moody's
as defaults, prepayments, cash flows, and losses,
Segment-level models are better suited to Analytics and the head of the Economic and Consumer
are modeled at the loan level. Heterogeneity Credit Analytics team in EMEA. Dr. Licari’s team
homogeneous portfolios. When a portfolio
of the loan characteristics can be easily provides consulting support to major industry players,
consists of heterogeneous assets, several
accommodated. builds econometric tools to model credit phenomena,
segments are needed in order to accurately
and has implemented several stress testing platforms
Building loan-level models requires reliable model the portfolio along multiple dimensions. to quantify portfolio risk exposure. He has a PhD
historical loan-level data. This can be onerous If portfolio composition changes through time or and an MA in economics from the University of
and expensive. If the loan-level data is not if assets migrate from one segment to another, Pennsylvania and graduated summa cum laude from
the National University of Cordoba in Argentina.
reliable, the models that are built may have to greater care is needed in segmenting the
be recalibrated. The implementation can also portfolio.
require additional resources. In situations where
The suitability of a particular model will thus
the portfolio consists of a large number of
depend on the type of collateral and the
homogeneous assets, a loan-level approach may
portfolio composition.
not be necessary.
A bottom-up approach has the advantage that the results are naturally
available at the highest level of granularity. Heterogeneity of the loan
characteristics can be easily accommodated.
Top-Down Approach Next, we turn our attention to specific guidelines

A top-down approach involves segmenting the in the IFRS 9 standard, how they apply to retail
portfolio by vintage and risk characteristics. portfolios, and how we can address them while
The segments can be as coarse or as granular building and implementing the models.
as required. For example, a large homogeneous
How to Use Forward-Looking Information
portfolio of retail credit cards could be modeled
One of the key issues identified with IAS 39
as a single segment. On the other hand, a
impairment regulation is that only past events
portfolio may be segmented by vintage,
and current conditions can be considered in
geographic regions, or the borrower’s risk profile.
measuring credit losses. This leads to a notable
Different segments can have different models.
weakness in the models developed under IAS 39
Results can be either aggregated further to arrive
standards: that there can be delayed recognition
at the portfolio-level results or calibrated further
of credit losses. In addition to considering
to instrument-level figures.

past events and current conditions, the new the change in credit risk cannot be assessed
standard requires that forecast information must simply by comparing the change in the absolute
be used in measuring Expected Credit Losses risk of a default occurring over time. For example,
(ECL) if available without undue cost or effort. if the risk of a default occurring for a financial
Forward-looking information is to be used for instrument with an expected life of 10 years at
stage allocation as well as for the calculation initial recognition is identical to the risk of a
of the ECL. We discuss how forward-looking default occurring on that financial instrument
information can be incorporated in the models. when its expected life in a subsequent period is
The use of this information for stage allocation only five years, that may indicate an increase in
and for calculation of the ECL is discussed in later credit risk. This is because the risk of a default
sections. occurring over the expected life usually decreases
An econometric model of the retail assets, as time passes if the credit risk is unchanged and
whether it is done at the cohort level or the the financial instrument is closer to maturity.”
loan level, involves relating the performance of An econometric panel data modeling approach
the assets to macroeconomic factors. Once this can help identify a forward-looking lifetime
relationship is established, forecasting the losses PD at the latest reporting date, but a question
or determining the lifetime PD requires using is raised concerning if and how to determine a
these models on prescribed scenarios. forward-looking lifetime PD at origination. To
One way to include forward-looking information use forward-looking information historically, one
is to incorporate econometric panel data models could either leverage historical macroeconomic
that will give risk parameter forecasts under scenarios on a monthly basis, or adjust
multiple scenarios. As the stage allocation origination PDs with historical macroeconomic
should use the change in the risk of a default data (i.e., utilizing what has actually happened in
occurring over the expected life of the financial the economy since origination).
Our recommendation is to incorporate forward-looking information into

the assessment of lifetime PDs for lending originated solely after the
implementation of IFRS 9.
instrument, banks need to determine the extent Our recommendation is to incorporate forward-
to which forward-looking information will be looking information into the assessment of
included in lifetime PDs. lifetime PDs for lending originated solely after
Our interpretation of the quantitative metric the implementation of IFRS 9. For lending
required for determining stage allocation is originated prior to the implementation of
the change in the lifetime PD of an instrument IFRS 9, lifetime PDs at origination can reflect
since origination, relative to age. The caveat the assessment of credit risk at the time of
that the change is relative to age is essential origination, which may not include forward-
for retail portfolios, as the lifetime PD will looking information. The usage of historical
depend on time until derecognition (retail origination PDs for instruments originated prior
portfolios show a strong lifecycle component: to IFRS 9 implementation is justified by the
nonlinear relationship between PD and time- following standards:
since-origination). This is highlighted in appendix »» Prior to the introduction of IFRS 9, there was
paragraph B5.5.11 of the standard: no explicit requirement for forward-looking
“Because of the relationship between the information to be used to adjust historical
expected life and the risk of a default occurring, estimates of PD.

SPOTLIGHT: IFRS 9
»» The work required to adjust historical PDs product type, region, or LTV band.
to incorporate forward-looking information
Factors influencing vintage-segment
would be considerable, going against the
performance can be conceptually divided
clause to “use information that is available
into three classifications: the lifecycle trends
without undue cost or effort” (B7.2.2).
depending on a loan’s age-on-books (seasoning),
The rationale behind stage allocation, which the factors indicating the quality of a vintage,
requires origination PD, is to compare the current and the characteristics of the current economic
view of default risk with the view that was held environment that depend only on calendar time.
when the lending was agreed and the product Other effects also operate across more than one
was priced. To adjust the origination PDs for of these main categories and can be modeled
forward-looking information (or any additional as interactions between them. An examination
data) would be inconsistent with this aim. of each of these types of effects in isolation is
The only adjustments that should be made to essential to understanding the multidimensional
origination PDs are ensuring these are unbiased nature of the data and the models used to
point-in-time best estimates of the lifetime PD. forecast it. Using this approach on a panel data
of marginal default rates and loss rates can help
In summary, our recommendation to address
provide 12-month and lifetime ECL.
the forward-looking aspect of the standard is to
use panel data (vintage or loan-level models) Consider a bank that already has 12-month
using macroeconomic drivers for retail portfolios. Point-in-Time (PiT) Basel models or 12-month
These granular-level outputs can be calibrated Through-the-Cycle (TTC) models with an easily
to instrument-level figures, if required, before extractable PiT component. The bank can
calculating instrument-level IFRS 9 impairment. achieve IFRS 9 compliance through a scaling
The inclusion of macroeconomic variables allows process that leverages the vintage-level outputs
the estimation of ECL under several different to provide account-level lifetime expected
scenarios and the generation of probability- credit losses that are consistent with the Basel
weighted outcomes. This approach captures both 12-month PiT outputs.
a range of forecasts and the non-linearity in the For banks that use 12-month TTC models, with
ECL calculation. no possibility of extracting PiT outputs, our
proposal is to use a vintage-level panel data
How to Calculate Unbiased Point-in-Time
Estimates modeling approach to estimate 12-month and
Paragraph 5.5.17(a) of the standard states lifetime ECL. The distribution of TTC PDs for a
that an entity shall measure ECL of a financial given vintage can then be used as a benchmark
instrument in a way that reflects an unbiased to define the ranking distribution of the account
amount. Therefore, banks need to consider PiT PDs around the new vintage-level mean.
if and how their existing capital models and When using a loan-level model, an unbiased
methodologies can be leveraged. This decision point-in-time estimate can be obtained by
will likely be driven by the level at which using models that incorporate lifecycle effects
downturn adjustments are incorporated into and macroeconomic factors. Using survival
their existing models. models, the dependence of the PD of a loan on
Banks that lack suitable models for IFRS its seasoning can be captured. Therefore, newly
9 purposes can use a panel data modeling originated loans will behave differently from
approach where the data is split by vintages. For seasoned loans. Loan-level models use past
PD modeling, the vintages refer to the month information through changes in macroeconomic
or quarter of origination, whereas for LGD variables such as home prices and
modeling, the vintages refer to the month or unemployment rates from loan origination to the
quarter of default. Data can be split by further reporting date, and forward-looking information
levels of client-specific segmentation such as such as future changes in macroeconomic

conditions. The output of the loan-level models financial asset transfer. This asset transfer can be
is a conditional PD or LGD. Through the use of identified by the transferral of the contractual
probability-weighted scenarios described later rights to receive the cash flows of the financial
in the paper, a point-in-time estimate of the ECL asset. There is also a case for asset transfer
can be obtained. where the rights are retained, as detailed in
paragraph 3.2.5 of the standard:
How to Define a Lifetime View of ECL
Paragraph 5.5.4 of the standard states: “When an entity retains the contractual rights
to receive the cash flows of a financial asset
“The objective of the impairment requirements is
(the ‘original asset’), but assumes a contractual
to recognise lifetime expected credit losses for all
obligation to pay those cash flows to one or more
financial instruments for which there have been
entities (the ‘eventual recipients’), the entity
significant increases in credit risk since initial
treats the transaction as a transfer of a financial
recognition.”
asset if, and only if, all of the following three
We address two challenges relating to providing conditions are met:
a lifetime view of risk parameters: first, how to
(a) “The entity has no obligation to pay
determine the length of an instrument’s lifetime,
amounts to the eventual recipients unless
and second, how to model risk parameters over
it collects equivalent amounts from the
the lifetime.
original asset. Short-term advances by
In regards to determining the length of an the entity with the right of full recovery of
instrument’s lifetime, the standard states in the amount lent plus accrued interest at
paragraph 5.5.19: market rates do not violate this condition.
“The maximum period to consider when (b) “The entity is prohibited by the terms of the
measuring expected credit losses is the maximum transfer contract from selling or pledging
contractual period (including extension options) the original asset other than as security to
over which the entity is exposed to credit risk and the eventual recipients for the obligation
not a longer period, even if that longer period is to pay them cash flows.
consistent with business practice.”
(c) “The entity has an obligation to remit
With retail products in mind, this raises the any cash flows it collects on behalf of
following two questions for how to define the the eventual recipients without material
lifetime: delay. In addition, the entity is not entitled
1. When the terms and conditions of an to reinvest such cash flows, except for
instrument are amended, should this lead to investments in cash or cash equivalents (as
the derecognition of the financial asset and defined in IAS 7 Statement of Cash Flows)
the recognition of a new asset? during the short settlement period from
the collection date to the date of required
2. For revolving products such as credit cards
remittance to the eventual recipients, and
or overdraft facilities where the contractual
interest earned on such investments is
period can be as little as one day, should the
passed to the eventual recipients.”
lifetime for these products only be one day?
Lifetime Definition – Question 2: Challenges in
Lifetime Definition – Question 1: Derecognizing
Retail Revolving Credit
Assets
The second question is addressed in paragraph
Chapter 3 of the standard answers the first
5.5.20:
question by defining both recognition and
derecognition. Derecognition occurs through “Some financial instruments include both a
both the expiration of the contractual period and loan and an undrawn commitment component

SPOTLIGHT: IFRS 9
and the entity’s contractual ability to demand of the standard defines the significant increase
repayment and cancel the undrawn commitment in credit risk as a significant “change in the risk
does not limit the entity’s exposure to credit of a default occurring over the expected life of
losses to the contractual notice period. For such the financial instrument.” This suggests that
financial instruments, and only those financial the decision should be based on the change
instruments, the entity shall measure expected in the lifetime PD since origination; however,
credit losses over the period that the entity there is little guidance around what quantifies
is exposed to credit risk and expected credit a significant change. We need to first define
losses would not be mitigated by credit risk this change in the risk of default and then set
management actions, even if that period extends a threshold to determine what constitutes a
beyond the maximum contractual period.” significant increase. There are many options
In order to measure over the expected period for the exact metric by which to allocate
that these entities are exposed to credit risk, instruments into stages for retail portfolios.
we propose a collective assessment of lifetime Absolute Change

length using behavioral data. A panel data
The absolute change in the lifetime PD since
modeling approach can be leveraged to model
origination is the simplest metric for calculating
the proportion of a vintage’s instruments that
the change, but would likely result in complexity
are recognized over age.
for the threshold assessment. The age of the
For non-revolving products, our interpretation instrument at the latest reporting date would be
for defining the lifetime view is more a key driver, leading to an age-specific threshold.
straightforward. The maturity date of the
Other dimensions that might warrant
contractual period helps provide the end date
consideration are the length of the remaining
for the calculation of each instrument’s lifetime
lifetime and the size of the lifetime PD at
length. Thus, for any given reporting date, the
origination. If the lifetime PD at origination is
remaining lifetime over which to determine risk
very low, the instrument could still be classified
parameters is simply the time to maturity.
within stage 1 after an increase in risk if the PD
Calculating Probability-Weighted Expected is still considered low risk. This is documented
Credit Losses in paragraph 5.5.10, which states, “An entity
The measurement of an expected credit loss may assume that the credit risk on a financial
requires calculation of expected present value instrument has not increased significantly since
of the cash shortfalls. These credit losses are initial recognition if the financial instrument
to be weighted using the probability of default. is determined to have low credit risk at the
Since the models for the PD and LGD use reporting date.” The approach of basing the
macroeconomic drivers and loan and borrower stage allocation off the absolute change in credit
characteristics, the calculation of the expected risk is not recommended as there are so many
credit losses involves projecting the PD, LGD, dimensions that would require assessment.
and cash flows for different macroeconomic
Relative Change
scenarios. An entity need not consider every
possible scenario. However, different scenarios The relative change in the lifetime PD since
with their probabilities of occurrence must be origination would similarly require a number
considered.1 of dimensions in assessing the threshold.
The assessment might be slightly simpler
Defining a Significant Increase in Credit Risk
than using the absolute change, as the size
Stage allocation requires determining if an asset
of the lifetime PD at origination might not
has undergone a significant increase in credit
need to be a dimension. However, the age at
risk since initial recognition. Paragraph 5.5.9
latest reporting date and length of remaining
1 Further discussion on generating scenarios and their associated probabilities can be found in Black, Levine, & Licari, Probability-
Weighted Outcomes Under IFRS 9: A Macroeconomic Approach, Risk Perspectives Magazine, June 2016, Moody’s Analytics.

lifetime would both need consideration, unless remainder of the instrument’s lifetime. Lifetime
combined into a field to measure the percentage views of the Basel PDs can subsequently be
of expected lifetime remaining. This approach extracted, with the lifetime PD curves calibrated
is not recommended due to the complexity to these points, as shown in the graph. The
of capturing the age differences between the metric that can subsequently be used for setting
lifetime PD at origination and lifetime PD at the threshold is the absolute distance between
latest reporting date. Basel-calibrated lifetime PD curves at the age of
Absolute Change in Age-Specific Lifetime PD the latest reporting date.
Forecast Relative Change in Age-Specific Lifetime PD
The absolute change in the age-specific lifetime Forecast
PD forecast compares the lifetime PD at the Whether the absolute distance differs across
latest reporting month (at age α) with the age would need to be assessed. It is likely that
lifetime PD forecast at origination for the the age dimension would remain essential in
instrument once it reaches age α. In Figure 1, a setting the threshold using the absolute change.
vintage of instruments are modeled using the Our recommendation is thus to consider the
panel data modeling approach to give a vintage relative change in the age-specific lifetime PD
lifetime PD curve. The panel data approach forecast. This has the caveat of paragraph 5.5.10
models marginal default rates that can be as mentioned, which states that low values at
accumulated to give the lifetime PD curve. origination can have notable increases but still
Similarly, the marginal default rates can be be classified as stage 1.
accumulated into 12-month PD curves over age.
Considering the options for determining the
The vintage 12-month PD curve can be calibrated
metric above has helped identify some of the
against Basel 12-month PDs to give instrument-
dimensions that need to be assessed when
specific 12-month PD curves through the
quantitatively setting a threshold for stage
Figure 1 Lifetime PD
Vintage Lifetime PD Account Lifetime PD b Distance to assess for

at Origination threshold
Account Lifetime PD at Data from Basel Model calibrated to
Latest Reporting Date unbiased PiT lifetime PDs
8.00%
7.00%
6.00%
5.00%
<
b
4.00%
<
3.00%
2.00%
1.00%
0.00%
0 10 20 30 40 50 60 70 80 90

SPOTLIGHT: IFRS 9
allocation. Below are some of the key areas for Concluding Remarks
consideration: In this paper, we have addressed several
»» The size of the lifetime PD at origination important considerations in the modeling and
implementation of the IFRS 9 standard for retail
»» The age of the instrument at the latest
portfolios. We have shown how these guidelines
reporting date
should be interpreted and how they can be
»» The length of the lifetime remaining incorporated into loan-level and segment/
»» The product type vintage-level models. We have looked to address
some of the key IFRS 9 issues facing banks with
Besides the quantitative assessment, entities a focus on a retail perspective. What is clear
also need to consider whether to rebut the from discussing these issues with peers across
presumption that a financial asset’s PD has the globe is that the standard leaves room for
increased significantly since initial recognition interpretation in defining methodologies to
when contractual payments are more than reach IFRS 9-compliant impairment models.
30 days past due. On top of this, a qualitative Time will tell as to whether any specificities
assessment is recommended for identifying any of methodology design become enforced by
changes in behavior that are not immediately regulators.
captured in an entity’s default definition.

PRINCIPLES AND
PRACTICES
HOW BANKS CAN RAISE THEIR GAME IN
SMALL BUSINESS LENDING
By Nancy Michael
Despite high demand for small business credit, small business

Nancy Michael
Senior Director – Product lending at banks remains depressed and many prospective borrowers
Strategy struggle to find financing. Small businesses cite onerous processes,
lack of transparency, and high search costs among the challenges of
Nancy works to conceive and build innovative obtaining credit through traditional banking channels. Alternative
solutions for credit assessment of small businesses. lenders are capitalizing on scoring and lending technologies
Drawing on her previous experience co-founding a
small business, she has built products and strategies to
developed in retail markets to generate profitable small business
help financial companies better serve the needs of their loans while optimizing the online customer experience, but they
customers. Nancy previously led the Client Solutions often lack deep expertise in risk management, loan monitoring, and
team for the Training and Certification division and
headed Strategy and Marketing for the company’s servicing. Traditional lenders have an opportunity to change their
training and consulting businesses. Nancy received approaches to technology, scoring, and the customer experience to
a BS in Economics from the Wharton School of the
University of Pennsylvania, with concentrations in
meet the competitive challenge of the new market entrants, conduct
Strategic Management and Legal Studies. profitable small business lending, and continue to serve the interests
of their small business customers and local communities.
An interesting phenomenon is occurring in met.1 Prospective small business borrowers cite

business funding in the US. Although the demand onerous processes, lack of transparency, and high
for small business credit remains high, banks search costs among the challenges of obtaining
– the traditional providers of financing for this credit. Research from the Federal Reserve found
segment – have continued a long and steady that small business borrowers spend an average
decline in their position in lending to small of 24 hours on paperwork for bank loans and
businesses. Small loans to businesses are down approach multiple banks during the application
about 15% at banks since the financial crisis. process.2 Banks and borrowers routinely cite
Small business loans represented just 20% of small business loan processing times of weeks or
business loan balances in 2015, continuing a months from completed application to approval,
consistent downward trend from 34% in 1995 not including the time to collect application
(Figure 1). That’s despite an uptick in 2015 that information or fund an approved loan.
almost brought loan balances back to the level These factors have contributed to the rapid
they were a decade ago. growth of online and alternative “marketplace”
Meanwhile, small businesses report persistent lending to small businesses. Marketplace
financing shortfalls, with only half indicating in a lenders focus on using technology to improve
recent survey that all of their credit needs were the customer experience and “time-to-money,”
1 2015 Small Business Credit Survey, Federal Reserve, March 2016.

2 Joint Small Business Credit Survey Report, 2014, Federal Reserve Banks of New York, Atlanta, Cleveland and Philadelphia, 2014
Quarter 3.

often offering turnaround times for credit by a down credit cycle. Delinquency rates and
decisions in less than a day. A recent Morgan net charge-offs for business loans at banks are
Stanley report estimated that online lenders the lowest they’ve been in more than three
granted nearly $8 billion in credit to small decades. Some online lenders are using new
businesses in 2015, reflecting year-on-year techniques and information, like social data and
growth of 68%. At the current rate, they
3
educational backgrounds of borrowers, that are
estimate online lending could claim as much as contributing to approval rates well above those
20% of the small business loan market in the at banks and credit unions (Figure 2).
next five years.
Figure 1 Business lending at US banks, 1995-2015
Total small bus. loans, $ mill Small bus. loans / Total bus. loans
800,000
34%
700,000
30%
600,000
30%
500,000 28%
400,000
20%
300,000
200,000
100,000
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
Source: FDIC
The Alternative Lending Landscape

Marketplace players have the advantage of While these new approaches are driving valuable
being unencumbered by legacy systems and old innovation in the industry, they are yet unproven
technology. They are scaling up quickly to tap in challenging credit conditions. More than a
into the unmet demand, and they are doing so few industry observers have expressed fear that
profitably. While this may start to raise alarm the rapid proliferation of online lenders, the
bells among community banks and large banks application of new lending standards based on
whose mainstay has been lending to the small non-traditional data, and the origination-for-sale
business segment, alternative lenders will face dynamics of this market are akin to the causes of
some headwinds that make it an opportune the mortgage crisis. Although many alternative
time for traditional lenders to shore up their lenders are basing their decisions on strong
capabilities to remain competitive in the small credit management techniques, the next credit
business space. cycle is likely to expose gaps in credit modeling
and lending expertise among some of the online
First, alternative lender portfolios have scaled up
players.
over the past five years and remain unchallenged
3 Rudegeair, Peter & James Sterngold, Online Lenders Deluge Small Business, The Wall Street Journal, October 1, 2015.

Second, while regulators have given alternative between half and three-quarters of approved
lenders some freedom for the past few years applicants said they were satisfied with their
in an effort to allow innovation, the period of bank or credit union, while only 15% said the
regulatory arbitrage for non-bank lenders may same of their online lender. The biggest reasons
be ending. The online lending industry has made for dissatisfaction with online lenders were rates
concerted and visible efforts to self-regulate, and payment terms (Figure 3).
such as by creating a Small Business Borrower’s
A number of banks have recognized the
Bill of Rights and committing to practices that
opportunity to leverage the technology model
promote fairness and transparency in lending.
Figure 2 Small business loan approval rates by type of lender, February 2016
70%
61%
60%
49%
50%
42%
40%
30%
23%
20%
10%
0%
Alternative Small banks Credit unions Large banks
lenders
Source: Biz2Credit Small Business Lending Index
Despite these activities, regulators and advocacy of marketplace lenders in combination with
groups are starting to put increased scrutiny on their traditional strengths in credit and risk
alternative lenders as loan volume has increased. management to better serve small businesses.
A recent paper by the Consumer Financial Some have begun developing similar process
Protection Bureau considers expanding the scope capabilities or partnering with online providers
of consumer protection to small business loans,4 for prospecting, onboarding, and information-
and the Treasury Department closed a comment gathering, but these activities are still in their
period on marketplace lending in September infancy. Traditional lenders that act now to
2015. More regulation is likely to follow, streamline and improve their small business
particularly in the areas of disclosure, predatory lending practices will be best equipped to
lending, and sales tactics. address the funding needs of their small
Third, while small business borrowers clearly business customers and remain competitive and
respond to the ease of getting credit quickly from profitable as new players emerge.
marketplace lenders, banks and credit unions Automation and Integration

far outperform online lenders when it comes to For the tier of small business borrowers above
satisfaction after a loan is approved. According branch-based microbusinesses, credit decisioning
to the 2015 Small Business Credit Survey, and monitoring in many banks are characterized
4 Outline of Proposals Under Consideration and Alternatives Considered, Consumer Financial Protection Bureau, March 2015.

by highly manual processes. Such manual integrated. In smaller organizations, these

processes are better suited to middle-market ”systems” sometimes include spreadsheets
and larger corporate lending, where exposures and word processing programs for analysis and
(and revenue per loan) are much larger and credit write-ups. In larger ones, elaborate legacy
volumes are lower. systems are often so deeply entrenched and
Banks often collect customer and third-party entangled that it can be nearly impossible to
information on paper or through static files adopt new technologies that could dramatically
that require data to be manually keyed into streamline workflows. Together, these
approaches contribute to labor-intensive
Figure 3 Lender satisfaction rates, and reasons for dissatisfaction
Lender satisfaction score by lender type

Small bank 75%
Credit union 56%
Large bank 51%
Online lender 15%
Other 33%
32% Online lender N=67

Lack of
transparency 22% Small bank N=119
33%
Large bank N=217
22%
Long wait for
credit decision 43%
45%
Difficult 21%
application 52%
process 51%
Unfavorable 51%
repayment 15%
terms 16%
70%
High interest
rate 15%
18%
Source: 2015 Small Business Credit Survey
bank systems. Compounding the problem, and time-consuming processing, incomplete

data often has to be re-keyed multiple times process tracking, and increased probability of
because systems used in different parts of the errors. These challenges extend to small business
origination and back-office processes are not borrowers as well. While a consumer can secure

a five- or six-figure car loan on the spot at a to prevent them from becoming obsolete and
dealership, a small business borrower seeking weighing down lenders as new technologies
the same amount of funding will have to go continue to emerge.
through time-consuming document-gathering;
Credit Scoring and Decisioning
have back-and-forth communications with their
Lenders of all sizes report a lack of good credit
lender in person and by phone, email, and fax;
decisioning tools for small business lending
and endure weeks of waiting time while their
above the retail level. The market for external
application makes its way through the black box
models is highly fragmented, and many rely
of the bank’s process. In contrast, the always-
primarily or entirely on behavioral data that is
available, fully automated, and fast-response
backward-looking, heavily weighted toward the
customer experience of online lenders seems
financial position of the proprietor rather than
appealing, even at potentially higher interest
of the business, or focused on the business’s
rates and less attractive payment terms.
propensity rather than capacity to repay. Banks
Many banks have focused on ways to increase are left to piece together different models’
lending efficiency with origination platforms and outputs to make decisions, or rely on judgmental
system integrations to reduce rework. According analysis for decisioning and approval, resulting in
to a survey by the American Banker, nearly 30% inefficiencies and inconsistent outcomes.
The always-available, fully automated, and fast-response customer

experience of online lenders seems appealing, even at potentially higher
interest rates and less attractive payment terms.
of bank CIOs plan to increase their technology The emergence of marketplace lenders as
spending on lending platforms in 2016.5 So far, significant competitors makes this a more
that spending has focused mostly on banks’ urgent problem. Some are starting to do auto-
internal processes, and often it is invested in decisioning on loans to small businesses and
rebuilding or replacing in-house proprietary moving up the size scale quickly, granting and
systems with similarly customized internal funding six-figure loans in days.
platforms that will face obsolescence in only a
Banks can address this challenge by
few years. Meanwhile, banks often continue to
standardizing their credit analyses of small
overlook the customer-facing systems and tools
businesses through adoption of automated
on which marketplace lenders win.
spreading and quantitative scoring models for
There is an opportunity to do a lot more. larger loan sizes. These models should heavily
Streamlining banks’ internal processes and weight the fundamental financial information
offering customer tools that match the for the business that has proven to be most
experience provided by online lenders will predictive of future default, while incorporating
require active divestment in legacy systems, basic behavioral elements that are indicative of
not just the introduction of new tools overlaid small business credit risk.
on dysfunctional or inefficient processes. To
Our research has found that complementing
avoid legacy problems, banks must design new,
financial data with limited, relatively easy-
modular systems and processes that leverage
to-provide information on the prospective
cloud technology, APIs, and web tools, and
borrower’s relationship with the bank, available
require less customization and implementation
5 Terris, Harry, Banks to Spend More on Tech in 2016 – Especially Security, American Banker, October 15, 2015.

credit line utilization, and past delinquency their customers do business can become more
substantially increases accuracy while preserving valuable partners to their customers and create
the efficiency of model-based decisioning. new lending opportunities in this segment.
Incorporating these elements leads to higher
Raising the Game
predictive power among small businesses than
Strong demand for small business credit
existing financial-only or behavioral-only models
is expected to continue unabated. Online
based on 30 years of historical data. Scoring
and marketplace lenders will continue to
tools that utilize this information can also be
disrupt the small business lending industry by
used more proactively for early warning and loan
changing customer expectations with high-
monitoring, helping lenders prioritize credits
speed, technology-facilitated loan decisions,
warranting more scrutiny based on changes in
more accessible credit information, and vastly
scores.
improved customer experiences. At the same
Borrower Education and Enablement time, alternative sources of funding will continue
The 2015 Small Business Credit Survey shows to expand the population of the “bankable”
that one-sixth of employer firms that didn’t beyond businesses and individuals with
apply for credit were discouraged either because traditional scores, making new types of data
they felt they would not qualify or because and new approaches to modeling increasingly
Complementing financial data with limited, relatively easy-to-provide

information about the prospective borrower substantially increases
accuracy while preserving the efficiency of model-based decisioning.
they thought the process would be too arduous important. Across the industry, the rapid
to justify the time commitment. At the same evolution of technology and new sources of
time, banks continue to be the primary and data will shorten the time to obsolescence of in-
most trusted sources of information for small house, customized, non-modular platforms and
business borrowers – 73% of applicants asked of existing decisioning models.
their bankers for financing advice according to
It is critical for banks to modernize the small
the Fed’s research.
business lending process now to remain
Banks have an opportunity to solidify their competitive. This means divesting of heavy,
position as trusted advisors to their business obsolete, customized systems and adopting
customers and prospects by providing education modular cloud-based technologies for rapid
and tools that help the borrower understand deployment and agility. It means automating
their credit standing before they apply for a processes that are typically done manually
loan. Resources that provide a credit score and leveraging workflow solutions that speed
and simple, accessible information to help the the process. It means buying into automated
user understand what is driving it would go scoring solutions and innovative use of data to
a long way toward addressing this untapped inform rapid and consistent decisioning, while
market. Consumer tools abound in this area, maintaining a position of strength in traditional
but there are few that equip businesses to credit and risk management. And most of all, it
better understand and manage their credit means investing in the customer experience to
positions. Banks that provide such tools in ways make it easy, fast, transparent, and adapted to
that are adapted to when, where, and how the way small businesses operate.

INNOVATION ZONE
ARE DEPOSITS SAFE UNDER NEGATIVE
INTEREST RATES?
By Dr. Anthony Hughes
In this article, I take a theoretical look at negative interest rates

Dr. Anthony Hughes
Managing Director of Credit as a means to stimulate the economy. I identify key factors that
Analytics may influence the volume of deposits held in the economy. I then
empirically describe the unique situation of negative interest rates.
Tony oversees the Moody’s Analytics credit analysis I delve into the asymmetries that exist in the relationship between
consulting projects for global lending institutions. An deposits and interest rates, considering Sweden as an example case.
expert applied econometrician, he has helped develop
approaches to stress testing and loss forecasting in
retail, C&I, and CRE portfolios and recently introduced As the global economy sags, with many regions has been commensurately squeezed. One would
a methodology for stress testing a bank’s deposit book. in or teetering on the brink of recession, central expect this process to be further enhanced as
He received his PhD in Econometrics from Monash
banks are looking for new ways to boost monetary authorities push deeper into negative
University in Melbourne, Australia.
economic activity and stave off deflation. The territory in a bid to boost their moribund
first approach attempted in the wake of the economies.
global financial crisis was quantitative easing,
The liabilities side provides an entirely different
which arguably enjoyed some modest success
set of challenges. Banks want to maintain a
in the US but which was viewed (fairly or
consistent deposit base so they do not need to
unfairly) as a failure in various other important
tap other more expensive sources of capital to
jurisdictions around the world.
fund their activities. They also want to carefully
Banks want to maintain a consistent deposit base so they can avoid using
other, more expensive sources of capital to fund their activities.
More recently, a few countries have tried a new control the interest and noninterest costs of
tack that involves charging banks for depositing retaining these depositors, especially given the
reserve funds in the vaults of the central bank. tight profit margins imposed from the asset
Such policies have been implemented in side of the ledger. Bear in mind that many large
Japan, which has been a pioneer in the use of banks are restricted from increasing the riskiness
radical monetary stimulation techniques; the of their asset holdings due to the firm grip
eurozone; Switzerland; Denmark; and Sweden. being applied by regulators in every advanced
The possibility that the US could chart a similar economy around the world.
course cannot be easily dismissed.
Serious questions emerge regarding the effect
On the asset side of the balance sheet, such of negative central bank deposit rates on the
a move is likely to have symmetric effects on volume of deposits actually held. If, under
commercial bank activity. With zero or razor-thin negative rates, households and businesses
deposit rates, interest revenue earned by banks continue to demand bank services to store their

INNOVATION ZONE
accumulated wealth in a risk-free manner, banks In more pointed analysis, Poi, Malone, Hughes,
will be able to divert their attention to more and Zandi3 considered the effect of quantitative
profitable lending activities. Some, though, easing policies (and their subsequent reversal) on
fear that negative rates will cause an exodus of the deposit base held by US and Japanese banks.
depositors, forcing banks to raise capital from Using a variety of champion and challenger
other places or to curtail their money-making models for both jurisdictions, they found only
operations. This paper seeks to cast some marginal overall effects of the radical policies on
light on these issues from a macroeconomic the size of the deposit pie. The implication is that
perspective. if central banks want to “push on the string” in a
very low interest rate environment, or to remove
Building on Previous Research
said impetus from said ligature, it will have little
Moody’s Analytics has previously engaged
overall effect on bank holdings of deposits.
in numerous studies of the effect of stressed
economies on the aggregate deposit base. For The analysis in this paper should be viewed as an
example, Hughes looked at developing CCAR-
1 extension of earlier work.
style stress testing models for total US deposits We will begin by taking a theoretical look at the
in various categories. While the mix of deposits problem of negative rates and try to identify
held in a range of products is acutely affected key factors that may influence the volume of
by macroeconomic effects, the overall level of deposits held in the economy. This discussion
funds held by banks tends to be only marginally will then be used to guide an empirical study
impacted by generic macro stress. Recessions do that will seek to tease out the asymmetries that
cause growth to slow, but typically with quite exist in the relationship between deposits and
a long lag. Lower interest rates, holding all else interest rates. We will follow a similar approach
equal, tend to push clients away from CDs and to that used by Poi et al. and seek to identify
term deposits in favor of more convenient forms the most useful test case for empirical analysis.
of on-demand services. “Deposit recessions” Specifically, here we will consider the case of
– situations where overall volumes actually Sweden, mainly because the Riksbank was the
shrink – are very difficult to generate even under first to employ negative rates as a key plank in
extreme macroeconomic duress. its monetary policy back in 2009. Sweden is
“Deposit recessions,” in which overall volumes actually shrink, are very rare
even under extreme macroeconomic duress.
In a similar vein, Hughes and Poi2 extended unique in that there were two distinct instances
this analysis to an individual institution with of negative deposit rates with a brief intervening
a long history of high-quality deposit data period of positive rates. Poi et al. used Japan for
(North Carolina State Employees Credit Union). the same reasons in their consideration of the
They found that account funds held by such impact of QE on overall deposit behavior.
an institution are likely to also grow in the
Theoretical Analysis
midst of a recession, though pricing of services
In their economic activities, businesses and
relative to the market plays an important role
households generate stores of wealth. Part of
in determining the size of the slice eventually
this is then reinvested in risky ventures (like
retained by the institution in question.
stocks and property) to generate additional
1 Hughes, Anthony, Deposit Stress Testing, Moody’s Analytics white paper, June 2013.
2 Hughes & Poi, Improved Deposit Modeling: Using the Moody’s Analytics Pre-Provision Net Revenue Factors Library to Augment
Internal Data, Moody’s Analytics white paper, July 2015.
3 Hughes, Malone, Poi, & Zandi, Quantitative Easing and Bank Deposits, Moody’s Analytics white paper, October 2015.

future income streams. The dictates of a will retrench their holdings of these assets and
balanced portfolio and risk hedging then seek to store their accumulated wealth in the
demand that part of the store of wealth be held form of cash. Pieces of paper depicting dead
in relatively riskless investment forms such presidents always earn precisely zero percent
as cash, government bonds, and insured bank annual interest regardless of any actions pursued
deposits. by the central bank.
In a negative interest rate environment, it is safe This simple analysis assumes that holding cash is
to assume that the real economy is performing as riskless as holding insured deposits.
poorly. This implies both that the overall wealth
Some media speculation has picked up on these
generation engine is sputtering and that the
theoretical musings, often with a comedic bent.
range of available attractive, risky investment
The authors will opine about the likelihood of
options is limited. The upshot of this, holding
companies and/or households withdrawing all
all else equal, is that investors will tend, at the
their funds and putting $100 bills in tin cans to
margin, to retrench their risky investments but
be buried in the garden. Such activities carry a
retain a high demand for the risk-free options.
variety of risks that belie the notion that cash is
Though the overall wealth pie may be shrinking
a riskless way to store wealth. Mattresses can
or stagnant, the riskless slice will tend to grow
burn, treasure maps can be stolen, and buried
in scale throughout the period of real economic
notes can suffer water damage or be forgotten.
misfortune.
On a more practical level, for households in
If interest rates on short-term government the modern age, paychecks are electronically
bonds and bank deposits are zero, in theory, deposited and bills are paid automatically by
people will be indifferent between holding cash, remote computers. It is hard to imagine anyone
bonds, or deposits. If interest rates on bonds and with a measure of accumulated savings ever
deposits are negative, again in theory, people truly going off the banking grid.
Figure 1 Effects of interest rates on various deposit types
Demand Sm Term/Savings Lg Term
0.8
0.4
-3 -2 -1 1 2 3 4 5
-0.4
-0.8

INNOVATION ZONE
Figure 2 Regression results: Demand deposits
Dependent Variable: (XM1MINUSM0-XM1MINUSM0(-12))/XM1MINUSM0(-12)

Method: Least Squares
Date: 03/18/16 | Time: 11:31
Sample (adjusted): 1999M02 2015M11
Included observations: 202 after adjustments
HAC standard errors & covariance (Bartlett kernel, Newey-West fixed bandwidth = 5.0000)
Variable Coefficient Std. Error t-Statistic Prob.

C -0.055756 0.072641 -0.767563 0.4437
XIRDEP 0.011356 0.008902 1.275604 0.2037
XIRDEP*DUMNEGRATES -0.110944 0.019911 -5.571941 0.0000
DUMNEGRATES -0.297946 0.196629 -1.515266 0.1314
XIR2Y-XIRDEP 0.050152 0.015872 3.159874 0.0018
(XRCONS(-1)-XRCONS(-13))/XRCONS(-13) 0.353362 0.224713 1.572499 0.1175
(XRGFC(-1)-XRGFC(-13))/XRGFC(-13) -0.020302 0.071561 -0.283701 0.7769
XIR10Y -0.022471 0.011718 -1.917549 0.0567
XFOREXEURO 0.014548 0.008855 1.642853 0.1021
XFOREXEURO*DUMNEGRATES 0.026587 0.019757 1.345717 0.1800
XRNETEX(-1)-XRNETEX(-13) -1.29E-06 1.08E-06 -1.191750 0.2348
XIRDEP-XIREURODEP 0.001477 0.010754 0.137381 0.8909
Figure 3 Regression results: Small time and savings deposits
Dependent Variable: (XM2MINUSM1-XM2MINUSM1(-12))/XM2MINUSM1(-12)

Date: 03/14/16 Time: 14:25
C -0.721739 0.208770 -3.457094 0.0007
XIRDEP 0.103110 0.028416 3.628622 0.0004
XIRDEP*DUMNEGRATES 0.248966 0.040490 6.148816 0.0000
DUMNEGRATES -0.074632 0.396999 -0.187991 0.8511
XIR2Y-XIRDEP 0.035101 0.042133 0.833100 0.4058
(XRCONS(-1)-XRCONS(-13))/XRCONS(-13) -1.104709 0.650064 -1.699386 0.0909
(XRGFC(-1)-XRGFC(-13))/XRGFC(-13) 0.263482 0.211152 1.247826 0.2136
XIR10Y -0.098638 0.032533 -3.031977 0.0028
XFOREXEURO 0.107095 0.023207 4.614723 0.0000
XIRDEP-XIREURODEP -0.052811 0.025884 -2.040287 0.0427

Figure 4 Regression results: Large time deposits
Dependent Variable: (XM3MINUSM2-XM3MINUSM2(-12))/XM3MINUSM2 (-12)

Date: 03/18/16 Time: 11:34
C -1.683298 0.551146 -3.054179 0.0026
XIRDEP 0.210180 0.051769 4.059956 0.0001
XIRDEP*DUMNEGRATES -0.133712 0.156853 -0.852469 0.3950
DUMNEGRATES -3.277907 1.299707 -2.522036 0.0125
XIR2Y-XIRDEP 0.430368 0.109546 3.928642 0.0001
(XRCONS(-1)-XRCONS(-13))/XRCONS(-13) -0.392407 2.576636 -0.152294 0.8791
(XRGFC(-1)-XRGFC(-13))/XRGFC(-13) 1.862032 0.567306 3.282233 0.0012
XIR10Y -0.222485 0.068092 -3.267405 0.0013
XFOREXEURO 0.178921 0.064444 2.776365 0.0060
XIRDEP-XIREURODEP 0.119530 0.063208 1.891063 0.0601
For businesses and companies, the operational due to a severe recession, the overnight deposit
meshing with banks is even more extreme than it rate in Sweden first entered negative territory in
is for households. the summer of 2009. As GDP growth bounced,
Moreover, imagine for a moment that Apple, the rate was then lifted in late 2010, though this
which reportedly holds more than $200 billion move was highly controversial at the time due
in cash reserves, decided to avoid paying the to the severe recession that was continuing to
negative interest rate deposit charges imposed rage across much of Europe. The doubters were
by their bankers and to instead bury 200 tons then proved right as economic clouds once more
of $100 banknotes in the hills of Cupertino. engulfed Sweden. Renewed recession led to a
Now imagine the meeting taking place between resumption of the negative rate policy in the
Apple’s CFO and the company’s army of external summer of 2014. This situation remains in place
auditors. today.
It’s just never going to happen. The set of circumstances endured by the Swedes
sets up an ideal test case for an assessment of
Given that a range of unobserved risks and
negative rates. The most important feature is
operational rigidities associated with holding
that we observe two substantial, distinct periods
cash remains, the question of the impact of
during which the external policy treatment was
negative rates on deposit holdings is squarely
applied, as well as two separate baseline periods
empirical in nature. We will consider other
during which more normal operations were
theoretical musings – notably the effect of the
undertaken. Moreover, the controversy regarding
carry trade – in our empirical discussion of the
the initial removal of the policy in 2010 implies
Swedish economy.
that the action can be viewed as exogenously
Empirical Findings undertaken by monetary authorities. Data
For a variety of technical and policy reasons, and sourced from Statistics Sweden is of high

INNOVATION ZONE
quality, with a long history allowing precise battle with deflation during the early 2000s. In
modeling to proceed. Regression results used the current environment, where rates are low
for the following analysis are included in the everywhere, we should expect the carry trade
accompanying figures. to be far less prominent. Nevertheless, we are
We identify various deposit categories by happy to sacrifice a couple of degrees of freedom
decomposing statistics for monetary aggregates. to control for its potential impact in the work
Sweden publishes M0 through M3 – four presented here.
separate categories that progressively aggregate The final set of controls focuses on the real
longer-duration forms of bank deposits. The M0 domestic economy. In a nutshell, we give
category is very narrowly defined, incorporating Swedish households and businesses the option of
highly liquid forms of central bank deposits, consuming their deposits or of reinvesting their
banknotes, and coins. M1 adds in deposits savings in more risky forms of investment. These
available on demand; M2 adds small time macroeconomic factors enter our models with
deposits and other forms of savings accounts; a lag to stave off any accusations of potential
and M3 adds large time deposits typically owned endogeneity.
by corporations and wealthy individuals. We
We transform all control variables to ensure an
therefore interpret M1-M0 as “demand deposits,”
absence of unit roots.
M2-M1 as “small time/savings deposits,” and
Our primary focus here is in assessing the
M3-M2 as “large time deposits.”
symmetry of the observed relationships
We augment this data using information
between short-term policy rates and various
regarding interest rates at a variety of terms.
deposit categories. We create a dummy variable
We seek to control for longer end yield curve
that is generally zero but switches to unity if
dynamics that will allow us to focus our
negative deposit rates prevail within the Swedish
attention on the specific effects of the overnight
monetary system at the time. The third variable
deposit rate. Importantly, we also consider
of key interest is an interaction between the
external channels through which the Swedish
deposit rate and the dummy.
economy may interact with other countries
If the marginal effect of a change in rates is
and regions. To do this, we include net exports
the same on either side of the zero frontier,
of goods and services, the exchange rate of the
the parameter on the interaction term in our
krona with the euro, and a variety of prevailing
regression will be precisely zero. Statistically,
European interest rates.
Negative interest rates seemingly cause a slight – but statistically

significant – reduction in total deposits held.
This set of variables allows us to consider the therefore, we can test the hypothesis that this
effect of the carry trade on the behavior of phenomenon prevails in the data by using a
domestic deposits. If safe returns at home are t-test on the estimated coefficient in the model.
elusive, one option available to depositors Similarly, the inclusion of the dummy variable
involves investing their funds in a foreign allows us to consider the specific marginal effect
currency-denominated account in which involved with crossing the zero frontier. If the act
positive interest will be paid. Such investments of moving from positive to negative rates causes
carry exchange rate risk, but these potential a change in deposit behavior we will observe
misfortunes are sometimes adequately a level shift in the deposit growth rate as this
compensated by the available interest rate unfolds. The included figures display the key
differential. The carry trade is generally regression results.
considered an important dynamic in Japan’s

Across the range of positive deposit rates, the This is the most interesting finding of the
behavior of the key marginal effects is in line empirical research. Though it is true that
with our prior expectations. All three deposit negative rates sap the performance of term
categories are sensitive to rate shifts such that deposits, the funds repatriated from this process
rate increases tend to accelerate the growth rate do not do so in the form of cash. Rather, they
of underlying deposits. Of the three categories, exit in the form of demand deposits.
again as expected, demand deposits are the least
For banks, there are two implications of these
sensitive to rate changes across the positive part
findings. One is that crossing the frontier
of the number line.
seemingly causes a slight – but statistically
As the zero frontier is hit, large time and demand significant – reduction in total deposits held. For
deposits both shift lower, though the latter Sweden, this initial effect amounted to around
effect is not statistically significant at the 5% 4% of the total deposit base. As further rate cuts
level. Small time/savings deposits, meanwhile, are effected, though, part of this reduction can
do not suffer a noticeable level shift. be clawed back in the form of perhaps higher
For all time deposits, the intuitive effect of than anticipated growth in demand deposits.
an increase in rates continues to hold on the The second implication is that the duration of
negative side of the zero rate boundary. For the deposit book will tend to decline as funds
large time deposits, abstracting from the are moved out of term deposits and CDs into
presence of the level shift, the effect of rates vehicles that allow funds to be drawn instantly.
is symmetric in the sense that marginal rate Conclusion

changes have the same deposit growth impact Radical central bank policy changes are rarely
on either side of zero. Going from, say, -1% to welcomed by bankers. With the global economy
-2% will have the same growth implications for still trying to shake off the lingering effects of
large time deposits as moving from 2% to 1%. the Great Recession more than seven years
For small time deposits, meanwhile, the rates post-Lehman Brothers, it is little wonder that
effect is more pronounced on the negative side, monetary authorities are seeking new directions
implying that small time depositors become in their stimulative efforts. Recently, this push
hypersensitive to rate cuts that make such has moved in the direction of punishing savers
holdings commensurately more expensive. for holding risk-free investment forms. Such an
Things get really interesting when we consider unprecedented move has banks worried for the
demand deposits. After noting the slight safety of their deposit holdings.
downshift in such instruments when the zero In a series of papers, Moody’s Analytics has
mark is crossed, we can further observe demand explored the empirical effects of radical policy
deposit volumes tending to increase with further shifts on deposits. In general, it is observed that
short-term rate cuts. Because demand deposits such moves often have a significant impact on
represent around 80% of all Swedish holdings, the form of deposits but not an especially large
it is this effect that is the most economically effect on their scale. The findings of this paper
significant of the findings across the three are fully consistent with these observations.
separate categories. Though negative rates will cause term holdings
As rates fall further below zero, total deposits to shrink, this effect will be more than offset by a
held by banks actually increase! rise in demand deposits.

Stress Testing.
Evolved.
The Moody’s Analytics Stress Testing Suite helps firms
implement collaborative, auditable, repeatable, and
transparent stress testing processes – helping achieve
program congruence and improved efficiency.
Learn how to better integrate your stress testing program

activities at MoodysAnalytics.com/StressTesting
73
SCENARIO PROBLEM: A 2016 CCAR CASE
STUDY
By Dr. Samuel W. Malone
This article introduces Credit Risk Cascades, a new model

Director, Specialized Modeling that forecasts probability of default of financial institutions
Group under compound scenarios. The model seamlessly integrates
macroeconomic, counterparty, and systemic risk projections. This
Dr. Malone develops novel risk and forecasting article illustrates the use of the Credit Risk Cascades tool via a case
solutions for financial institutions while providing study for US financial firms under the CCAR 2016 baseline and
thought leadership on related trends in global financial
markets. He also frequently leads consulting projects
severely adverse macroeconomic scenarios.
taken on by the Specialized Modeling Group. He is
coauthor of the highly-cited book Macrofinancial
Introduction streamlined quantitative tool to facilitate stress
Risk Analysis, which offered the first operational,
validated systemic risk framework in the midst of Thinking in terms of compound scenarios, testing under compound scenarios.
the global financial crisis, as well as numerous peer- which are comprised of macroeconomic,
In this article, I introduce a new model – Credit
reviewed academic articles in applied and financial counterparty, and systemic risk components,
Risk Cascades (CRC) – that forecasts PDs of
econometrics. He has BS degrees in Mathematics and represents the natural next step of stress testing.
Economics from Duke University, where he graduated financial institutions under compound scenarios
For eight bank holding companies, the Fed
summa cum laude, and a doctorate in economics from involving economic, counterparty, and systemic
the University of Oxford. now requires layering a counterparty default
risk components. The CRC model incorporates
scenario onto standard CCAR exercises, and
these three main effects:
has recently suggested that next-generation
»» Direct dependence of PD forecasts on the
stress tests could feature shocks to bank
economic scenario
interconnectedness. In fact, as reported in
American Banker, Fed governor Daniel K. Tarullo »» Credit risk spillovers via network linkages
has recently gone on record stating that he “… »» A user-specified path for financial sector
envisions the stress-testing process moving interconnectedness
beyond just examining the performance of
individual banks' capital levels under stress and This is accomplished with the following
also include the interconnection of institutions techniques: 2
under stress as well.” When forecasting
1
»» Using Moody’s Stressed EDF to obtain EDF
probability of default (PD) for such purposes, paths under a user-selected economic
banks require a model that seamlessly integrates scenario
macroeconomic, counterparty, and systemic
»» Using a dynamic network model to obtain
risk projections. In fact, some large banks that
PD forecasts under the counterparty shock
conduct sophisticated “war game” exercises
relative to an appropriate no-shock baseline
as part of their risk management activities
already think in these terms, but lack a model »» Adjusting the Stressed EDFs to take into
that allows a fully coherent evaluation of such account the network model results
shocks together. The banking industry needs a
1 Heltman, 2015.
2 The full documentation and validation of Credit Risk Cascades can be found in Malone (2015).

INNOVATION ZONE
Here, I illustrate the use of the Credit Risk and Federal Deposit Insurance Corporation
Cascades tool via a case study for US financial (FDIC) recently said that the resolution plans
firms under the CCAR 2016 baseline and severely submitted in 2015 by five of the largest US banks
adverse macroeconomic scenarios. EDF forecasts failed to pass muster and must be rewritten by
under these scenarios, obtained from Stressed the banks prior to October 1, 2016.4 Among the
EDF, are modified in response to the default of a faults found in the living wills, regulators cited
small set of counterparties in a way that allows the failure to fully address the material financial
credit risk shocks to propagate more strongly interconnections between banks and broker-
when bank interconnectedness is high. After dealers that would be relevant when winding
discussing other regulatory applications of CRC down trading portfolios. Network methodologies
and providing a brief methodology overview, the such as the one behind CRC are fit-for-purpose
case study illustrates the effects that different for quantifying the strength and direction of such
counterparty and systemic risk events could interconnections under economic stress.
have for a select group of major US banks under
Methodology Overview
alternative paths for the macroeconomy.
The CRC model builds on Systemic Risk Monitor5
Regulatory Context: Other Applications of CRC for network model estimates, and on Stressed
While the counterparty default scenario that EDF6 for projections of EDFs conditioned only on
forms the focus of this article is a key regulatory the macroeconomic scenario. CRC combines the
application of CRC, it is not the only one. information in the network model and Stressed
Two other regulatory applications include EDF to arrive at the CRC-EDF forecast paths for
single counterparty credit limits (SCCL) and financial institution PDs under the compound
compliance issues related to the “living will” scenario, which includes the macroeconomic
portion of Dodd-Frank. scenario as well as the systemic risk path and
counterparty shocks specified by the user. All of
Related to the first point, the Fed has recently
the aforementioned methodologies employ the
proposed an update to the rules governing
most recent iteration of CreditEdge, the EDF9
SCCLs for US bank holding companies and
measure, as the core input. The construction of
foreign banking organizations with at least
this newest iteration of the EDF metric of default
$50 billion in total consolidated assets. As
probability and differences with the previous
part of the proposed rules, whenever a bank
version are documented in detail in Chen et al.
holding company’s net credit exposure to an
(2015).
unaffiliated counterparty exceeds 5% of the
bank’s eligible capital, the bank must determine The financial network I use to drive this
whether the counterparty is economically application of CRC is the Global Megabanks
interdependent with any of the bank’s other network, which consists of all publicly traded
unaffiliated counterparties. The analysis
3
financial institutions whose book assets exceed
required to determine such a relationship turns $100 billion in value. Historical estimates of this
directly upon whether the financial distress of network’s structure and associated systemic risk
one counterparty is likely to impair the ability of measures are provided in Hughes and Malone
other bank counterparties to make good on their (2015), and were updated for this study using
liabilities to the bank. As the case study of this the most recent available CreditEdge data.
article illustrates, CRC quantifies precisely the The Global Megabanks network is the natural
extent to which the distress of one counterparty peer group of global CCAR-sized institutions
will transfer to another counterparty of interest in general and the institutions subject to the
under a variety of economic scenarios. counterparty default scenario in particular.
Regarding the living wills required of major US Systemic Risk Monitor estimates the extent and
banks under Dodd-Frank, the Federal Reserve strength of Granger causal connections between
3 An unaffiliated counterparty is generally understood to be one that the bank does not control, is not controlled by the bank, and
is not under common control with the bank that is the subject of SCCL regulation.
4 Hamilton & Dexheimer, 2016.
5 This is documented in Hughes & Malone (2015).
6 This is documented in Ferry, Hughes, & Ding (2012).

Figure 1 Interconnectedness measure under network stasis and network stress
DGC.plus under network stress DGC.plus under network stasis
0.16
0.14
0.12
0.10
0.08
0.06
0.04
0.02
0.00
1 3 5 7 9 11 13 15 17 19 21 23 25 27
Forecast Horizon
the EDF series of all distinct pairs of financial In CRC, counterparty shocks take the form of
institutions in the network, and this information perfect-foresight PD paths. In other words, the
is passed to CRC for forecasting. user specifies the time series that a given bank’s
PD will follow over the course of the forecast
Case Study
horizon, and the CRC algorithm solves for the
The base date for our compound counterparty
PDs of other banks endogenously using model
default scenario is February 2016. For
averaging to construct forecasts. In this case
illustrative purposes, I consider shocks to two
study, I assume that from month 0 (February
counterparties: Fifth Third Bancorp and US
2016) until month 27 (May 2018) of the forecast
Bancorp. Both of these firms are subject to
horizon, the PDs of both Fifth Third Bancorp
CCAR stress tests but are not required to do
and US Bancorp follow the path taken by the
the counterparty default scenario. I adopt a
EDF of Lehman Brothers during the 28 -month
27-month time horizon and forecast default
period from January 2008 to April 2010. During
probabilities for each month during that
this period, Lehman Brothers’ EDF reached the
period for two firms that are required to do
maximum value for financial firms of 35% for the
counterparty default scenarios under CCAR:
first time in September 2008, or month 8, and
Bank of New York Mellon and JP Morgan Chase.
remained there for the rest of the time horizon.
As stated in the previous section, I measure
Compound scenarios are closed by specifying
PDs using the Moody’s CreditEdge EDF metric,
macroeconomic and systemic risk components.
which provides an estimate of the 1-year ahead
The macro scenario choice set for our purposes
probability of default. CRC and Stressed EDF
consists of the 2016 Fed baseline scenario and
both forecast EDFs under scenarios, and I
the 2016 Fed severely adverse scenario. For
will refer to these forecasts as CRC-EDFs and
the systemic risk scenario choice set, I consider
SEDFs, respectively. CRC-EDFs are compound
two alternatives: network stasis and network
scenario forecasts, whereas SEDFs are forecasts
stress. Under network stasis, it is assumed that
conditional on only the macroeconomic
the DGC.plus interconnectedness measure for
component of the scenario. For the same
the Global Megabanks network will remain at
underlying Fed macro scenario, their difference
its February 2016 value over the scenario time
can be attributed to counterparty shocks
horizon. Under network stress, in contrast, the
and the effect of changes in financial sector
DGC.plus measure is assumed to follow the path
interconnectedness.

INNOVATION ZONE
it took during the turbulent period from January more than that of JP Morgan Chase in response
2008 to April 2010, to coincide with the same to a US Bancorp default. These findings illustrate
period from which the Lehman Brothers EDF the role that Credit Risk Cascades might play in
values are sourced for the counterparty shock. counterparty selection in the context of bank
The time paths of the DGC.plus measure under stress testing.
network stasis and stress are depicted in Figure 1.
It is useful to compare Figures 2 and 4. In Figure
The results are organized into four figures of four 4, the small deviations from Stressed EDF under
plots each. Each individual plot displays the CRC- the Fed baseline scenario are magnified when we
EDF and SEDF of either Bank of New York Mellon increase the level of interconnectedness in the
or JP Morgan Chase under a specific compound financial network, and the net effect of a Fifth
scenario. Each of the four figures corresponds Third Bancorp shock on JP Morgan Chase’s EDF
to a choice of macroeconomic scenario (Fed forecast turns positive under network stress.
baseline or Fed severely adverse) and a choice This result illustrates the importance of risk
of systemic risk scenario (network stasis or transmission via the intermediate counterparties
network stress). For a given figure, each of four connecting the two banks in the network.
plots corresponds to one of the four choices of
Additionally, a comparison of the right-hand
shocked counterparty bank (Fifth Third Bancorp
columns of Figures 2 and 4, as well as of Figures
or US Bancorp) and bank whose shocked CRC-
3 and 5, shows that a heightened level of
EDFs and SEDFs are displayed (Bank of New York
interconnectedness in the financial network
Mellon or JP Morgan Chase).
would exacerbate the effect of the US Bancorp
Results counterparty shock on both banks. In general,
Results are displayed in Figures 2 through 5. the counterparty-systemic interactions impact
Figure 2 shows CRC-EDF9 and SEDF9 paths BNY Mellon and JP Morgan Chase differently
under the Fed baseline macro scenario and depending on which counterparty we shock.
network stasis. Figure 3 shows these series under These results illustrate the potential for
the Fed severely adverse macro scenario and counterparty shocks to be more damaging
Banks could incorporate private information on their own direct

counterparty exposures to further increase the realism of results.
network stasis. Figure 4 shows results for the when the strength and prevalence of credit risk
Fed baseline macro scenario and network stress, spillovers in the financial network increase, as is
and Figure 5 shows results for the Fed severely the case during periods of market stress.
adverse macro scenario and network stress. We can gain specific insight on the relative
As seen in the left column of Figure 2, a shock to contributions of the three components of the
Fifth Third Bancorp has a small negative effect compound scenario to the resulting CRC-EDF
on the CRC-EDF of BNY Mellon relative to its of a bank by decomposing the PD forecast on a
Stressed EDF. The Fifth Third Bancorp shock specific date. Let us take as an example the case
produces virtually no effect on the CRC-EDF of BNY Mellon in February 2017, one year after
forecast relative to the Stressed EDF for JP the scenario begins.
Morgan. In contrast to the case of the Fifth Third In February 2017, BNY Mellon’s CRC-EDF under
Bancorp shock, the distress and subsequent the compound scenario involving (a) the Fed
default of US Bancorp would nontrivially raise severely adverse macroeconomic scenario,
the default probabilities of both JP Morgan Chase (b) a counterparty shock to US Bancorp, and
and BNY Mellon, even under network stasis. This (c) network stress – located in the upper right
can be seen in the right column of Figure 2. corner of Figure 5 – is approximately 4.2 percent.
As is evident in the right-hand columns of all four This compares to the Stressed EDF on the same
figures, BNY Mellon’s EDF would be impacted date under the Fed baseline scenario of just

Figure 2 CRC-EDF9 and SEDF9 paths under Fed baseline and network stasis
BNY MELLON, counterparty shock: FIFTH THIRD BANCORP BNY MELLON, counterparty shock: US BANCORP
CRC SEDF9 CRC SEDF9
1.00 1.00
0.90 0.90
0.80 0.80
0.70 0.70
0.60 0.60
0.50 0.50
0.40 0.40
0.30 0.30
2016-02 2016-08 2017-02 2017-08 2018-02 2016-02 2016-08 2017-02 2017-08 2018-02
JP MORGAN CHASE, counterparty shock: FIFTH THIRD BANCORP JP MORGAN CHASE, counterparty shock: US BANCORP
CRC SEDF9 CRC SEDF9
1.00 1.00
0.90 0.90
0.80 0.80
0.70 0.70
0.60 0.60
0.50 0.50
0.40 0.40
0.30 0.30
2016-02 2016-08 2017-02 2017-08 2018-02 2016-02 2016-08 2017-02 2017-08 2018-02
under 0.5 percent, as shown for example in the remaining 0.5 percentage point increase, to bring
upper right corner of Figure 2. Let’s decompose us to the CRC-EDF of 4.2 percent under the most
this difference of 3.7 percentage points. stressful compound scenario, shown in Figure 5.
Compared to the SEDF under the Fed baseline, The relative magnitudes of these contributions
adding macroeconomic stress alone accounts for to the CRC-EDF accord with intuition about their
an increase of 2.2 percentage points in the EDF. relative importance in most situations.
This gets us to a SEDF under the Fed severely CRC does not use information on direct
adverse scenario of 0.5 + 2.2 = 2.7%, as shown counterparty exposures on banks’ books. Rather,
in the upper right-hand corner of Figure 3. it uses information from the last five years of
From there, the US Bancorp counterparty shock history to estimate the presence and strength
accounts for an additional 1 percentage point of spillovers between the PDs of all pairs of
increase in the EDF, taking us to 3.7 percent banks in the financial network. This produces
under the scenario involving the Fed severely real-time, market-based conditional forecasts of
adverse macro component, the US Bancorp financial institutions' PDs under macroeconomic,
counterparty shock, and network stasis. This systemic, and counterparty stress.
result can be seen by inspecting the CRC line in
Without a doubt, banks could incorporate
the upper right-hand corner of Figure 3. Layering
private information on their own direct
on financial network stress accounts for the

INNOVATION ZONE
Figure 3 CRC-EDF9 and SEDF9 paths under Fed severely adverse and network stasis
CRC SEDF9 CRC SEDF9
4.00 4.00
3.50 3.50
3.00 3.00
2.50 2.50
2.00 2.00
1.50 1.50
1.00 1.00
0.50 0.50
0.00 0.00
2016-02 2016-08 2017-02 2017-08 2018-02 2016-02 2016-08 2017-02 2017-08 2018-02
CRC SEDF9 CRC SEDF9
4.00 4.00
3.50 3.50
3.00 3.00
2.50 2.50
2.00 2.00
1.50 1.50
1.00 1.00
0.50 0.50
0.00 0.00
2016-02 2016-08 2017-02 2017-08 2018-02 2016-02 2016-08 2017-02 2017-08 2018-02
counterparty exposures to further increase the user to take a granular view of how the PDs
the realism of results, such as by shocking of different counterparties might evolve over the
their assets directly in the face of a projected forecast horizon as their projected default events
counterparty default. The shocked bank assets draw near.
would flow through the structural formula that
Finally, while banks know who their own
drives the bank EDF, and this direct impact
counterparties are and their exposure to them,
could be overlaid on the results provided by
they know comparatively little about their
CRC. Alternatively, banks could use their own
counterparties’ counterparties. This lack of
PD series to calibrate the model instead of using
knowledge can be most dangerous precisely
CreditEdge EDFs, so long as their PD series are
when it matters most: in times of market stress.
available with sufficient frequency (monthly)
Banks take it for granted that the default of a
and history (at least five years is recommended).
direct counterparty will affect them, but often
An important advantage of CRC is that it allows gloss over the potential for that default (or
risk managers to view counterparty default as default by a third party) to take down a second
the culmination of a process of deteriorating of their important counterparties. CRC uses
credit quality. Such deterioration manifests network models and model averaging techniques
itself in the form of persistent increases in the to create conditional forecasts of financial firm
probability of default over time, and CRC allows PDs. In a backtest using data from the financial

Figure 4 CRC-EDF9 and SEDF9 paths under Fed baseline and network stress
CRC SEDF9 CRC SEDF9
1.00 1.00
0.90 0.90
0.80 0.80
0.70 0.70
0.60 0.60
0.50 0.50
0.40 0.40
0.30 0.30
2016-02 2016-08 2017-02 2017-08 2018-02 2016-02 2016-08 2017-02 2017-08 2018-02
CRC SEDF9 CRC SEDF9
1.00 1.00
0.90 0.90
0.80 0.80
0.70 0.70
0.60 0.60
0.50 0.50
0.40 0.40
0.30 0.30
2016-02 2016-08 2017-02 2017-08 2018-02 2016-02 2016-08 2017-02 2017-08 2018-02
crisis, forecasts conditional on appropriately systemic risk component, and a counterparty

selected compound scenarios proved superior distress or default scenario for one or multiple
to PD forecasts conditional on macroeconomic counterparties of interest. CRC integrates the
stress alone (Malone, 2015). effects of these three types of shocks together
seamlessly to forecast the PDs of all firms in a
Conclusion
user-specified financial network.
This article introduces a new solution, Credit
Risk Cascades, for the Fed’s counterparty default Using data on default probabilities from
scenario requirement in the latest installment of Moody’s CreditEdge, bespoke networks can be
While banks know who their own counterparties are and their exposure
to them, they know comparatively little about their counterparties’
counterparties. This lack of knowledge can be most dangerous precisely
when it matters most: in times of market stress.
CCAR. The CRC model allows users to evaluate constructed and shocked for publicly traded
the impact on firm default probabilities of financial institutions around the world. CRC
compound scenarios. Such scenarios involve can be used to select counterparties for default
a macroeconomic scenario component, a scenarios in terms of their impact, and in this

INNOVATION ZONE
Figure 5 CRC-EDF9 and SEDF9 paths under Fed severely adverse and network stress
CRC SEDF9 CRC SEDF9
4.00 4.00
3.50 3.50
3.00 3.00
2.50 2.50
2.00 2.00
1.50 1.50
1.00 1.00
0.50 0.50
0.00 0.00
2016-02 2016-08 2017-02 2017-08 2018-02 2016-02 2016-08 2017-02 2017-08 2018-02
CRC SEDF9 CRC SEDF9
4.00 4.00
3.50 3.50
3.00 3.00
2.50 2.50
2.00 2.00
1.50 1.50
1.00 1.00
0.50 0.50
0.00 0.00
2016-02 2016-08 2017-02 2017-08 2018-02 2016-02 2016-08 2017-02 2017-08 2018-02
way can help banks construct more effective Finally, the regulatory applications of CRC are
counterparty default scenarios. The case study not limited to the counterparty default scenario
presented here, based on the network of Global alone. Single counterparty credit limit (SCCL)
Megabanks with at least $100 billion in assets, regulations and the need to better take into
shows that both appropriate counterparty account interconnectedness when devising
selection and a rise in financial sector resolution plans, or “living wills,” also provide
interconnectedness can cause material changes highly relevant applications for this network-
in projected credit quality for banks subject to based solution.
CCAR.
Chen, Nan, Houman Dehghan, Min Ding, Jian Du, Douglas Dwyer, James Edwards, Danielle Ferry, Pooya Nazeran, Zhao Sun, Jing
Zhang, and Sue Zhang, EDF9: Introduction and Overview, Moody’s Analytics white paper, June 2015.
Ferry, Danielle, Anthony Hughes, and Min Ding, Stressed EDF Credit Measures for North America, Moody’s Analytics white paper,
May 2012.
Hamilton, Jesse, and Elizabeth Dexheimer, Five Big Banks’ Living Wills are Rejected by U.S. Regulators, Bloomberg News, April 13,
2016.
Heltman, John, “How the Fed Might Change Stress Tests in Version 2.0”, The American Banker, December 28, 2015.
Hughes, Anthony and Samuel W. Malone, Systemic Risk Monitor 1.0: A Network Approach, Moody's Analytics white paper, June
2015.
Malone, Samuel W., Credit Risk Cascades: Forecasting PDs under compound scenarios, Moody's Analytics white paper, November
2015.

REGULATORY
REVIEW
READING THE TEA LEAVES OF RECENT
REGULATORY GUIDANCE
By Anna Krayn, Ed Young, and David Little
In this article, we review the common themes reflected in recent

Anna Krayn
Senior Director and Team Lead, regulatory guidelines released by the Federal Reserve and the
Capital Planning and Stress BCBS. Regulators are putting greater emphasis on: unified risk
Testing
management infrastructures, enhanced capital planning and ALLL
Anna is responsible for the business development projections, and improved decision-making with sensitivity and
of stress testing and capital planning solutions. scenario analysis. Banks will need to reevaluate and update internal
Her clients include a variety of financial services
institutions, including those in the insurance, banking,
processes in order to align with new regulations.
and consumer finance sectors across the Americas.
December 2015 was a busy month for recent regulatory guidance and proposals, there
regulatory agencies and global standard setters. are themes that will compel financial institutions
Throughout 2015, the industry had been to take another critical view of their information
waiting for additional guidance on high-impact systems. Additionally, these publications confirm
Ed Young
Senior Director, Capital topics including capital planning and allowance that scenario-driven analysis is spreading
Planning and Stress Testing methodologies, and in the final stretch of the from stress testing to business-as-usual risk
year, both the Federal Reserve and the Basel management, including allowance processes.
Committee on Banking Supervision (BCBS) While much of the recent guidance will require
Ed is a Senior Director on the Stress Testing and
complied. This paper will primarily focus on interpretation over the coming months, we
Capital Planning Team. In this capacity, he focuses on
structuring solutions that bring together capabilities
common themes in these two releases: review the common themes, summarized in
across Moody’s Analytics to support robust capital 1. The Federal Reserve’s “Guidance on Figure 1, and their interconnectedness across
planning and stress testing processes. His primary organizational silos of the finance and risk
Supervisory Assessment of Capital Planning
focus is on clients in the banking and insurance sectors functions.
across the Americas. and Positions” (SR 15-18 and SR 15-19)
2. The BCBS’s “Guidance on Credit Risk and A common thread consistently emphasized by
Accounting for Expected Credit Losses” the regulators is for financial institutions to
improve risk identification and measurement at
Recent guidance across a range of jurisdictions the enterprise level. The traditional delineation
David Little builds on previously voiced recognition that of responsibilities between chief financial officers
Managing Director, Head of the technology systems at many banks need and chief risk officers has led to a segregation
US Enterprise Risk Solutions and
Stress Testing Sales Teams
David is responsible for helping financial institutions A common thread consistently emphasized by regulators is for financial
worldwide with their enterprise risk management, institutions to improve risk identification and measurement at the
liquidity, and stress testing solutions. Since joining enterprise level.
Moody’s in 2002, David has been Managing Director
in the Product Strategy group, responsible for a global
portfolio of research, data, and analytic products
across all fixed income asset classes, led end-of-day improvement. Technology infrastructure has of duties that has greatly enhanced risk
pricing business at Moody’s Evaluations Inc., and been stifled by legacy mergers and acquisitions management practices at large firms. However,
headed the Global Structured Finance Sales team at that led to tactical system integrations, and the changes have also contributed to fragmented
Moody’s Analytics.
by patched solutions to address immediate risk reporting that in turn obfuscates the “top of
regulatory requirements. We observe that in the house” view of a firm’s risk profile.

REGULATORY REVIEW
Figure 1 Unification of Risk Management Infrastructure
Governance/
Controls
Finance
Risk
Risk
Management
Infrastructure
Benchmark Sensitivity/
Data/Models Scenario
Analysis
THEMES FOR FIRMS TO CONSIDER:

1. Focus on governance and internal controls to unify risk management infrastructure.
2. Leverage benchmark data and models to support and enhance your capital planning
and ALLL projections.
3. Consider a range of potential outcomes using sensitivity and scenario analysis to

improve decision-making.
Managing the complexity of data and models to identify and manage risk while maximizing
across business lines and developing a efficiency. In this paper, we will address three
comprehensive strategy that is aligned with key themes that are elements in achieving these
a firm’s risk appetite is a challenging task. To objectives.
maintain an effective process, management
Theme #1: Focus on governance and
must focus on organizational planning, internal controls to unify risk management
communication, and implementing robust infrastructure.
information systems. As a result, a need for When considering how to meet the spirit of the
cross-organization coordination is imperative recent regulatory standards, viewing the various

standards through the lens of an effective the process is challenged by the multitude of
governance framework can help identify handoffs required to complete risk management
connection points within the landscape of the and regulatory reports. The ad hoc nature of
organization. This first theme is a natural place these handoffs and the resulting loss of data
to begin, because governance is the glue that granularity, often through reliance on Excel-
unequivocally binds together the components of based manual processes, create operational risk.
a bank’s enterprise risk management system. Increased regulatory emphasis on the active role
Capital Planning Guidance Leading to of internal audit is creating constant pressure
Infrastructure Reassessment for firms to enhance their information systems.

The adopted changes to the FR Y-14A/Q/M
In its 2009 report for the Financial Stability
reports create additional pressure to enhance
Board, the Senior Supervisors Group stated:
information systems, albeit at a staggered
“Supervisors believe that considerable work
schedule, as follows:
remains in the areas of governance, incentives,
internal controls, and infrastructure.”1 In the 1. Beginning with FR Y-14 M/Q/A reports as of
subsequent years, supervisors have become December 31, 2016, LISCC firms will attest to
more and more vocal about their expectations internal controls.
for governance. The recent Fed guidance notes 2. Beginning with monthly, quarterly, and
that firms must have “integrated management semi-annual reports as of January 31, 2017,
information systems, effective reporting, and there will be an additional attestation to the
change control processes.” This message can accuracy of the reported data, conformance
be directly linked with the concurrent Agency with FR Y-14 instructions, and agreement to
Information Collection Proposal that stated
2
report material weaknesses and any material
“all respondents to the FR Y–14A/Q/M reports errors.
should meet the Federal Reserve’s expectations
3. Beginning December 31, 2017, LISCC firms will
for internal controls.” The proposal was recently
attest to the effectiveness of internal controls
approved3 and requires Chief Financial Officers of
around the FR Y-14A/M/Q (as a replacement
LISCC4 firms to attest to the quality of FR Y–14A/
to attestation described under #1 above).
Q/M reporting “in order to encourage large firms
to improve their systems for developing data As a result of the new requirements, many
necessary for the stress tests and CCAR.” These firms will need to reduce the number of ad
latest releases can be viewed as an effort by the hoc manual processes, replacing them with
regulators to remediate lingering issues initially automated solutions that serve as a foundation
outlined in the wake of the financial crisis. for a transparent and auditable capital planning
process, credible stress testing results, and risk
At many banks, internal audits of the stress
appetite framework quantification.
testing processes have prompted management
to take initial steps to trace data lineage for each Revised Allowance Processes Guidance Leading
FR Y-14A report line item. However, a significant to Infrastructure Reassessment
challenge for banks to overcome is the array of Concurrently, accounting standard-setters
decentralized systems (often dozens) that feed around the globe are in the process of adopting
the capital planning process. The governance of
1 Senior Supervisors Group, Risk management lessons from the global banking crisis of 2008, October 2009
2 Federal Register / Vol. 80, No. 179 / Wednesday, September 16, 2015 / Notices, Request for Comment on Information Collection
Proposal.
3 Federal Register / January 14, 2016 / Capital Assessments and Stress Testing information collection.
4 Firms supervised by Large Institution Supervision Coordinating Committee of the Federal Reserve. As of this writing, the LISCC
Portfolio includes American International Group, Inc., Bank of America Corporation, The Bank of New York Mellon Corporation,
Barclays PLC, Citigroup Inc., Credit Suisse Group AG, Deutsche Bank AG, General Electric Capital Corporation, The Goldman
Sachs Group, Inc., JP Morgan Chase & Co., MetLife, Inc., Morgan Stanley, Prudential Financial, Inc., State Street Corporation,
UBS AG, and Wells Fargo & Company.

REGULATORY REVIEW
standards that aim to address the many highlights that the ECL framework may lead to
documented failures of the incurred loss model significant investment:
by requiring forward-looking credit loss models.
“While the implementation of ECL accounting
The global IFRS 9 standard was published in July
frameworks may require an investment in
2014, and in the US, the Financial Accounting
both resources and system developments/
Standards Board’s Current Expected Credit Loss
upgrades, standard setters have given (or are
(CECL) standard is expected to be released in
expected to give) firms a considerable time
June 2016. With a targeted implementation in
period to transition to the updated accounting
2018 for IFRS 9 and essentially 2020 for CECL,
requirements. On that basis, the Committee has
we are only beginning to see the impact these
significantly heightened supervisory expectations
changes will have on processes and controls, and
that internationally active banks will have a high-
therefore information systems at banks.
quality implementation of an ECL accounting
The move to forward-looking measures to framework.”
inform the allowance will push firms to better
The implications are significant, and are
integrate allowance methodologies with stress
summarized in Figure 2.
testing processes. That said, initially, the greatest
Today, many financial institutions employ
challenge for the banking industry will be
different systems and processes across their
interpretation of the standards. In December,
accounting, capital planning, and credit risk
the BCBS published Guidance on Credit Risk and
management groups. This leads to different
Accounting for Expected Credit Losses (ECL)
Figure 2 Key considerations that will impact implementation and systems:
1. Use of economic scenarios may cause volatility of the provision expense, driving a need to run a
multitude of scenarios (as highlighted in the IFRS Transition Resource Group staff paper) more
frequently than current scenario analysis practices.5
2. Depending on interpretation, the ECL-driven allowance calculation may require more granular
data, which will in turn put pressure on processing time.
3. Capital calculation and reporting will require monthly reconciliation, putting pressure on monthly
data collection / cleansing activities and processing time.
4. Auditability of the results and transparency of the process is key, in particular since “the
Committee … expects management to apply its experienced credit judgment to consider
future scenarios … and the resulting impact on the measurement of ECL” and use of temporary
adjustments and overrides will require “appropriate documentation, and subject to appropriate
governance processes.”
5. Since there is obvious commonality in data and processes used for allowance and capital
adequacy calculations, “The Committee expects banks to leverage and integrate common
processes” to “reduce cost and potential bias and also encourage consistency in the
measurement, management and reporting.”
6. IFRS 9 stage allocation (i.e., the movement from “bucket” 1 to “bucket” 2, and to “bucket” 3) will
create significant workflow requirements.
5 Transition Resource Group for Impairment of Financial Instruments, Incorporation of forward-looking scenarios, December 11,
2015.

Figure 3 Differentiated standards for external models and data
LISCC and Large, Complex Firms Large, Noncomplex Firms
Use of External Data in Model Development Permitted in some cases: Permitted:
“A firm should use internal data to estimate losses “A firm may use either internal or external data to
and PPNR as part of its enterprise-wide stress estimate losses and PPNR as part of its enterprise-wide
testing and capital planning practices. However, it stress testing and capital planning practices.”
may be appropriate for a firm to use external data if
internal data limitations exist as a result of systems
limitations, acquisitions, or new products, or other
factors that may cause internal data to be less
relevant for developing stressed estimates.”
Challenger Model Expectations Expected: Not expected:
“A firm should use benchmark or challenger models “… A firm should use a variety of methods to assess
to assess the performance of its primary models performance of material models and gain comfort
for all material portfolios or to supplement, where with material model estimates. However, a firm is
appropriate, the primary models.” not expected to use benchmark models in its capital
planning process.”
“A firm should also use benchmark models during
validation as an additional check on the primary
model and its results.”
Use of Vendor Models Permitted with “heightened expectations for… Permitted, with expectations that “any vendor or other
controls around the use of vendor models.” third-party models [used] are sound, appropriate for the
given task, and implemented properly.”
data hierarchies and levels of granularity. Theme #2: Leverage benchmark data and
Ongoing work to interpret the IFRS 9 standard models to support and enhance capital planning
and ALLL projections.
and eventual CECL standard will drive future
The use of external data and models is common
systems requirements, but it is evident that in
practice for financial institutions of all sizes.
many instances current information systems
Reasons to employ external data or third-party
will not suffice. There will be opportunities
models include:
in some cases to leverage common data and
models across an organization. Enhancing these 1. The cost efficiency of leveraging industry-
linkages will improve governance, transparency, tested solutions versus those developed
and efficiency for firms. However, many firms in-house
will not have the technology infrastructure 2. The time savings of implementing an out-of-
to seamlessly implement efficient processes. the-box or customized third-party tool
While existing Basel and Stress Testing / Capital
3. The supplementation of internally developed
Planning infrastructures may serve as starting
solutions that may have insufficient internal
points, both will need enhancement to meet
data for modeling due to portfolio changes or
the new standards. Moving forward, governance
lack of internal historical data
expectations from regulators across the globe
will guide firms to strengthen the symbiotic Additionally, external data and models are
relationship between stress testing frameworks often used as benchmarks to meet industry
and business as usual risk management systems. model risk management standards. However,

REGULATORY REVIEW
bank supervisors’ more rigorous model risk Theme #3: Consider a range of potential
management expectations have raised the bar outcomes using sensitivity and scenario
analysis to improve decision-making.
for implementing these solutions with stricter
In the wake of the financial crisis, risk managers
“fit for use” criteria.
have been inundated with questions from
In the “Assessment of Capital Planning and regulators about how their bank gauges
Positions” guidance, the Fed aimed to clarify uncertainty and how they incorporate
how it tailors expectations for large and complex uncertainty into their pro-forma estimates.
firms versus noncomplex firms (e.g., generally However, until recently, regulators released
large regional banks with assets between $50 very little public guidance on how banks should
billion and $250 billion). Figure 3 is a summary specifically address uncertainty and include
of the differentiated standard as it relates to the “difficult to quantify” risks in their stress
use of external data and models. It highlights scenarios. The recent publications from the Fed
areas subject to interpretation. For example, and the BCBS continue to highlight uncertainty
noncomplex firms, through conversations with as a concern. Fortunately, the Fed has provided
the three regulatory agencies, will need to some details for minimum expectations on the
ascertain the interplay between Principle 2 of topic (as it pertains to Capital Planning). While
the interagency guidance on stress testing that the new guidance leaves questions as to how
requires “multiple conceptually sound stress to incorporate various difficult-to-quantify
testing activities and approaches” and SR 15-19 risks, there is a clear theme of using sensitivity
that as a minimum expectation eliminates the and scenario analysis to provide perspective
mandate for benchmark model use. on the pro-forma results. Banks are expected
There are many open questions on the use to leverage scenario analysis and sensitivity
of external data and models with respect to analysis to broadly capture uncertainty in
ECL-based allowance implementation, as their estimates due to the inherent limitations
well. BCBS guidance on ECL states that robust embedded within a single deterministic stress
allowance frameworks will generally “consider scenario.
the relevant internal and external factors, that For capital planning, the Fed also outlined
may affect ECL estimates, such as … changes in the need to address the uncertainty of model
industry, geographical, economic and political outputs through sensitivity analysis of key
factors.” Some types of models applicable assumptions. These expectations span the entire
under IFRS 9 and expected under CECL may capital planning process, and include identifying
require more granular historical data with and sensitivity testing key assumptions in
longer time series than available internally (to individual models, as well as collectively
establish relationship with macroeconomic at the aggregated level to “inform senior
variables). The introduction of forward-looking management and the board of directors about
credit loss models will inevitably increase the potential uncertainty” associated with the firm’s
volatility of allowance calculations under ECL. projections. This will require firms to ensure they
The increased complexity of the calculation, have strong assumption management processes
coupled with the volatility of a forward-looking in place, and also an established, transparent
measure will have a direct impact to earnings. and auditable process to “justify, document, and
Thus, the lack of recognition for the increased appropriately challenge” assumptions.
allowance in key capital ratios will likely drive
Scenario analysis has emerged as an effective
firms to conduct additional sensitivity analysis
forward-looking tool to manage risk. However,
around key assumptions and increase the use
the time, effort, and technology needed to
of benchmarks. At the same time, any use of
conduct a bottom-up assessment of many
external tools will be subject to both an external
different scenarios is daunting. For allowance
and internal audit assessment of “reasonable
calculations, it appears that the BCBS and the
and supportable tools.”

IASB6 are expecting banks to run a multitude and will define parameters for technology
of scenarios to ensure that ECL calculations enhancements.
are sound. Additionally, to meet the Fed’s
In 2016, banks will need to critically assess
expectations related to scenario design, banks
whether existing systems across risk and
will need to run more scenarios to “collectively
finance functions support compliance with
address all material risks to which the firm is
the upcoming capital planning and accounting
exposed over the course of an annual planning
guidance. This assessment will need to include
cycle.” The Fed discusses the importance of
a multitude of facets, including data lineage
utilizing multiple scenarios “to assess a broad
capabilities, operational efficiency, linkage to
range of risks, stressful conditions or events,”
business-as-usual processes, robust internal
and evaluating its impact on the capital position
controls, and the ability to support a range of
of the firm. New, efficient tools will need to
end users. Systems will need to be able to trace
be developed to address these expectations
loss estimates back to their sources (including
and ensure banks are capturing all material
loss models and overlays), as well as incorporate
risks in their bottom-up stress scenario
critical “top of the house” capabilities, such as
analysis. Options to consider include applying
establishing the linkage between the firm’s risk
simulation-driven portfolio management
appetite statement and the risk profile of the
tools for enterprise-level sensitivity analysis
current positions and the pro-forma estimates.
and developing less granular top-down stress
For many institutions, this will mean kicking
scenario models that are based on the firm’s
off multi-year transformational projects that
more detailed bottom-up approaches.
will shape the future of their organization.
Conclusion Understanding the key linkages between
Contrasting recent news articles with the finance and risk will be important in developing
recently published regulatory guidance suggests information systems that can meet the needs
that while front offices buzz with excitement of of the many internal stakeholders. It may take
fintech and blockchain, back offices of financial years to fully realize the value of these changes.
institutions continue to need significant However, firms will have the opportunity for
infrastructure improvements. The libretto of many “quick wins” that will lead to cost savings
how banks move from the current state to the and better business decisions along the way,
end-state infrastructure that meets the “new and that will ultimately lead to using their
normal” of regulatory expectations is still being stress testing processes in a more efficient and
written. That said, the themes have been framed strategic manner.
6 Transition Resource Group for Impairment of Financial Instruments, Incorporation of forward-looking scenarios, December 11,
2015.

Another
Standout Year
Globally and locally acknowledged
for award-winning tools to measure
and manage risk.
MoodysAnalytics.com/Awards
RiskTech ®
2016 Category Winner

Enterprise Stress Testing
Liquidity Risk Technology Ranked Top-5 in 5 out of 11 Front-Line Customer Service #1 Enterprise-Wide Credit Winner Enterprise
Implementation of the Year performance categories: 2011, Team of the Year – Financial Risk Management and Stress Testing
2012, 2014, 2015 Expectations Services Industries #1 Economic and Regulatory Overall #7 out of 100
(1-, 2-, 3-, 4-Year Horizons) Risk Capital Calculation
Best Solvency II Software #1 Regulatory Capital Recognized as a Top Recognized as a Top #1 Risk Management
Best Economic Scenario Calculation and Management Solution Provider Solution Provider Regulatory/Economic
Generation Software and #1 Economic Capital Capital Calculation
Calculation and Management
91
ANACREDIT: A NEW APPROACH TO BANKING
REGULATORY COMPLIANCE
By Michael van Steen
The European Central Bank’s Analytic Credit (AnaCredit) regulation

Michael van Steen
Senior Director, Enterprise Risk significantly expands European banks’ regulatory reporting
Solutions architecture and will drive more granular data reporting. It is
expected that this granular reporting will likely be replicated
Michael helps deliver advanced portfolio credit risk, in many, if not most, regulatory jurisdictions. To help better
stress testing, correlation, and valuation solutions understand this specific effort and its larger consequences, this
to global financial institutions and regulatory
organizations. He is the practice lead for origination
article summarizes AnaCredit’s rationale, presents its historic and
services in the Americas, developing and managing future timelines, and highlights its features and challenges. The
services around stress testing, lending workflows, article concludes by offering some guidance on how institutions can
pricing, and limit setting.
best meet the challenges of and benefit from the work required by
AnaCredit.
During the crisis, eurozone banks were unable to identify and aggregate
credit exposures, despite the widespread availability of credit data and
elaborate reporting architectures.
Since the banking crisis, there has been an for banks to submit comprehensive, granular
extensive overhaul of the global banking datasets for detailed analysis by regulators.
regulatory framework. The overhaul has been AnaCredit will run in parallel to the existing
led by Basel III but also encompasses regimes frameworks.
including IFRS 9 and stress testing.
AnaCredit will require significantly more
An overarching theme of the changes is exposure-by-exposure credit risk information,
much more extensive reporting, and banks such as collateral values and probabilities of
are now required to create a much wider set defaults (PDs), as well as significantly more
of compliance reports. They must complete accounting metrics. It will be rolled out in phases
significantly more work to consolidate data, starting in September 2018, and will feature
calculate results, and submit reports to results and provisions computed according
regulators, using a predefined format. to the new IFRS 9 accounting standard being
Analytical Credit Datasets, or AnaCredit for introduced in January 2018.
short, is a new regulatory framework that will Although AnaCredit will be for the eurozone, it
be introduced in the eurozone. It will take a is anticipated that its granular approach will be
different approach to regulating banks. Rather replicated in most regulatory jurisdictions.
than submitting a report for review, it calls
This article summarizes the rationale behind

REGULATORY REVIEW
AnaCredit, as well as relevant timelines, features, other element, AnaCredit, is a much more
and challenges. It concludes by offering some comprehensive regulatory initiative which
guidance on how institutions can best meet emphasizes more frequent and much more
AnaCredit challenges and benefit from the work granular data submissions.
required.
The goal of AnaCredit – to create a eurozone-
Background wide Central Credit Registry – will be a new
Credit institutions around the world are in near- development in the European regulatory
constant communication with their regulators. landscape. Before this, many national rules and
This communication provides regulators technology hurdles inhibited the aggregation of
with information about institutions’ liquidity data outside of individual states. Post-crisis, the
conditions, capital levels, and credit exposures. arrival of a European Union (EU) banking union
Lessons from the banking crisis have driven and the emergence of new powerful statistical
AnaCredit will fundamentally transform the European regulatory landscape.
regulators to have access to more specific, and data-handling technologies (“big data”) have
credible, and timely information so they can enabled the creation of AnaCredit.
identify and address real and potential issues. AnaCredit will fundamentally transform the
During the crisis, eurozone banks were unable European regulatory landscape. It will push
to identify and aggregate credit exposures, frequent, fine-grained, and comprehensive data
despite the widespread availability of credit submissions to the center of regulation and
data and elaborate reporting architectures. compliance. Regulators will use this data as their
Credit exposure data gaps around particular primary means to monitor and mitigate credit
branches, or the total borrowings of a firm across issues at the institutions they regulate. It is
institutions, persist to this day. In addition, critically important for these credit institutions
the constantly evolving regulatory reporting to implement the systems and data processes
architecture is a significant burden on regulated needed to successfully deploy AnaCredit.
institutions, which are responsible for both
AnaCredit Overview
reporting their data and monitoring regulatory
The European System of Central Banks (ESCB),
changes.
comprised of the ECB and the National Central
To address these problems, the European Banks (NCBs) of EU member states, is driving the
Central Bank (ECB) has established a high-level AnaCredit framework.
roadmap. One element focuses on incremental
AnaCredit builds on Central Credit Registers
efforts, such as common data dictionaries
(CCR) now used in many eurozone countries
and a common data reporting framework. The
by NCBs to collect credit data and monitor and

manage credit risk. Financial institutions under The regulation proposes reporting a minimum
the Single Supervisory Mechanism (SSM) will of 95 credit risk and accounting attributes,
need to submit much more granular data, on along with seven identifiers. However, at
more obligors, more frequently to their NCBs. national discretion, some authorities may
The NCBs, in turn, will provide this data to the require reporting of additional data attributes,
ECB to form AnaCredit. potentially 100 or more data elements for each
AnaCredit Stages credit exposure.
The AnaCredit will be established in stages. AnaCredit Data
The first stage will start on September 1, 2018. The ECB proposes reporting 10 interrelated
Submissions during this first stage will include datasets, each organized around individual
information on debtors who are legal entities instruments or a single counterparty, as shown
and who have instruments which 1) give rise in Figure 1. All datasets would include internal
to credit risk and 2) total €25,000 or greater. identifiers, which are intended to have no
All credit instruments of these debtors will be meaning outside of AnaCredit. These internal
reported. identifiers would allow data to be cross-
Two subsequent stages, though not currently referenced and uniquely identified.
spelled out in the regulation, are likely to be AnaCredit Reporting Frequency

implemented in 2020 and 2021, respectively.
The proposed reporting frequency for these data
Stage Two will likely include Information on
types is monthly, quarterly, or when the data
instruments such as financial derivatives,
changes. A significant challenge will be to keep
other accounts receivables, and off-balance
track of the changes in data, to avoid reporting
sheet exposures. Stage Three will likely include
the data set twice, before and after the change.
anonymized information on mortgage loans
Complementary Efforts
to households and credit granted to sole
proprietors. AnaCredit will require extensive harmonization
with other frameworks with regard to data and
AnaCredit Scope
reporting concepts and definitions. Related
Credit institutions operating within the
frameworks include:
eurozone, along with the resident foreign
»» The European Reporting Framework (ERF):
branches of credit institutions, are subject
This harmonization of primary regulatory
to AnaCredit rules. It is not anticipated that
reporting covers credit, balance sheet,
institutions would have to report loans booked
income, and interest rate reporting, among
at a branch headquartered outside the eurozone;
others.
however, countries may have discretion
over this requirement. Additionally, in some »» The Banks’ Integrated Reporting Dictionary
circumstances, national central banks may opt (BIRD): This covers common banking data
to exempt certain small credit institutions from terminology and data transformations.
reporting. »» The ECB’s Single Data Dictionary (SDD):
AnaCredit Data and Reporting This defines common data terminology and
The key highlight of AnaCredit is that data transformations for use in reporting within
is taking the place of reporting. Instead of the ECB and the NCBs.
specifying a report layout, the proposed »» The ongoing LEI effort to standardize obligor
regulation specifies data tables and data fields identification.
that need to be reported. Credit instruments are
the centerpiece of this proposed data model, Banks will need to utilize advanced analytic and
and the data is proposed to be collected on a ‘’big data” tools to manage this significantly
loan-by-loan and borrower-by-borrower basis. increased data.

REGULATORY REVIEW
Figure 1 AnaCredit datasets
Counterparty Joint Liabilities

Financial Data
Risk Data Data
Counterparty Reference Counterparty-

Instrument Data
Data Instrument Data
Counterparty Protection Received Instrument-Protection

Accounting Data
Default Data Data Received Data
Data Set Composition

Instrument Data Includes attributes such as instrument type, inception date, payment
frequency, settlement date, and interest rate-related data
Financial Data Contains an instrument’s financial characteristics, such as drawn
and undrawn amounts for credit lines
Accounting Data Contains an instrument’s accounting characteristics, such as the
accounting classification of the instrument
Counterparty Instrument Data Contains information about the counterparties of each instrument
Joint Liabilities Data Contains the joint liability amount for each debtor in relation to
an instrument, for any instrument where each debtor is liable at
national discretion
Instrument Protection Received Data Describes all the protection received in relation to the protected
instrument
Protection Received Data Includes protection type and value, approach and date of
protection valuation, and real estate collateral location, as well
as the original value, valuation date, and maturity date of the
protection
Counterparty Reference Data Contains various information about a counterparty, ideally
including a unique counterparty identifier, like a Legal Entity
Identifier (LEI), for each counterparty in the instrument
Counterparty Risk Data Contains counterparty risk data as probability of default
Counterparty Default Data Allows for a quick identification of counterparties in default, and
includes default status and date of default
Source: European Central Bank

AnaCredit Changes and Challenges challenges for banks, as they will likely need
AnaCredit presents institutions with significant to enhance their data management and
technological, data management, and reporting processes. Banks have already
operational challenges, including the following: made extensive investments in managing
»» Consolidating and processing data from compliance with Basel III, stress testing, and
different sources in the institution. AnaCredit capital planning, and now will need to do the
requires banks to consolidate and process same with AnaCredit.
data from across a range of systems which

There may be a light at the end of the tunnel.
can be highly complex. Banks will likely need
The ECB hopes that the comprehensiveness
to revisit their processes and data aggregation
and granularity of AnaCredit will help reduce
tools.
the volatility of reporting changes. This should
»» Maintaining data quality. AnaCredit requires help to reduce reporting implementation costs
an extensive set of information for individual going forward. The ECB also hopes that banks
credit exposures. Some information may be will benefit from the standardized loan-level
either very hard to find, or never collected taxonomy of AnaCredit which will simplify the
in a systematic fashion. Banks will need to acquisition of new entities and ease integration
carefully consider how to fill in, proxy, or of this data into new systems.
otherwise account for this missing data. This
Banks also hope that modern data and statistical
will likely be subject to detailed regulatory
management tools that can scale to analyze
scrutiny.
terabytes and petabytes of volatile data can be
»» Implementing accurate and robust reporting used in AnaCredit initiatives.
systems. Banks must ensure the data
A Path Forward with AnaCredit
assembled for AnaCredit produces results
A key challenge for banks is finding a way to
that are consistent with all other reports of
balance the need for expanded and more closely
the same information, including COREP and
monitored data requirements and more frequent
Pillar 3 disclosures. The size and frequency of
reporting, with an efficient and effective data
AnaCredit submissions requires a reliable and
management and reporting process.
robust reporting platform.
The key highlight of AnaCredit is that data is taking the place of reporting.
»» Aggregating and presenting group and An optimal solution for implementing AnaCredit
standalone data. AnaCredit focuses on should deliver the following results:
individual exposures as well as differing
»» The seamless consolidation of risk and finance
national Central Credit Registries across
data from different sources into a central
the eurozone. A bank with an obligor with
internal repository
borrowings across several countries in the
»» Accuracy and consistency so that common
eurozone will need to report these exposures
data definitions and calculations shared
in different ways on a country-by-country
between AnaCredit, Basel III, and IFRS are
basis.
always consistent
»» Complex implementation. AnaCredit
»» Automation to drive efficient, accurate, and
significantly expands the data that
cost-effective compliance and reporting
institutions must provide. This presents

REGULATORY REVIEW
At its core, the optimal solution needs to AnaCredit reporting templates to streamline the
consolidate all the loan-level and counterparty reporting process and assure accuracy.
data into a unified data set to provide solid
This consolidated approach to AnaCredit
foundations for AnaCredit calculations and
reporting can lend itself to integrated reporting
reports. It needs to have powerful data
for Basel III, IFRS 9, and stress testing.
At its core, the optimal solution needs to consolidate all the loan-level and
counterparty data into a unified data set.
cleansing capabilities, so managers can quickly Consolidated reporting helps a bank deliver a
identify and address data that does not meet the consistent, accurate message across multiple
bank’s data quality standards. The solution also regulatory regimes, while leveraging a single
needs to be open and flexible, so it can import data source to deliver cost-effective regulatory
risk and balance sheet information quickly and compliance and reporting.
easily.
Conclusions
The solution also needs to have a fully AnaCredit is a major change to eurozone
integrated, credit risk-weighted asset calculation reporting architectures as it moves regulatory
engine so banks can compute their credit risks compliance away from ever-changing reports to
at loan level. Furthermore, the solution needs to a more data-rich submissions framework.
have an integrated IFRS 9 calculation engine that
In the short- to medium-term, focus should be
can calculate the expected credit loss provisions
on ensuring reporting architecture is as efficient
at loan level.
and transparent, both internally and externally,
Finally, the solution should leverage a powerful, as possible.
integrated data publishing solution that can
In the longer term, there is potential for reduced
consolidate risk, finance, and other results
reporting costs due to more stable reporting
from across the business to meet the frequent
processes, greater use of automation, and the
reporting requirements of AnaCredit. The
application of powerful new tools.
solution should also allow banks to apply
AnaCredit project, European Central Bank, 2016.

Draft: Regulation on the collection of granular credit and credit risk data, European Central Bank, December 2015.

SUBJECT MATTER EXPERTS
Barnaby Black Cayetano Gea-Carrasco
Assistant Director, Economist Managing Director – Advisory Services
Barnaby Black is an Assistant Director in Cayetano Gea-Carrasco works with financial

the Moody's Analytics London office. His institutions to address their technology and
responsibilities involve leading client projects enterprise risk management needs. Previously,
regarding IFRS 9 impairment, capital and he held leadership positions at various
impairment forecasting, and stress testing, institutions and global banks. He is a regular
particularly for retail and SME portfolios. speaker at international conferences and
He has an MS in Operational Research and has published several articles in the areas of
Statistics from Southampton University and risk management, financial technology, and
a BS in the same field from the University of derivatives pricing. Cayetano holds a BSc. and
Kent. MSc. in Telecommunication Engineering, a
[email protected] master’s in Economics and Finance, and an MSc.

in Financial Mathematics, with distinction, from
King’s College London.
Dr. Shirish Chinchalkar
Managing Director – Quantitative Finance [email protected]
MoodysAnalytics.com/CayetanoGea-Carrasco
Dr. Shirish Chinchalkar heads the group
responsible for building credit models for
different retail asset classes such as US, UK, Christian Henkel
Senior Director – Risk Management
and Dutch residential mortgages, US auto
loans, and credit cards, as well as for several Christian Henkel is a Senior Director in
different US and EMEA asset-backed securities. the Moody’s Analytics Risk Measurement
His group also implements the models in the Services group. He leads a national team of
Portfolio Analyzer platform which is used risk consultants who work closely with banks,
for stress testing and risk management of insurers, and other financial institutions. Having
retail and structured portfolios. He has a PhD spent most of his career in commercial banking,
from Cornell University and a BTech from IIT Christian has a unique blend of business and
Bombay. academic experience across the financial services
industry – including expertise in areas such
[email protected]
as commercial credit and financial analysis,
MoodysAnalytics.com/ShirishChinchalkar
portfolio management, and asset quality. He has
a BA and an MBA from the University of Texas
Christopher Crossen
at Dallas in Finance and Accounting, and he
Associate Director – Research
graduated valedictorian from the Southwestern
Christopher Crossen is the Research Writer Graduate School of Banking at SMU.
for Moody’s Analytics Quantitative Research
[email protected]
Group. He has worked for Moody’s Analytics
MoodysAnalytics.com/ChristianHenkel
since 2008. Prior to Moody’s, he worked as
an analyst in equity research and portfolio
management. He earned advanced degrees
from UC Berkeley’s Haas School of Business
and the University of Notre Dame.
[email protected]

Dr. Anthony Hughes Dr. Juan M. Licari
Managing Director of Credit Analytics Managing Director, Chief International Economist
Tony oversees the Moody’s Analytics credit Dr. Juan M. Licari is a Managing Director at
analysis consulting projects for global lending Moody's Analytics and the head of the Economic
institutions. An expert applied econometrician, and Consumer Credit Analytics team in EMEA. Dr.
he has helped develop approaches to stress Licari’s team provides consulting support to major
testing and loss forecasting in retail, C&I, industry players, builds econometric tools to
and CRE portfolios and recently introduced a model credit phenomena, and has implemented
methodology for stress testing a bank’s deposit several stress testing platforms to quantify
book. He received his PhD in Econometrics from portfolio risk exposure. He has a PhD and an MA
Monash University in Melbourne, Australia. in economics from the University of Pennsylvania
[email protected] and graduated summa cum laude from the
MoodysAnalytics.com/TonyHughes National University of Cordoba in Argentina.
[email protected]
Anna Krayn MoodysAnalytics.com/JuanLicari
Senior Director and Team Lead, Capital Planning
and Stress Testing
David Little
Anna is responsible for the business development Managing Director, Head of the US Enterprise Risk
of stress testing and capital planning solutions. Solutions and Stress Testing Sales Teams
Her clients include a variety of financial services David is responsible for helping financial
institutions, including those in the insurance, institutions worldwide with their enterprise
banking, and consumer finance sectors across risk management, liquidity, and stress testing
the Americas. solutions. Since joining Moody’s in 2002, David
[email protected] has been Managing Director in the Product
MoodysAnalytics.com/AnnaKrayn Strategy group, responsible for a global portfolio
of research, data, and analytic products across all
fixed income asset classes, led end-of-day pricing
Glenn Levine
business at Moody’s Evaluations Inc., and headed
Associate Director, Senior Research Analyst
the Global Structured Finance Sales team at
Glenn Levine is an Associate Director in the
Moody’s Analytics.
Moody’s Analytics Capital Markets Research
[email protected]
Group. He provides support for the EDF product
MoodysAnalytics.com/DavidLittle
suite and is the lead researcher for Stressed
EDF. Prior to his current role, he was a Senior
Economist in the Economics and Consumer
Credit division based in Sydney, Australia.
He holds an MS from the London School of
Economics and a bachelor’s degree from the
University of New South Wales.
[email protected]
MoodysAnalytics.com/GlennLevine

Emil Lopez Nancy Michael
Director – Risk Measurement Senior Director – Product Strategy
Emil Lopez is a Director in the Moody’s Analytics Nancy works to conceive and build innovative
Risk Measurement Group, where he leads risk solutions for credit assessment of small
modeling advisory engagements and manages businesses. Drawing on her previous experience
the team's data quality, risk reporting, and IFRS co-founding a small business, she has built
9 research. Prior to joining the group, he oversaw products and strategies to help financial
operations for Moody's Analytics Credit Research companies better serve the needs of their
Database, one of the world's largest private firm customers. Nancy previously led the Client
credit risk data repositories. Emil has extensive Solutions team for the Training and Certification
experience in credit risk modeling and reporting, division and headed Strategy and Marketing
data sourcing, and quality control. Emil has for the company’s training and consulting
an MBA from New York University and a BS in businesses. Nancy received a BS in Economics
Finance and Business Administration from the from the Wharton School of the University of
University of Vermont. Pennsylvania, with concentrations in Strategic
[email protected] Management and Legal Studies.
MoodysAnalytics.com/EmilLopez [email protected]
MoodysAnalytics.com/NancyMichael
Director, Specialized Modeling Group Nihil Patel
Senior Director – Product Management
Dr. Malone develops novel risk and forecasting
solutions for financial institutions while Nihil Patel is a Senior Director within the
providing thought leadership on related trends Enterprise Risk Services division at Moody's
in global financial markets. He also frequently Analytics. He serves as the business lead driving
leads consulting projects taken on by the our product strategy related to credit portfolio
Specialized Modeling Group. He is coauthor analytics. Nihil has broad experience in research,
of the highly-cited book Macrofinancial Risk modeling, service delivery, and customer
Analysis, which offered the first operational, engagement. Prior to his current role, Nihil spent
validated systemic risk framework in the midst nine years in the Research organization leading
of the global financial crisis, as well as numerous the Portfolio Modeling Services team as well as
peer-reviewed academic articles in applied the Correlation Research team. Nihil holds a MSE
and financial econometrics. He has BS degrees in Operations Research and Financial Engineering
in Mathematics and Economics from Duke from Princeton University and a BS in Industrial
University, where he graduated summa cum Engineering and Operations Research from UC
laude, and a doctorate in economics from the Berkeley.
University of Oxford. [email protected]
[email protected] MoodysAnalytics.com/NihilPatel
MoodysAnalytics.com/SamMalone

Michael van Steen Ed Young
Senior Director, Enterprise Risk Solutions Senior Director, Capital Planning and Stress Testing
Michael helps deliver advanced portfolio credit Ed is a Senior Director on the Stress Testing
risk, stress testing, correlation, and valuation and Capital Planning Team. In this capacity,
solutions to global financial institutions and he focuses on structuring solutions that bring
regulatory organizations. He is the practice together capabilities across Moody’s Analytics
lead for origination services in the Americas, to support robust capital planning and stress
developing and managing services around stress testing processes. His primary focus is on clients
testing, lending workflows, pricing, and limit in the banking and insurance sectors across the
setting. Americas.
[email protected] [email protected]
MoodysAnalytics.com/MichaelvanSteen MoodysAnalytics.com/EdYoung
Dr. Yashan Wang

Senior Director – Research
Dr. Yashan Wang is a Senior Director at Moody’s

Analytics where he leads the Research and
Quantitative Modeling team for portfolio
valuation and balance sheet analytics. He
has led several research initiatives in asset
valuation, credit migration, joint credit and
interest rate dynamics, and balance sheet
analytics. Yashan has also worked with global
clients and provided training and advice on
enterprise risk management, asset and liability
management, PPNR, and stress testing. Prior
to joining Moody’s Analytics, Yashan was an
Assistant Professor at the MIT Sloan School of
Management. He has a PhD in Management
Science from Columbia University.
[email protected]
MoodysAnalytics.com/YashanWang

CONNECT WITH US
Get More Insight from

Moody’s Analytics
There are many ways to view our content online. Register just once and you’ll
gain free access to all of our valuable insight, including Risk Perspectives articles,
webinars, white papers, and more.
»» View previous editions of Risk Perspectives at

MoodysAnalytics.com/RiskPerspectives
»» Sign up for our newsletters at
MoodysAnalytics.com/Register
»» Learn about our events at
MoodysAnalytics.com/Events
»» Stay in touch with our Subject Matter Experts at
MoodysAnalytics.com/SME
»» Follow our social channels for the latest insight at
MoodysAnalytics.com/Connect
About Risk Perspectives

Each edition of Risk Perspectives magazine explores an industry or regulatory topic
in depth, presenting a wide range of views, best practices, techniques, and
approaches, all with one larger goal in mind – to deliver essential insight to the
global financial markets.
ABOUT US
Moody’s Analytics offers award-winning solutions and best practices for We help organizations answer critical risk-related questions, combining
measuring and managing risk through expertise and experience in credit best-in-class software, analytics, data and services, and models –
analysis, economic research, and financial risk management. By providing empowering banks, insurers, asset managers, corporate entities, and
leading-edge software, advisory services, data, and research, we deliver governments to make informed decisions for allocating capital and
comprehensive investment, risk management, and workforce solutions. maximizing opportunities. Through training, education, and certifications,
As the exclusive distributor of all Moody’s Investors Service content, we help organizations maximize the capabilities of their professional staff
we offer investment research, analytics, and tools to help debt capital so they can make a positive, measurable impact on their business. More
markets and risk management professionals worldwide respond to an information is available at MoodysAnalytics.com.
evolving marketplace with confidence.

GLOSSARY OF TERMS
AIRB Advanced Internal Ratings-Based FIRB Foundation Internal Ratings-Based
ALLL Allowance for Loan and Lease Losses FVOCI Fair Value through Other Comprehensive Income
API Application Program Interface FVTPL Fair Value through Profit or Loss
ASC Accounting Standards Codification GAAP Generally Accepted Accounting Principles
BCBS Basel Committee on Banking Supervision GDP Gross Domestic Product
BIRD Banks’ Integrated Reporting Dictionary GL General Ledger
BHC Bank Holding Company IAS International Accounting Standard
CCAR Comprehensive Capital Analysis and Review IASB International Accounting Standards Board
CDF Cumulative Distribution Function IFRS International Financial Reporting Standard
CECL Current Expected Credit Loss IRB Internal Ratings-Based
COREP Common Reporting LEI Legal Entity Identifier
CRC Credit Risk Cascades LGD Loss Given Default
CRE Commercial Real Estate LISCC Large Institution Supervision Coordinating Committee
DFAST Dodd-Frank Act Stress Tests LTV Loan-to-Value
EAD Exposure at Default NCO Net Charge-Off
EBA European Banking Authority PCAOB Public Company Accounting Oversight Board
ECB European Central Bank PD Probability of Default
ECL Expected Credit Loss PDF Probability Density Function
EDF Expected Default Frequency PIT Point-in-Time
EIR Effective Interest Rate PPNR Pre-Provision Net Revenue
EL Expected Loss QE Quantitative Easing
ERF European Reporting Framework QIS Quantitative Impact Study
ESCB European System of Central Banks SDD Single Data Dictionary
EU European Union SEC Securities and Exchange Commission
FAS Financial Accounting Standard SEDF Stressed Expected Default Frequency
FASB Financial Accounting Standards Board SCCL Single Counterparty Credit Limit
FATCA Foreign Accounting Tax Compliance Act SSM Single Supervisory Mechanism
FCAG Financial Crisis Advisory Group TTC Through-the-Cycle

© 2016 Moody’s Corporation, Moody’s Investors Service, Inc., Moody’s Analytics, Inc. and/or their licensors and affiliates (collectively, “MOODY’S”). All rights reserved.
CREDIT RATINGS ISSUED BY MOODY’S INVESTORS SERVICE, INC. AND ITS RATINGS AFFILIATES (“MIS”) ARE MOODY’S CURRENT OPINIONS OF THE RELATIVE FUTURE CREDIT RISK OF
ENTITIES, CREDIT COMMITMENTS, OR DEBT OR DEBT-LIKE SECURITIES, AND CREDIT RATINGS AND RESEARCH PUBLICATIONS PUBLISHED BY MOODY’S (“MOODY’S PUBLICATIONS”)
MAY INCLUDE MOODY’S CURRENT OPINIONS OF THE RELATIVE FUTURE CREDIT RISK OF ENTITIES, CREDIT COMMITMENTS, OR DEBT OR DEBT-LIKE SECURITIES. MOODY’S DEFINES
CREDIT RISK AS THE RISK THAT AN ENTITY MAY NOT MEET ITS CONTRACTUAL, FINANCIAL OBLIGATIONS AS THEY COME DUE AND ANY ESTIMATED FINANCIAL LOSS IN THE EVENT
OF DEFAULT. CREDIT RATINGS DO NOT ADDRESS ANY OTHER RISK, INCLUDING BUT NOT LIMITED TO: LIQUIDITY RISK, MARKET VALUE RISK, OR PRICE VOLATILITY. CREDIT RATINGS
AND MOODY’S OPINIONS INCLUDED IN MOODY’S PUBLICATIONS ARE NOT STATEMENTS OF CURRENT OR HISTORICAL FACT. MOODY’S PUBLICATIONS MAY ALSO INCLUDE
QUANTITATIVE MODEL-BASED ESTIMATES OF CREDIT RISK AND RELATED OPINIONS OR COMMENTARY PUBLISHED BY MOODY’S ANALYTICS, INC. CREDIT RATINGS AND MOODY’S
PUBLICATIONS DO NOT CONSTITUTE OR PROVIDE INVESTMENT OR FINANCIAL ADVICE, AND CREDIT RATINGS AND MOODY’S PUBLICATIONS ARE NOT AND DO NOT PROVIDE
RECOMMENDATIONS TO PURCHASE, SELL, OR HOLD PARTICULAR SECURITIES. NEITHER CREDIT RATINGS NOR MOODY’S PUBLICATIONS COMMENT ON THE SUITABILITY OF AN
INVESTMENT FOR ANY PARTICULAR INVESTOR. MOODY’S ISSUES ITS CREDIT RATINGS AND PUBLISHES MOODY’S PUBLICATIONS WITH THE EXPECTATION AND UNDERSTANDING
THAT EACH INVESTOR WILL, WITH DUE CARE, MAKE ITS OWN STUDY AND EVALUATION OF EACH SECURITY THAT IS UNDER CONSIDERATION FOR PURCHASE, HOLDING, OR SALE.
MOODY’S CREDIT RATINGS AND MOODY’S PUBLICATIONS ARE NOT INTENDED FOR USE BY RETAIL INVESTORS AND IT WOULD BE RECKLESS AND INAPPROPRIATE FOR RETAIL
INVESTORS TO USE MOODY’S CREDIT RATINGS OR MOODY’S PUBLICATIONS WHEN MAKING AN INVESTMENT DECISION. IF IN DOUBT YOU SHOULD CONTACT YOUR FINANCIAL OR
OTHER PROFESSIONAL ADVISER.
ALL INFORMATION CONTAINED HEREIN IS PROTECTED BY LAW, INCLUDING BUT NOT LIMITED TO, COPYRIGHT LAW, AND NONE OF SUCH INFORMATION MAY BE COPIED OR
OTHERWISE REPRODUCED, REPACKAGED, FURTHER TRANSMITTED, TRANSFERRED, DISSEMINATED, REDISTRIBUTED OR RESOLD, OR STORED FOR SUBSEQUENT USE FOR ANY SUCH
PURPOSE, IN WHOLE OR IN PART, IN ANY FORM OR MANNER OR BY ANY MEANS WHATSOEVER, BY ANY PERSON WITHOUT MOODY’S PRIOR WRITTEN CONSENT.
All information contained herein is obtained by MOODY’S from sources believed by it to be accurate and reliable. Because of the possibility of human or mechanical error as well as other
factors, however, all information contained herein is provided “AS IS” without warranty of any kind. MOODY’S adopts all necessary measures so that the information it uses in assigning a
credit rating is of sufficient quality and from sources MOODY’S considers to be reliable including, when appropriate, independent third-party sources. However, MOODY’S is not an auditor
and cannot in every instance independently verify or validate information received in the rating process or in preparing the Moody’s Publications.
To the extent permitted by law, MOODY’S and its directors, officers, employees, agents, representatives, licensors and suppliers disclaim liability to any person or entity for any indirect,
special, consequential, or incidental losses or damages whatsoever arising from or in connection with the information contained herein or the use of or inability to use any such information,
even if MOODY’S or any of its directors, officers, employees, agents, representatives, licensors or suppliers is advised in advance of the possibility of such losses or damages, including but
not limited to: (a) any loss of present or prospective profits or (b) any loss or damage arising where the relevant financial instrument is not the subject of a particular credit rating assigned by
MOODY’S.
To the extent permitted by law, MOODY’S and its directors, officers, employees, agents, representatives, licensors and suppliers disclaim liability for any direct or compensatory losses or
damages caused to any person or entity, including but not limited to by any negligence (but excluding fraud, willful misconduct or any other type of liability that, for the avoidance of doubt,
by law cannot be excluded) on the part of, or any contingency within or beyond the control of, MOODY’S or any of its directors, officers, employees, agents, representatives, licensors or
suppliers, arising from or in connection with the information contained herein or the use of or inability to use any such information.
NO WARRANTY, EXPRESS OR IMPLIED, AS TO THE ACCURACY, TIMELINESS, COMPLETENESS, MERCHANTABILITY OR FITNESS FOR ANY PARTICULAR PURPOSE OF ANY SUCH RATING
OR OTHER OPINION OR INFORMATION IS GIVEN OR MADE BY MOODY’S IN ANY FORM OR MANNER WHATSOEVER.
Moody’s Investors Service, Inc., a wholly-owned credit rating agency subsidiary of Moody’s Corporation (“MCO”), hereby discloses that most issuers of debt securities (including corporate
and municipal bonds, debentures, notes and commercial paper) and preferred stock rated by Moody’s Investors Service, Inc. have, prior to assignment of any rating, agreed to pay to Moody’s
Investors Service, Inc. for appraisal and rating services rendered by it fees ranging from $1,500 to approximately $2,500,000. MCO and MIS also maintain policies and procedures to address
the independence of MIS’s ratings and rating processes. Information regarding certain affiliations that may exist between directors of MCO and rated entities, and between entities who hold
ratings from MIS and have also publicly reported to the SEC an ownership interest in MCO of more than 5%, is posted annually at www.moodys.com under the heading “Investor Relations —
Corporate Governance — Director and Shareholder Affiliation Policy.”

THE CONVERGENCE OF RISK, FINANCE, AND ACCOUNTING VOL 7 | JUNE 2016
THE CONVERGENCE OF RISK, FINANCE, AND ACCOUNTING VOL 7

FEATURED ARTICLES:
ONLINE
IMPLEMENTING AN IFRS 9 SOLUTION 8

Challenges faced by financial institutions
Find additional risk, finance, and accounting articles, interviews, IFRS 9 SURVEY RESULTS 28
and multimedia content at MoodysAnalytics.com/RiskPerspectives Preparedness for and opinions on IFRS 9 guidelines

SCENARIO PROBLEM 74
JUNE 2016
A 2016 CCAR case study
MARP_RFAC_COVER_FINAL_Print_edit_0.25spine.indd 1 6/14/2016 4:54:38 PM

Moodys Analytics Risk Perspectives Risk Finance Accounting

Uploaded by

Copyright:

Available Formats

Moodys Analytics Risk Perspectives Risk Finance Accounting

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

Moodys Analytics Risk Perspectives Risk Finance Accounting

Uploaded by

Copyright:

Available Formats

THE CONVERGENCE OF RISK, FINANCE, AND ACCOUNTING VOL 7 | JUNE 2016

THE CONVERGENCE OF RISK, FINANCE, AND ACCOUNTING VOL 7

IMPLEMENTING AN IFRS 9 SOLUTION 8

SOLVING THE COUNTERPARTY DEFAULT

A 2016 CCAR case study

MARP_RFAC_COVER_FINAL_Print_edit_0.25spine.indd 1 6/14/2016 4:54:38 PM

FROM THE EDITOR 1

Preparing for the New Impairment Requirements: Practitioner's View 11

IFRS 9 Survey Results 28

IFRS 9 Impairment Regulations: Implementation Challenges 34

Probability-Weighted Outcomes Under IFRS 9: A Macroeconomic Approach 42

Complying with IFRS 9 Impairment Calculations for Retail Portfolios 48

PRINCIPLES AND PRACTICES

AnaCredit: A New Approach to Banking Regulatory Compliance 92

4 MOODY’S ANALYTICS RISK PERSPECTIVES

THE CONVERGENCE OF RISK, FINANCE, AND ACCOUNTING | JUNE 2016 5

IFRS 9 is a new international standard set forth to address the

»» Classification and Measurement: IFRS »» The measurement of allowance of credit loss

8 MOODY’S ANALYTICS RISK PERSPECTIVES

credit losses on historical, current, and required by IFRS 9 in a cost-effective, scalable

»» The forward-looking provision framework and expected modeling challenges. It must

subsequent reporting date. capable of modeling or importing PD, LGD,

IFRS 9 provision calculation will require integration of multiple processes

a driver of provision and toward quantitative

and the implementation of IFRS 9 solutions at impairment metrics. Financial institutions

financial institutions. usually have several general ledgers within a

due to effect on earnings) will be critical. In requirements: The IFRS 9 forward-looking

calculation. In terms of architecture design, testing. Institutions will want to do facility-

layer, model risk management and workflow processes, will be imperative.

For financial institutions transitioning to IFRS 9, »» Underwriting, risk-adjusted pricing, and

»» Systems, processes, and automation: or lifetime) and monitor for ongoing

Systems will need to change significantly deterioration of credit quality.

in order to calculate and record changes »» Risk-adjusted pricing metrics: Pricing

THE CONVERGENCE OF RISK, FINANCE, AND ACCOUNTING | JUNE 2016 9

requirements will lead to related challenges, make better-informed decisions, proactively

including: manage provisions and effects on capital plans,

10 MOODY’S ANALYTICS RISK PERSPECTIVES

PREPARING FOR THE NEW IMPAIRMENT

banks should align with the new CECL impairment standards,

THE CONVERGENCE OF RISK, FINANCE, AND ACCOUNTING | JUNE 2016 11

Federal Reserve Standards

Aim to improve financial

1917 Establishes the Securities and Exchange Commission (SEC)

1921 1929 1965

Revenue Act Fraud Exposed Loan Loss Estimation

Recognizes reserves Major corporate accounting Process established

12 MOODY’S ANALYTICS RISK PERSPECTIVES

Joint Working Group

Banking supervisory agencies

All banks with total Interagency policy CECL

See Figure 6 for 2008-2016 details.

1973 1986 1994 2006

FASB Public Statement General Accounting Office Policy Update

THE CONVERGENCE OF RISK, FINANCE, AND ACCOUNTING | JUNE 2016 13

Assets 2015 2014

Cash and due from banks $ 251,258 $ 261,544

Interest-bearing deposits 155,907 132,695

Securities available-for-sale, at fair value 2,984,631 2,793,873

Securities held-to-maturity 3,923,052 3,872,955

Loans held for sale 37,091 67,952