Stress Testing Webinar Series: Macroeconomic Conditional Pre-Provision Net Revenue (PPNR) ForecastingJanuary 28, 2014

Presented by: Moody’s Analytics

2

Agenda

1. Introduction

2. Regulatory Expectations: PPNR

3. Emerging Quantitative Methodologies: New Ideas on Old Processes

4. Practical Implementation Issues: Innovation and the Road Ahead

5. Conclusion: Better Start than Good

6. Next Webinar:

Stress Testing Methodologies: Enhancing Data and Loss Estimation for DFAST Banks March 18, 2014 | 12:00pm EST

3

Introduction 1

4

Overall Progress: Integrated Financial Risk Forecasting

» Post Financial Crisis, it was clear that the manner in which risk was analyzed resulted in wreckage.

» Many bank managers and supervisory authorities were essentially flying blind.

» Data wasn’t actionable and risk could not be aggregated and analyzed on demand.

» New ideas were needed.

» From the wreckage, the DFA mandated regular stress-tests, and the FRB designed and executed.

» Very little discussion around alternative design, perhaps hampering innovation.

» Current tools are providing some lift; however, legacy processes were not designed for integrated financial and risk analytics. Current state remains brittle.

» Banking is changing. Banks need to be fast, agile, and able to run away from the competition.

» Innovation is needed – new ideas that allow the bank to maintain performance in good times and in bad.

» In order to design the proper “risk management platform” and analytics, new ideas are needed – ideas that can break outmoded barriers to effective risk management.

Movie: “Flight of the

Phoenix”

5

Stress-Testing and Capital Planning

Financial and Risk Forecast» Pro-forma balance sheet (under scenarios)» PPNR» Losses, charge-offs, and recoveries» Valuations» Operational risk(s)» Accounting measures (e.g., DTA, Goodwill)» Documentation and Validation

Commercial Lending

Retail Lending

Discretionary Portfolio

Finance and Accounting

Treasury

Funding

Credit Risk

Trading

Capital Planning

Industry Observations:

» The stress-testing process requires an unprecedented amount of coordination and collaboration across numerous front, middle, and back office functions.

» Communication, documentation, and well defined business processes are required, and assumptions made to conditional forecasts require justification.

» Governance of the process can be as important as the result(s).

» Risk quantification is critical at all levels, with challenger approaches considered sound practice.

» PPNR estimates are notoriously complex in that centralized estimates may miss necessary SME input from lines of business (e.g., how the business(es) would actually react under stress). Quantification processes are generally preferred, with overlays well justified.

» Creating increased efficiency in the process is necessary, create cost savings, and improve operational resilience.

6

CCAR/DFAST: Process Complexity – PPNR Issues

Treasury/ALM and Finance

Balances, Revenues, Expenses,

Accounting

Economics Group

Credit Risk

Other

Stress-Testing and Capital Planning Committee

Financial Forecast

Q1, Q2, Q3, …Q13

Loss Estimates, NPAs, Delinquencies,

Ratings, etc.

Portfolio and Credit Research

Scenario(s) and Economic Research

PPNR

Commercial Lending

Region 1 Region 2 Region 3 Region 4

Line of Business: Challenge Models and Results

Finance, Treasury, and Risk: Develop Forecast and Risk Estimates

PPNR, loss estimates, charge-off estimates, rating distribution(s), non-performing levels, new originations and new origination spreads, capital estimates, operational losses, and other measures.

Scenario Analyzer

»Workflow»Messaging»Document

Management»On-Demand

Collaboration»Assumption

Management»Auditability»Transparency

»Model Management»Input/Output

Management»Scenario Management»Data exchange»Regulatory Reporting»Dashboard Reporting»System Integration

» Process Governance, Automation, Assumption, Model, and Results Management are critical for an effective CCAR/DFAST program.

» Assembling stress-test reports, and validating results from the bottom-up, requires structured processes.

» For PPNR, validating and agreeing estimates should also be “bottom-up” and leverage LOB models and SME input.

Workflow 1

Workflow 2

Workflow “n”

Results Data-mart

FRY-14A

Wo

rk P

acka

ge

Val

idat

ion

an

d C

hal

len

ge

PPNR, Losses, Charge-off, Recovery, etc

1.2.

3.

4.

7

Stylized Workflow for DFAST/CCAR Exercise» While presented as a sequential workflow, this is not realistic or practical. The CCAR/DFAST workflow must be

instantiated to work in an asynchronous fashion and robustly address numerous hand-offs, edits checks, task schedules, and interactions. The entire “chain of custody” must be transparent and auditable.

Data Pull as of Sept-30

Fill-in “Missing” Data with Proxy Data (inc. Tags)

Populate Required Fields for FRY-

14M/Q

Document Workflow, Version, and Audit the Data

Receive ScenariosExpand and

“Regionalize” Scenarios

Ensure Market Data is Consistent with

the ScenarioTailor Scenarios

Calculate Conditional ELs

Across All Assets

Determine Business Strategy in Each Scenario

Create Proper Assumption Input

for Integrated PPNR

Calculate Expected NII/NIM and

Balance Sheet for Each Scenario

Calculated Expected NIR and

NIE in Each Scenario

Determine Charge-Off and ALLL in Each Scenario

Assess and Apply Other Losses,

Including Ops Risk

Calculate Appropriate Pro-

Forma Regulatory Capital

Populate Required Regulatory

Reporting Forms

Reconcile Reports to FRY-9C and

Other Reporting

Assess and Validate Results

Apply Measures to Capital Plan

Data

ScenarioDesign

Analytics

Reporting

8

PPNR = “Interest Income” – “Interest Expense” + “Non-Interest Income” – “Non-Interest Expense”

PPNR Requires a “N” Quarter Forecast: Full Balance Sheet

Net Interest Income + Non-interest Income - Non-interest Expense = Pre-provision Net Revenue (PPNR)

Note: PPNR includes Losses from Operational Risk Events, Mortgage Put-back Losses, and OREO Costs

PPNR + Other Revenue - Provisions - AFS/HTM Securities Losses -Trading and Counterparty Losses - Other Losses (Gains)

= Pre-tax Net Income Note: Provisions = Change in the Allowance for Loan and Lease + Net

Charge-offs

Pre-tax Net Income - Taxes + Extraordinary Items Net of Taxes = After-tax Net Income

After-tax Net Income - Net Distributions to Common and Preferred Shareholders and Other Net Reductions to Shareholder's Equity

= Change in Equity Capital

Change in Equity Capital - Deductions from Regulatory Capital + Other Additions to Regulatory Capital

= Change in Regulatory Capital

9

Pre-Provision Net Revenue (PPNR)

» One of the most challenging components of the stress-testing exercise – an emerging area of practice with little available research.

– Biggest areas of challenge: 1) joint modeling of credit , interest rate, and capital risk in a unified framework and/or calculation, 2) data, 3) NIR and NIE, and 4) conditional balance sheet dynamics

» An area of note by the Federal Reserve as “lacking coherence” between credit loss estimates and the resulting impact on net interest income, and other areas of income and expense.

» Banks are required to forecast quarterly by FRB defined business segment, as well as a BHC view. Revenues should tie to the FRY9C net of any valuation adjustment for the firm’s own debt and operational expenses.

– May require new dimensions within a firm’s ALM Chart of Accounts

– Various metrics required, such as average yields, average rates on interest bearing liabilities, WAM, deposit repricing betas and estimated WAL of non-maturity deposits

– Significant historical PPNR data and metrics are also required to be submitted

» The NII by business segment must be FTP adjusted, based on the firm’s own internal FTP pricing methodologies.

10

End-State Goals: Areas for Consideration (2014)

» Developing tangible, practical business uses for stress-testing investments. For example, the same process that creates stressed measures should be capable of:– Sensitivity analysis around “expected” results, not just major systemic shocks– Computation of many more scenarios than the extreme shocks required for the regulatory

exercise– Integrating analytical capabilities into useful tools for on-going deal and relationship analysis– Creating side-by-side views of economic and regulatory returns on capital, at any required

dimension

» A single “run-time” compute that accommodates monthly credit loss and PPNR coherence, by scenario and by asset class.– Provides coherence among interest income, FTP interest expense, prepayment, credit loss,

credit migration, economic and regulatory capital calculations

» A single environment to manage data and work packages that are sequenced through Treasury, Finance, and Credit Risk. The environment should permit: – Management of multiple hierarchies across numerous lines of business and entities– Use and re-use of current and historical scenarios, market data, and instrument data– Serve as a single point of entry for management and use of multiple models, with input and output

results versioned and persisted– Act as the main aggregation area for regulatory and management reporting

» Enhancements to conditional volume and spread estimates

» Enhancements to conditional estimates of NIR and NIE

11

Calculation Engine: Unified Credit and Interest Rate » For some hard to model asset classes, creating a unified calculation capability can be managed by

calling a separate library that directly incorporates primary and challenger credit models. Inputs to the credit model may be: 1) PD/LGD/EAD (monthly/loan level), 2) parameter estimates (e.g., bank internal credit risk models), and/or 3) native (library) credit model.

ALCO Report

FRY-14A

Risk Report

Deal Analysis

» C&I» CRE» Other Asset Classes

Interest Rate and PPmt Process

Credit Risk Model

Calibration Framework

Migration Matrice(s)In

terf

aceJoint interest rate and

credit dynamics

Results Data-mart

» P&I Cash Flows» Credit/non-credit

adjusted» Prepayments» Additional property

sets» Pro-rata NIR and

NIE allocation» RWA» Regulatory and

Economic Capital

ETL Process

Use of result output for sensitivity analysis

Import cash flows to

ALM/FP&A

Import cash flows to

regulatory reporting

Pricing and Performance measurement

» ETL out process

» Data consolidation and reporting

Loan level. Ability to roll-up to any hierarchy level. Supports all

reporting and business processes.

Scenario Data

Market Data

Contractual Terms

» FRY-14Q/M (subset/monthly)

» Monthly stress-scenarios» Monthly calibrated market

data

Consistent input data

Internal Models

12

Example: Integrated Dashboard Report – Current State

1

2

3

4

5

6

13

Regulatory Expectations: PPNR2

14

Regulatory Methodological Expectations

» PPNR must be estimated over the same range of scenarios used for loss estimation

– Implies that market data used for calculations are consistent with the economic conditions

» Banks must consider scenario impact on current position business as well as how origination strategy may change in different scenarios. Banks are expected to model the balance sheet using contractual terms and capture behavioral characteristics.

– Deposit growth, new business pricing, balances, line usage, changing fees, expenses, etc

– Quantitative techniques help support more subjective estimates

– Baseline estimates should be consistent with internal plans and ALM assumptions, and proper adjustments to optimistic baseline plans must be considered in the scenarios

» Pro-forma RWA calculations should consider how management actions may impact capital ratios

– Can require the modeling of the credit quality of new origination, and losses that may be attributed to those balances

» Balance sheet and income statement projections should present a “coherent story”

» Better practice involves a robust interaction between FP&A, credit risk/business lines, and central treasury. Challenge processes are normally used.

» Clear mapping between internal projections and the FRY14 categories

15

Where to Start: Creating Tactical Value

» Demand and supply functions by asset class

– What available lending will prevail in the various macroeconomic scenarios?

– What is the assumed credit quality of these balances and how are losses estimated?

– How are they allocated to various business lines?

– What are the earnings (interest and non-interest) on these balances? That is, how does credit spread change across scenarios, product type and assumed credit quality?

» Deposit growth and pricing

» Full incorporation of Basel I and III estimates, inclusive of changing credit and mix

» Integrating credit loss estimates into a coherent calculation of net interest income

– Top-down adjustment(s) to asset balances based on aggregate loss estimates

– Transition matrices (quarterly) by asset class, by scenario, scaled to target projected non-performing asset levels indicated by quantitative and qualitative assessment

– Direct integration of loss model into loan-level cash flow compute (i.e., treating default as a proper behavioral option)

» Scenario conditioned non-interest revenue and expense modeling

16

Emerging Quantitative Methodologies: New Ideas on Old Processes3

17

PPNR = Interest Income – Interest Expense + Non-Interest Income – Non-Interest Expense

Interest Income*

» Loans– Existing Book– Less all run-off– Plus new loans

» Securities– Existing Book– Less all run-off– Plus new securities

Interest Expense*

» Deposits– Interest vs. non-interest bearing– By Line of Business / Product– Client vs. wholesale funded– Term structures

» Bonds– Existing– Funding gap for additional bond issuances

Non-Interest Income

» Credit Related Fees– By Product / Line of Business– Origination vs. Servicing (esp. for resi mortgage)– Credit Card

» Non-Credit Related– Investment Banking– Investment Management / Trust– Deposit Service Fees– Trading

Non-Interest Expense

» Employee Compensation– Salary– Benefits– Bonuses

» Processing / Software» Occupancy (Plant, Property & Equipment)» Credit / Collections» Residential Mortgage Repurchases

* Note: Interest Income less Interest Expense = Net Interest Income (NII)

PPNR consists of numerous components from income and expenses from various areas of a bank

18

Two Approaches to estimate Interest Income and Balance: Direct and Granular

Direct Approach Granular Approach

» Moody’s models Total Balance for each segment directly. For revolvers, the balance model estimates the total commitment, but together with Moody’s Usage model obtains estimates for the outstanding drawn amount.

» Simplifying assumptions are required for the interest earned on the balance

» Moody’s models Usage, New Origination, and Runoff (prepayment, maturity and amortization, provisioning, etc.). Together, these models produce an estimate for balance

» Moody’s models Interest Rate Charged for New Origination. Together with the rates paid by the surviving loans from previous period, this model can be used to calculate an estimate for total interest earned

Seg

men

tati

on

» Direct and Granular approaches to modeling balance both allow for consistency across the balance sheet and income statement if applied to both PPNR and Loss models.

19

Example: Direct Balance Model for Term Loans

,

- are the selected CCAR macroeconomic variables used

-

- are segment characteristics

Q3 2003

Q2 2004

Q1 2005

Q4 2005

Q3 2006

Q2 2007

Q1 2008

Q4 2008

Q3 2009

Q2 2010

Q1 2011

Q4 2011

Q3 2012

Q2 2013

Q1 2014

Q4 2014

Q3 20158090

100110120130140

Term Loans

Actual Fitted BaseAdverse Severe

QuarterBala

nce

(200

3 Q

3) =

100

20

Modeling Runoff for the Granular Approach

» The granular approach involves modeling the different components responsible for balance development over time. For example, Term Loans:

» Runoff includes balance depletion due to Prepayment, Maturity, Amortization, Provisioning, …

» Modeling Runoff

– Derive Runoff using the Balance and New Originations models:

– Explicit runoff modeling allows for differentiation across Runoff components:

» Maturity and Amortization: Model the relationship between maturity/amortization of new origination and the macro environment at origination using Moody’s CRD data (LAS dataset) and the institution’s own data

– Can be combined with segmentation by Tenor for the Interest Charged model to refine the interest earned projections

» Provisioning: Leverage Loss stress testing models

» Prepayment: Leverage Moody’s Analytics lattice model (details in next slide)

21

Using the Moody’s Lattice* to Model Prepayments

» Prepayments for floating loans are mostly driven by improvements to borrower’s credit quality

» The Moody’s Lattice model captures borrowers’ credit migration dynamics, and can produce prepayment rates (even at the individual loan level) given the current credit state and contractual interest charged

» The Lattice model also accounts for prepayment penalty/cost, which can be calibrated to empirical prepayment data

Valuation Lattice

0

3

6

9

12

15

0 1 2 3 4 5

Time (Year)

Cre

dit

Sta

te

Prepay

6 (Default)

1

3

4

5

2

*Moody’s Lattice is available in RiskFrontierTM

22

Modeling C&I Balance, New Origination, Interest Charged, Usage and EAD with the Credit Research Database (CRD)

» World’s Largest Historical Time Series of Private Firm Middle Market Data for C&I Loans– Consortium of 49 Banks Operating Globally including 19 from the US

– Defaulted and Non-defaulted Private Firm Financial Statement Data

– Obligor & Loan Level Accounting Data

» Allows for segmentation based on risk factors that can mimic the institution’s portfolio– Borrower PD

– Industry: for example, Financial vs. Non-Financial

– Firm Size

– Geographical location

– Loan Tenor

– New vs. Old borrower

23

Segmentation by Credit Quality for Balance and New Origination

» In general, high and low credit quality segments exhibit different time dynamics, suggesting it’s beneficial to model them separately.

» Balance of the High PD segment increased significantly in 2006-2008.

» During the crisis, both PD segments exhibit a sharp decline in balance (slightly steeper for low PD firms).

» However, post-crisis lending to low PD (high quality) firms has recovered much faster than to high PD (low quality) firms.

2003

Q3

2004

Q1

2004

Q3

2005

Q1

2005

Q3

2006

Q1

2006

Q3

2007

Q1

2007

Q3

2008

Q1

2008

Q3

2009

Q1

2009

Q3

2010

Q1

2010

Q3

2011

Q1

2011

Q3

2012

Q1

0

20

40

60

80

100

120

140

160

Term Loan Balance

Low_PD High_PD All_PD

Quarter

Bala

nce

(200

3 Q

3 =1

00)

-0.02

1.04083408558608E-17

0.02

0.04

0.06

0.08

0.1

0.12

0.14

Term Loan New Orig. over Balance (NoB)

All_PD High_PD Low_PD

Quarter

NoB

24

Segmentation by Credit Quality for Interest Charged

2001

Q1

2001

Q3

2002

Q1

2002

Q3

2003

Q1

2003

Q3

2004

Q1

2004

Q3

2005

Q1

2005

Q3

2006

Q1

2006

Q3

2007

Q1

2007

Q3

2008

Q1

2008

Q3

2009

Q1

2009

Q3

2010

Q1

2010

Q3

2011

Q1

2011

Q3

2012

Q1

2012

Q3

0

0.01

0.02

0.03

0.04

0.05

0.06

Average Term Loan Spread by Credit Quality

All_PD Low_PD High_PD

Quarter

Spre

ad

» In general, High PD borrowers are charged higher spreads than Low PD borrowers.

– This is especially pronounced during the financial crisis, where the Market Price of Risk was highest.

25

Segmentation by Financial vs. Non-Financial20

03 Q

3

2004

Q1

2004

Q3

2005

Q1

2005

Q3

2006

Q1

2006

Q3

2007

Q1

2007

Q3

2008

Q1

2008

Q3

2009

Q1

2009

Q3

2010

Q1

2010

Q3

2011

Q1

2011

Q3

2012

Q1

020406080

100120140160

Term Loan Balance

All Financial Non FinancialQuarter

Bala

nce

(200

3 Q

3 =1

00)

» Prior to the crisis, both segments exhibit a similar growth pattern, with financial firms growing faster right before the crisis.

» The financial crisis seems to have affected new origination to Financials more severely than Non-Financials:– During the crisis, lending to financials starts

shrinking a few quarters earlier than lending to non-financials, and at a faster pace. Non-financials exhibit a drop only after the Lehman Brothers collapse.

– Even though financials tend to be larger and safer, lending to them has remained constant since 2010, while lending to non-financial firms has recovered.

2001

Q1

2001

Q3

2002

Q1

2002

Q3

2003

Q1

2003

Q3

2004

Q1

2004

Q3

2005

Q1

2005

Q3

2006

Q1

2006

Q3

2007

Q1

2007

Q3

2008

Q1

2008

Q3

2009

Q1

2009

Q3

2010

Q1

2010

Q3

2011

Q1

2011

Q3

2012

Q1

00.020.040.060.08

0.10.120.14

Term Loan New Orig. over Balance (NoB)

All Financials Non Financials

Quarter

NoB

For Financials Term Loan

% Large (>80MM) 39% % High PD 26%

For Non-Financials Term Loan

% Large (>80MM) 5% % High PD 46%

26

Segmentation by Size for Balance and New Origination» In general, both segments exhibit different time

dynamics, suggesting it is beneficial to model them separately.

» Prior to the crisis, both segments experienced steady growth, with large firms experiencing a higher increase than small firms right before the crisis.

» During the crisis, lending to both segments decreased.

» Large firms show fast recovery after the crisis

» On the contrary, lending to small firms has remained low since the crisis - resulting in continued decrease in balance, and increasing the gap between the two segments.

Note: for financial firms, large firms have total assets > 80MM. For non-financial firms, large firms have total sales > 80MM.

2003

Q3

2004

Q1

2004

Q3

2005

Q1

2005

Q3

2006

Q1

2006

Q3

2007

Q1

2007

Q3

2008

Q1

2008

Q3

2009

Q1

2009

Q3

2010

Q1

2010

Q3

2011

Q1

2011

Q3

2012

Q1

020406080

100120140160

Term Loan Balance

Small Large All_size

Quarter

Bala

nce

(200

3 Q

3=10

0)

2001

Q1

2001

Q3

2002

Q1

2002

Q3

2003

Q1

2003

Q3

2004

Q1

2004

Q3

2005

Q1

2005

Q3

2006

Q1

2006

Q3

2007

Q1

2007

Q3

2008

Q1

2008

Q3

2009

Q1

2009

Q3

2010

Q1

2010

Q3

2011

Q1

2011

Q3

2012

Q1

00.020.040.060.08

0.10.120.140.16

Term Loan New Orig. over Balance (NoB)

All_size Large Small

Quarter

NoB

For Large (>80MM) Term Loan

% Financial 64% % High PD 21%

For Small Term Loan

% Financial 15% % High PD 48%

27

Data and Modeling Approach for CRE Interest Income

» For CRE Interest Income, the same modeling approaches described for C&I Interest Income (i.e., Direct and Granular) can be used, except for– CRE lines of credit are uncommon, so there is no need for a Usage model

– Different segmentation is recommended for CRE loans (by property type)

» Data for CRE modeling– Balance Model: Call Reports and FR Y-9C data on CRE loan growth rates segmented by

property type for large commercial banks and BHCs

– New Origination Model: Mortgage Banker Association‘s New Origination Index

– Interest Charged Model: Moody’s CMM provides credit quality measures (PD, LGD), that can be translated to loan-level spreads» Loan-level granularity on interest income forecasts

» Loan-level spread models for new origination, based on credit quality (LTV, DSCR etc) and terms

» Integration with runoff estimates (or assumptions)

» Consistent integration with stress testing losses: same PD, LGD, Balance and New Origination models used in loss and income calculations

28

Projecting Deposit Interest Expense

» Banks typically have tactic models that forecast the runoff of existing deposits and link new deposit volumes to interest rates offered and non-interest expenses

» Statistical models that project total deposit balances under various macro-economic scenarios, can be used in combination of a bank’s tactic models to project interest expenses onto existing stock and on new volumes

» Possible modeling approaches include:– Historical deposit data of an individual bank – often reflect idiosyncratic events of the bank and may

not sufficiently capture how future macro-economic factors would impact deposit volumes.

– Call report data from peer institutions can be used to develop deposit balance models for broad deposit categories: interest checking, non-interest checking, MMDA, other savings, time deposits

» Deposit balances often exhibit seasonality; season dummy variables are often useful and significant in regression analyses

» Regression coefficients are estimated based on historical data Parameters estimated from historical data are used to produce deposit balance projections under various future scenarios:–

29

Sample Deposit Balance Modeling Results

30

Modeling Non-Interest Income and Expense

Data Type Entity Type Primary Source Description

Line Items for FR Y-14A

FDIC insured subsidiaries

Call Reports, SDI data

» Data from 2001 onward.

» Mergers and acquisitions can be accounted for by consolidating historical statements of merged institutions.

» Macro factors are based on CCAR scenarios

BHCs FR Y-9C

Bank-specific

FR Y-14Q » FR Y-14Qs allow us to adjust model outputs to the level of granularity needed to populate FR Y-14A.

Macro variables All Federal Reserve CCAR 2013 Scenarios

31

Example: Modeling Overhead Expense Using Peer Groups

,

- are the selected CCAR macroeconomic variables

- is the BHC’s asset size at time t.

2001Q1

2001Q4

2002Q3

2003Q2

2004Q1

2004Q4

2005Q3

2006Q2

2007Q1

2007Q4

2008Q3

2009Q2

2010Q1

2010Q4

2011Q3

2012Q2

2013Q1

2013Q4

2014Q3

2015Q2

0.30%

0.40%

0.50%

0.60%

0.70%

0.80%

0.90%

1.00%

Overhead Expense over Assets (BHCs Data)historical 1-quarter projection baseline adverse severe

Year, Quarter

OE

/ A

sset

32

Practical Implementation Issues: Innovation and the Road Ahead4

33

Leveraging Balance Sheet Management Systems for CCAR» CCAR is a daunting task for any financial institution. For the first time banks are required

to assemble enterprise projections for earnings and capital that model the joint dynamics of market and credit risk under multiple macro-economic scenarios.

» Existing Balance Sheet Management (BSM) systems can serve as a source for CCAR stress testing outputs

» Macro-economic data» Market data i.e.

rates/prices» Detailed rate/maturity data» Detailed repricing data» New volume assumptions

including rates and spreads

» Prepayment assumptions» PD/LGD Assumptions» Charge offs

Enterprise Data Warehouse

Net Interest Income:» Forecast balances» Interest income

Credit Exposure:

CCAR INPUTS CCAR OutputsPROCESSING

Cash Flow Engine

Risk Data Scenarios Reporting

Chart of Accounts Credit Default/Loss

Market Data Mgr.

Behavior Models

FTP

Pro-forma Bal. Sheet

Formula Builder

Value at Risk

» PD/LGD/EAD» Expected loss

Analytics:» Credit Migration» Fair value» Charge offs/impairment

34

Modeling the Pro-forma Balance Sheet» The Balance Sheet Strategy (BSS) is a practical way to model the joint dynamics of the enterprise

level balance sheet.

» New volumes may be modeled in great detail i.e. by term, credit rating, etc. within a single account yielding a better and more accurate loss, income, and capital forecast

» Contractual product features may be controlled by forecast period permitting more dexterity in terms of responding to macro-economic forces with contractual and option-like features.

35

Modeling Demand Functions» Formula builders can automate the impact of macro-economic variables on new volumes

» A powerful language based syntax can allow the user to express logic and mathematical equations

» Formulas allow cross product references for balances, market data, and economic variables. Allows user to specify scenario based demand functions.

» Many formula builders are interpreted. The MA platform formula builder is compiled adding speed and flexibility.

Macro-economic indices

Lagged Market Data

Dynamic Credit Metrics

Segment Balances

36

The Impact of changes in Credit State on Cash Flows

» The Fed is very explicit about incorporating the impact of market volatility and credit forces on cash flows:- “The methods BHCs use to project their net interest income should be able to capture dynamic conditions

for both current and projected balance sheet positions. Such conditions include but are not limited to prepayment rates, new business spreads, re-pricing rates due to changes in yield curves, behavior of embedded optionality that Capture FAS 91 adjustments related to prepayment changes as caps or floors, call options, and/or changes in loan performance (that is, transition to nonperforming or default status) consistent with loss estimates.”; Capital Planning at Large Bank Holding Companies: Supervisory Expectations and Range of Current Practice; Board of Governors of the Federal Reserve System; August 2013; Page 33

Credit State

State specific prepayment assumptions

37

Other Assets/Liabilities and Non Interest Income/Expense

» The accounting identity ‘Assets = Liabilities + Equity’ must be true; if not, the pro-forma balance sheet and NI projection fall into question.

» Good BSM systems have the ability to natively incorporate ‘systems accounts’ but many banks do not use them fully. Examples include:- Accrued interest receivable/payable- Accrued principle receivable- AFS/HTM gains and losses/impairment- Charge Offs and Provision

» For stand alone applications like pre-trade analytics, IRR quantification, FTP, or capital management, the bare minimum was good enough. However, in CCAR, the regulatory community is saying that BSM needs to more prospective and holistic. Therefore, all of the macro-economic, risk factor, and accounting interrelationships matter.

» BHCs should clearly establish and incorporate into their scenario analysis the relationships among and between revenue, expense, and on- and off-balance sheet items under stressful conditions. Most BHCs used asset-liability management (ALM) software as a part of their enterprise-wide scenario-analysis toolkit, which helps integrate these items. BHCs that do not use ALM software must have a process that integrates balance sheet projections with revenue, loss, and new business projections. BHCs with more tightly integrated procedures were better able to ensure appropriate relationships among the scenario conditions, losses, expenses, revenue, and balances. Capital Planning at Large Bank Holding Companies: Supervisory Expectations and Range of Current Practice; Board of Governors of the Federal Reserve System; August 2013; Page 37

- Balancer accounts- Retained earnings- Taxes/Deferred tax liability- Dividends

38

Non Interest Income and Expense

» Modeling Non Interest Income and expense items can be tough because the GL may not match the granular account structure of BSM systems.

» However, many BSM systems have income/expense accounts that can be specified either as interest earning/interest costing or non-interest earning/non-interest costing.

» BSM systems typically have features that permit the user to allocate income/ expense items from aggregates to detailed income expense items. Therefore, if desired, very rich and detailed fee schedules can be created.

» In addition, the Formula builder can be used to create models that generate fee and expense schedules based on balances or other financial results i.e. deposit servicing fees.

39

ALL, Provision, and Charge Offs

» Allowance for Loan and Lease Losses (ALLL) is an asset contra account where provisions are capitalized on the balance sheet until the corresponding assets default and are charged off.

» Provision can be specified as a function of both an ALLL target and forecast charge-offs. Some BSM systems have the ability to target the provision and perform re-provisioning from pre-tax net income based on native loss calculations and custom frequencies. Therefore, a high degree of automation and consistency among results is possible at many institutions.

» The timing of charge offs may vary based on the asset class. Therefore, a BSM system that is used to produce charge offs must have the ability to forecast loan status and have rules that determine when assets should be charged off and when ALLL needs to be re-provisioned.

» Some of the advantages of modeling the whole balance including mark to market gains and losses and provision in a single BSM engine are efficiency, consistency across multiple risk management functions, and the capability to capture balance sheet interrelationships including the compounding of equity.

40

Calculation Engine: Unified Credit and Interest Rate » For some hard to model asset classes, creating a unified calculation capability can be managed by

calling a separate library that directly incorporates primary and challenger credit models. Inputs to the credit model may be: 1) PD/LGD/EAD (monthly/loan level), 2) parameter estimates (e.g., bank internal credit risk models), and/or 3) native (library) credit model.

ALCO Report

FRY-14A

Risk Report

Deal Analysis

» C&I» CRE» Other Asset Classes

Interest Rate and PPmt Process

Credit Risk Model

Calibration Framework

Migration Matrice(s)In

terf

aceJoint interest rate and

credit dynamics

Results Data-mart

» P&I Cash Flows» Credit/non-credit

adjusted» Prepayments» Additional property

sets» Pro-rata NIR and

NIE allocation» RWA» Regulatory and

Economic Capital

ETL Process

Use of result output for sensitivity analysis

Import cash flows to

ALM/FP&A

Import cash flows to

regulatory reporting

Pricing and Performance measurement

» ETL out process

» Data consolidation and reporting

Loan level. Ability to roll-up to any hierarchy level. Supports all

reporting and business processes.

Scenario Data

Market Data

Contractual Terms

» FRY-14Q/M (subset/monthly)

» Monthly stress-scenarios» Monthly calibrated market

data

Consistent input data

Internal Models

41

Consistency Across Basel III and Treasury Risk Management Functions

Cash Flows & Behavior Models

42

Conclusions: Better Start than Good5

43

Integrated Financial and Risk ForecastingThree-tier (and “N” tier) architecture is fundamental to good systems design. A proper platform is modular and

Comprehensive, and creates a “future proof” design that embraces internal and 3rd party technologies.

Analytic Layer: For DFAST/CCAR purposes, best practice is to begin with the analytical layer and supporting models while working towards automation of data and reporting.

1.

Data Layer: For DFAST/CCAR purposes, and to target required data reporting, many banks must launch a DataFoundation data project. The goal is to target a single data platform to support risk, finance, credit, and regulatory reporting and capital planning needs.

2.

Reporting Layer: The DFAST/CCAR reports are complex, and must be reconciled to FRY-9C, FFIEC 031/041, Basel FFIEC 101, and other internal management reports. Automating this process must leverage work performed from the Analytic Layer and the Data Layer.

3.

44

Fully Integrated Architectural DesignModular, Flexible and Comprehensive – Allowing for Straight Through Risk Processing

Spreading SystemRiskAnalyst / RiskOrigins

Core Systems(e.g. GL, Loan Accounting)

Risk and Finance Datamart(Inputs and Results)

DA

TA L

AY

ER

- RiskFoundation Datamart as an integrated risk and finance data layer is the foundation for stress testing

- RiskFoundation can be integrated with various data sources, including enterprise data warehouses and core banking systems

- Our solution design accommodates comprehensive regulatory reporting, internal risk and LOB reporting, plus dimension / hierarchy management:

- Executive and board-level reporting

- Instantiation of the organization’s Risk Appetite Framework(s)

- Existing and expected liquidity risk reporting

- Drill-through and scenario dependent PPNR, balance sheet and new business volume

- Comprehensive wholesale and retail credit portfolio reporting

Management Reporting / Dashboard

Risk & Performance Management

RE

PO

RT

ING

Regulatory Reporting

- Part of Potential Moody’s Solution

- Bank’s Internal / Third Party Systems

- Moody’s is able to work with existing Treasury, FP&A and Risk systems to coordinate, enhance and improve stressed cashflow calculation and transparency

- By linking results from point solutions to the reporting layer, Moody’s can empower the bank by providing key linkage between input data and output results.

Risk Management and ALM System Data

Credit Models(Wholesale & Retail)

Budgeting & Planning System Output

AN

ALY

TIC

LA

YE

R

NCOs ALLL PPNR

SCENARIO ANALYZER TM

RWA

RiskAuthority

45

Next Webinar6

46

Moody’s Analytics Stress Testing Webinar Series

Stress-Testing Methodologies: Enhancing Data and Loss Estimation for DFAST Banks

March 18, 2014 at 12:00pm EST

Topics to be covered include:

» Regulatory expectations surrounding data and loss estimation for DFAST banks

» Common themes and issues:  Rating systems, origination and scoring systems, and use of models

» Conditional measures using macroeconomic conditioned correlation models

Register at: http://www.cvent.com/d/z4qplc/4W

47

Questions7

48

moodysanalytics.com

Thomas DaySenior Director

Direct: 404.617.8718Thomas.Day@moodys.com

7 World Trade Center at250 Greenwich StreetNew York, NY 10007www.moodysanalytics.com

Amnon Levy, PhDManaging Director

Direct: 415.874.6279Amnon.Levy@moodys.com

405 Howard StreetSuite 300San Francisco, CA 94105www.moodysanalytics.com

Robert Wyle, CFASenior Director

Direct: 415.874.6603Robert.Wyle@moodys.com

405 Howard StreetSuite 300San Francisco, CA 94105www.moodysanalytics.com

49

Find out more about our award-winning solutions

www.moodysanalytics.com

50

@MoodysAnalyticsStay current with the latest risk management and assessment news, insights, events, and more.

@dismalscientistView global economic data, analysis and commentary by Mark Zandi and the Moody's Analytics’ economics team.

@CSIGlobalEdRead the latest financial services education information

@MA_CapitalMktsKeep up to date on credit and equity market signals reflecting investment risk and opportunities for issuers and sectors.

7 World Trade Center250 Greenwich StreetNew York, NY 10007(212) 553-1653

121 North Walnut StreetSuite 500West Chester PA 19380(610) 235-5299

405 Howard StreetSuite 300San Francisco, CA 94105(415) 874-6000

www.moodysanalytics.com

Moody's AnalyticsFollow our company page to view risk management content, such as white papers, articles, webinars, and other insightful content and news.

The Economic OutlookThis group features insightful discussions and knowledge sharing among business, economics, and policy professionals regarding the economic outlook. 

Risk Practitioner Community This group brings together risk management practitioners from around the world to discuss best practices, share ideas and insights, and gain networking opportunities. 

51

© 2014 Moody’s Analytics, Inc. and/or its licensors and affiliates (collectively, “MOODY’S”). All rights reserved.

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. Under no circumstances shall MOODY’S have any liability to any person or entity for (a) any loss or damage in whole or in part caused by, resulting from, or relating to, any error (negligent or otherwise) or other circumstance or contingency within or outside the control of MOODY’S or any of its directors, officers, employees or agents in connection with the procurement, collection, compilation, analysis, interpretation, communication, publication or delivery of any such information, or (b) any direct, indirect, special, consequential, compensatory or incidental damages whatsoever (including without limitation, lost profits), even if MOODY’S is advised in advance of the possibility of such damages, resulting from the use of or inability to use, any such information. The ratings, financial reporting analysis, projections, and other observations, if any, constituting part of the information contained herein are, and must be construed solely as, statements of opinion and not statements of fact or recommendations to purchase, sell or hold any securities.

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.

Each rating or other opinion must be weighed solely as one factor in any investment decision made by or on behalf of any user of the information contained herein, and each such user must accordingly make its own study and evaluation of each security and of each issuer and guarantor of, and each provider of credit support for, each security that it may consider purchasing, holding, or selling.

Any publication into Australia of this document is pursuant to the Australian Financial Services License of Moody’s Analytics Australia Pty Ltd ABN 94 105 136 972 AFSL 383569. This document is intended to be provided only to “wholesale clients” within the meaning of section 761G of the Corporations Act 2001. By continuing to access this document from within Australia, you represent to MOODY’S that you are, or are accessing the document as a representative of, a “wholesale client” and that neither you nor the entity you represent will directly or indirectly disseminate this document or its contents to “retail clients” within the meaning of section 761G of the Corporations Act 2001.