Interest rate risk modeling day sun_gard_ambit banking

3/20/2011

1

SunGard BankingIRR Modeling and Management

[Business or Company that Needs to Know More aboutBalance Sheet Management] – ask me to come help

DATELOCATION

1

Standard Disclaimer and Ground Rules

The views, expressions and ideas of this presentation are those of the author and do not necessarily reflect the views and opinions of Ambit or SunGard.

Thank you to the [COMPANY] SunGard and all

2

Thank you to the [COMPANY], SunGard and all the participants and organizers of this conference.

Ground rules: Network: Make friends and stay in touch

Engage: Ask many questions

Apply: How do we make a difference tomorrow when on-site?

Who is SunGard?

3

Financial ServicesFinancial Services

3/20/2011

2

Ambit Risk Academy: Bio of Speaker

Thomas DaySunGard Ambit – Risk & PerformanceDirector, Risk & Policy(617) 780-4140 – [email protected]

Role at SunGard:Tom Day is the Managing Director of Risk Solutions and Policy at SunGard Ambit, where he is a key source of thought leadership for the company and its financial institution clients. Leveraging an extensive twenty-year career in banking, risk management and bank supervision, Mr. Day assists clients in navigating through diverse market challenges, regulatory policy

4

changes and business opportunities. His overall technical knowledge across bank balance sheets, capital management and enterprise-risk is comprehensive, and Mr. Day is well known for practical solutions to financial problems from a best-practice, business, accounting and strategic perspective.

Background prior to SunGard:Prior to re-joining SunGard, Mr. Day was Senior Risk and Policy Advisor within the US Treasury Department. In this role, Mr. Day oversaw the TARP program for various chartered institutions and was a member of the interagency review committee that recommended billions of dollars of capital investments into banks and financial institutions. He has also served in senior roles at the Office of the Comptroller of the Currency, the Federal Reserve Board of Governors, AmSouth Bancorporation, SouthTrust Bancorp and Barnett Bank. He previously served as the SVP and head of product management and development at SunGard BancWare, one of the world’s leading providers of risk management software.

Mr. Day has a BS in Economics from Auburn University and is a Commissioned National Bank Examiner and a Commissioned Federal Bank and Financial Holding Company Examiner. He is on the Board of the Professional Risk Managers’ International Association (PRMIA), the Regional Director of the D.C. Chapter of PRMIA, a frequent speaker at various industry conferences and events, a routinely published writer, and an avid fan of American football (particularly the SEC).

Ambit Risk & Performance: ALM, Capital/Stress-Testing, Liquidity, Budgeting, Planning, and FTP

Basel 3

Systemic Risk

European Systemic Risk Board

Ambit Risk Academy Advice, Tools, and SMEs for a Changing Financial Landscape

The Dodd-Frank ActCEBS

Too Big to Fail

Incurred v Expected Losses

SIFI and G-SIFIBalance Sheet Management: Never More Critical

Interest Rate and Liquidity Risk ManagementCapital Buffers

Incentive Compensation and Performance Measurement

G-20 and FSB

Financial Stability Oversight Council

International Harmonization

Cross-Border ResolutionInterconnectedness

Incurred v. Expected Losses

Stress-Testing and Capital Planning

http://www.centerforcapitalmarkets.com/resources/dodd-frank-wall-street-reform-and-consumer-protection-act-of-2010-regulatory-authority/

3/20/2011

3

Impact of Financial Reform on Risk Management?

• Risk Management Impact?

• Full employment at:

7

p y Deloitte, KPMG, PwC, BAH,

Promontory Have regulatory experience,

will hire• Major areas of need:

Infrastructure issues ERM is “for keeps” Massive need for education Cultural challenges Regulatory challenges

Major Areas of Impact: Balance Sheet Management

Agenda: Interest Rate Risk Modeling and Measurement

Supervisory and Business Expectations

ALM5 Capabilities and Limitations

Using the Tool: Risk versus Compliance

IRR Architecture and Key Issues

Summary Review and Conclusions

Sungard Account Management

3/20/2011

4








Convergence: Balance Sheet and Enterprise-Wide Risk Management

Gap Analyses

Duration Analysis

NII Simulation

Economic Value Analysis

Dynamic Simulation Techniques

Fair Value Accounting / Credit Adjustment

Integration of

1970s 1980s 1990s 2000s

Value-at-Risk (VaR)

Sensitivity Analysis

Stress Tests

Monte-Carlo-VaR

Integration ofRisk Management

Black-Scholes, Greeks

Basel I RiskMetrics Basel II

The ALM Risk Domain

Supervisors expect that IRRM will cover: SHORT-TERM RISK to EARNINGS:

Risk to margin, net interest income (NII), and net income

Includes base-runoff cash-flows and new business projections

LONG-TERM RISK to VALUE: Often called “economic value of equity” (EVE); may be other


Often called economic value of equity (EVE); may be other approaches that are acceptable

Bank supervisors (and others) often forget ALM should also encompass: Liquidity

FTP, profitability, performance and budget

Certain aspects of credit risk

Certain aspects of operational risk

3/20/2011

5

BS&R Expectations and Examination Requirements

With regard to interest rate risk management, BS&R should be as interested in the risk-management process as the model (i.e., 80/20 rule)

SunGard has considerable experience with the all regulatory groups and many examiners are familiar with ALM5ALM5

Consistent with SR 95-51, SR 96-13, and the January 6, 2010 Advisory, the regulators should be looking at: Board and senior management oversight

Policies, procedures and limits Not just ALM and IRR

Model risk management, investment portfolio and liquidity

Risk monitoring systems, controls and infrastructure

Internal and external audit, reporting and disclosure13

Good Starting Point: Pragmatic Approach

Models are analytical approximations, or abstractions, of reality that i lif l h d l ti hi i t th ti l

“No matter how good the theory, the first question any good quant should ask of a model: Under what conditions will it fail. The

second: When will it occur? You see, your organization needs to be aware that your models are wrong. Some people use the word

mis-specified, but I don’t like that because there is no obvious Platonic specification of the financial world.” - Emmanual Derman

simplify complex phenomena and relationships into the essential elements that drive risk.

Models are designed to solve or provide direction towards the solution of a particular problem.

What is model risk? The risk that is derived from using assorted variables and mathematical specifications within a model class to produce estimates, forecasts and solutions that do not conform to real world outcomes.

The impact of such risk is felt by an organization when decision-makers make economic commitments based on the results of such output.


Summary of Key Guidance


3/20/2011

6

The Evolution of IRR Supervisory Guidance

The Federal Deposit Insurance Corporation Improvement Act of 1991

• Section 305

• Focus on “Capital Charge”

• OTS adopts a unique approach

• Internal Models Approach (all other Agencies)

Earnings versus Value Debate

• NII forecast


• EVE (beauty is in the eye of the beholder)

Interagency Policy Statement (1996)

• The Board should…

• Senior management should…

• Treasury management should…

• Reporting processes should…

• Rate risk systems should…

• Internal audit should…

The January 2010 Guidance on IRR

Corporate Governance Regular, timely, broad range of participation

Policies and Procedures No mention of credit spreads: value and earnings impacts

Measurement and Monitoring of IRR Input data, cash-flow, aggregation, forecast horizon (5-7 year issue), p , , gg g , ( y ),

dynamic and “static” simulations

Stress-Testing Broader range of scenarios, call for deterministic scenarios

Assumptions Sensitivity analysis, special attention around behavioral models

Risk-mitigation

Internal Controls and Validation









3/20/2011

7

Accrual Book Maturity Model: Maturity Map

• End-state target environment• Governance excellence• Consolidated and

comprehensive on-and off-balance sheet coverage

• Numerous stress, sensitivity, ctic

e

Leadership Quadrant

Proactive Quadrant

What Does Maturity Mean?

Leadership Target

Accrual Book IRR: Maturity Map

2011 ( l)


, y,and strategy scenarios: earnings and value

• Decision-support/ computational speed

• Monthly builds/ model “weight”

• FTP method ownership• Integrated liquidity analysis• Resource strength (human

and technical)• Validation and control

Firm-Level Maturity

Ind

ust

ry B

est

Pra

c

Early-stages Reactive Quadrant

20092010

2011 (goal)

Strategic Balance Sheet Management

1

ALM Process • Import position and market data• Calculate risk measures (income and value)

• Generate ALM report package (min frequency

monthly)

Analyze Reports• EVE sensitivities

(contractual cash flows)• Standard EVE

Risk Committees• Active balance sheet risk

management • Enhance risk/return profile• Modify plans – tactical and

t t i • MVPE• DV01/KRD

• NII Simulation• Re-pricing Gap analysis

2

strategic

4

3

What-if Simulation• Analyze risk exposure – forecast and stress• Measure impact of hedging strategy• Measure sensitivity to key assumptions• Understand behavioral dynamics

Governance: Critical Success Factor

Organizational Design is critical to effective process management

• Effective governance drives the entire balance sheet management process

• ALCO is the action-vehicle

• Treasury must have a clear mandate with well articulated policies and procedures

• Line of business buy-in and cooperation is essential

• Modeling is designed to support the governance process; anything else is /ancillary/secondary

• Reporting and feedback align various components

• Enforce accountability

• Create action items, e.g. hedging

• Value-added, e.g. pricing optimization, elimination of expensive options, funding optimization

All players must understand roles and responsibilities

• Education/Qualification of resources

• Creation of risk management culture

3/20/2011

8

Governance Framework

• Governance

o Comprehensive oversight

o ALCO is action vehicle

• Organizational Structure and Responsibilities

o ERM sponsorship

o Treasury ownership

o LOB cooperation

• Data and Assumptions

o Ownership

o Quality Assurance/Control

• Risk Models

o Design

o Efficiency/Usability vs. Granularity

o Consistent management/analysis across all line models

• Reporting

o Comprehensive presentation of risk limits and exposures

o Actionable information

Effective Business-Level ALM: Communication

ompe

titiv

e P

ositi

onin

g

Risks and OpportunitiesP

lans and ForecastsALM/

F ti

3rd Parties

Retail

A t Fi

Subsidiaries On-going Communication

catio

n

On-going C

omm

u

Str

ateg

ic F

ocus

and

Co

ForecastingUnit

Auto Finance

CRE

Leasing/Factoring

C&I

Investment BankingOn-going Communication

On-

goin

g C

omm

unic

unication

ALM is the “huddle” that brings all business activities together into an overall picture of balance sheet strength and weakness. Ultimately, ALM is an enterprise-risk function

Range of Practice at Large Banks

3/20/2011

9

Range of Practice at Large Banks (continued)

Range of Practice at Large Banks (continued)

Some Interest in Stochastic NII

• Produce a range of NII outcomes based on sampling around the tails rather than the mean.

• Interesting practice for stress-testing; however, terribly difficult to interpret and communicate the usefulness given business


usefulness given business contingencies and other factors.

• Ambit encourages a wide-range of approaches with the goal of establishing a range of views as to the impact of rate and spread movements on the firm’s overall structural position risk.

3/20/2011

10








ALM5 is a dynamic balance sheet income simulation and valuation model Can fully meet all business and regulatory requirements

Can scale to meet any size and complexity requirements

Embeds best-of-breed analytics, access to third-party prepayment and structure models, and valuation technologies

Used to ensure consistency between risk, budgeting/planning, FTP, and liquidity risk management groups. Increasingly used for macro credit

l i

How ALM5 is Used within Financial Organizations

analysis.

The model is used to create multiple market scenarios to ascertain sensitivity to all sources of IRR Repricing, Basis, Yield Curve, and Options Risk

Is used to understand sensitivities to business plan, spread, prepayment, funding and customer behavioral risks

Drives better business decisions through precision of cash flow calculations

Common Use Case: Defined More Clearly

$

CP$

New Business Roll

Forecast Growth

Static balance sheet (a.k.a., flat balance

Forecast error. Ties to budget and planning.

Net Interest Income from this point is typical “risk” measure from S0 to/against S1, S2, Sn

27%

”X”%

Valuation from this point is the typical “risk” measure from Scenario0 (S0) against S1, S2, Sn

t0 t12

Baserunoffsheet)

ALM5 can measure all of these risk components. The manner in which cash flows are modeled (from both an income and valuation view) requires a wide range of assumptions related to the market environment, data, calculations, and risk measures (i.e., output).

t3 t6 t9

73%What is “VaR” at this future point? Forward looking risk measures. T12 or any Tn

3/20/2011

11

Pragmatism in ALM: Necessary, but not Sufficient

ALM5 is a forward-looking, state-of-the-art risk management system used to identify and understand risk → A/L management is not an accounting exercise.

Future balances and behavioral assumptions are forecast with uncertainty, but must foot to plan.

There is a tradeoff between granularity and efficiency.

Too much granularity (i.e. not enough aggregation) will hinder efficiency and slow-down effective decision-making.

The trade-off between accuracy and risk-management pragmatism is an art, but best practices exist as well as supervisory/regulatory expectations. Need to reinforce each other.

Treasury and Risk Management Organization

Implementing improvements in data workflows, ownership of assumptions, and utilization of industry leading practice risk measures will support an organization’s ALCO Process, linking risk groups and decisions in a practical and efficient manner.

This will also successfully address regulatory requirements for adequacy of IRR management.adequacy of IRR management.

ALM5 has all necessary capabilities to support an enterprise-wide ALCO and balance-sheet management process.

It is critical that staff – bank and supervisory – understand that the processes that surround the balance sheet management process are as (often more) important than the model itself. Too many get trapped on the vendor model rather than internal

resourcing, staffing, policies and procedures

All income and all valuation scenarios Spot and future

Static and new business

Daily versus monthly

Integrated liquidity risk management Deposit shock scenarios

What can ALM5 do?(and a side-note about ALM6)

Integration of Contingency Funding Plan

Haircuts to pledged and non-pledged assets

Designed to be actively used Not a compliance tool, but a decision-making tool

Rapid compute times are a priority. Scales linearly with number of CPUs added to analytical server farm

Be adjusted to accommodate unique business problems and needs; designed to be an “open” system

3/20/2011

12








Risk Tensions

Senior Management and ALM goals often can conflict with each other

Sr. Mgt: Give me more earnings. NOW!


ALM: No problem. Here is some more risk.

Sr. Mgt: Wait a minute. Give me less risk!

ALM: No problem. Here is less earnings.

The Role of ALCO: Process is Important

To manage the structure of the bank’s balance sheet and the level of banking book market risk(s)

Ensure effective measurement and reporting systems

Include heads of all major LOBs May also include marketing and investor relationsy g

Balance LOB optimism on business plans with practical financial realities (i.e., a forum for debate across LOBs)

Ensure Treasury’s “discretionary” actions are in alignment with core banking plans and activities

When using correspondent services, all of the above are even more critical for the examiner to review

3/20/2011

13

Key Risks Traditional Covered by IRRM

INTEREST-RATE RISK

E i B d M

Maturity and Repricing Mismatch

The risk that “gaps” between maturity or repricing dates will affect earnings and value.

Yield-curve RiskBasis RiskEarnings-Based Measures

Economic-Based Measures

The risk to changes in the slope of the yield-curve

The risk that the correlations between rates upon which my assets and liabilities are based will move in a fashion which is detrimental to an organization’s earnings and value

Options Risk

The risk that embedded and explicit options present to an organization

Volatility?

Dimensions of Risk

Static Risk How do changes to risk factors impact my inherent risk position? EXAMPLE:

Economic value of equity, as traditionally applied, is a static risk measure Gap reporting, as traditionally applied, is a static risk measure

Dynamic Risk How do changes in risk factors – over time – impact my risk position –

over time? Note: risk factors are moving AND my risk position is moving This is equivalent to a holding period measure of risk where position

risk is likely changing relative the movement of the risk factors EXAMPLE:

Net interest income simulation is a dynamic measure of risk wherein the risk factors AND the position risk is changing over the holding period

Typical Model Data Flow

Balance sheet data and aggregation, repricing

Specific rate forecasts, future Interest Rate Risk

Loan

Systems

Investmnt

Systems

Other

Systems

General

Ledger

Deposit

System

Earnings gg g , p ginformation and other inputs and assumptions

,growth assumptions and other key inputs

Model

Beginning Balance Sheet

Month 1 Balance Sheet

Month 2 Balance Sheet

Etc. Ending Balance Sheet

Performance for Month 1

Performance for Month 2

Periodic Performance

Etc.

Simulations

Value

Simulations

3/20/2011

14

Interest Rate Risk Management Concepts

1. Repricing Gap

2. Net Income Simulation and Earnings at Risk (EaR)

3. Economic Value of Equity (EVE)3. Economic Value of Equity (EVE)

4. Scenario/Sensitivity Analysis

5. Data Integrity and Assumptions

6. Limits and Controls


Repricing Gap Analysis

Repricing Gap provides a simple, yet limited, view of re-pricing mismatches along the yield curve. Best Practices

Availability of re-pricing details on all balance sheet instruments

Significant granularity of re-pricing cash flow, e.g. 0-3, 3-6, 6-9, 9-12 months, 1, 2, 3, 4, 5, 6-10, 11-15, 16-20, 21-30 years

Multiple shock scenarios in order to observe mismatch volatility

Critical Assumptions Distribution of non-maturity deposit balances should be consistent with

FTP process; risk management and incentive comp should be aligned

Limitations Analysis is limited to current balance sheet; ignores future business

Inadequate for capturing impact of optionality, e.g. caps, floors, swaptions, convexity, etc.

Levered vs. Unlevered


Example: Repricing Gap Report


3/20/2011

15


1. Repricing Gap







Net Interest Income (NII) Simulation

NII simulation provides an estimate of near-term earnings sensitivity to changes in market interest rates. Intuitive and easily understood by ALCO and Exec. Management

Best Practices Dynamic balance sheet derived from LOB forecasts of future business

LOB ownership of key model assumptions

Sensitivity analysis to key model assumptions, e.g. deposit re-pricing, y y y p , g p p g,prepayment speeds, credit performance

Large number of rate scenarios including shocks, ramps, twists

Analysis of below-the-margin line items in order to achieve a NI or EPS sensitivity measure

Regular model validation and back-testing of near-term projections

Critical Assumptions Static versus dynamic balance sheet

Flat versus forward curve

Static (constant speeds) versus dynamic prepayment functions (ADCo)Sungard Account Management

Net Interest Income (NII) Simulation (cont’d)

Limitations Short term in nature; common practice is 1-2 year horizon

Misses impact of options that are beyond model horizon, e.g. mortgage prepay/extension risk, long-term FHLB optionality, TRuP options

Longer term, e.g. 5-7 years, horizons are recommended by regulators, but become very assumption driven

Ignores changes in value of MTM instruments (and potential earnings impact of impairment charges), e.g. trading book, MTM hedging portfolio

Deterministic vs Probability-based Yield Curve evolution Traditional methods rely upon single path rate scenarios

More current methods utilize stochastic yield curve models to derive a range of earnings estimates Provides more granular analysis of cash flow variability


3/20/2011

16

Example: Net Interest Income Report


Example: Net Interest Income Report



1. Repricing Gap







3/20/2011

17

Economic Value of Equity (EVE) Measures

EVE analysis provides an estimate of the sensitivity of the “economic value” of balance sheet equity to changes in market interest rates Captures all cash flows associated with current balance sheet

Better captures exposure to optionality risk

Inconsistent views across regulatory authorities FDIC prefers LT income simulation over EVE FDIC prefers LT income simulation over EVE

OCC somewhat ambivalent about EVE; have been known to give a “pass” to some LFIs on EVE in lieu of long-term income simulation

FRB more interested in EVE

12 CFR § 3.10 provides that national banks can be assessed higher minimum capital ratios based on significant exposures to declines in the economic value of its capital.

Proposed new accounting rules may make EVE considerably more important as it will translate directly into reported earnings volatility


Economic Value of Equity (EVE) Measures (cont’d)

Best Practices Stochastic interest rate model

Dynamic prepayment functionality

Two-factor yield curve models

Incorporates credit-spread volatility

Unique credit spreads for different asset classes

Limitations Does not translate/correlate directly to an intuitive earnings-based

measure of risk

Does not translate/correlate directly to market cap sensitivity

Does not indicate optimal hedging strategies

Ignores the impact of future business

All of the above lead to difficulty in limit setting

Requires significant computing power

Not all balance sheet instruments can be valued accurately with a single type model


Other Valuation Based Measures

DV01 Dollar value impact of 1 bp move in an interest rate

Common risk measure in fixed-income portfolio management

Not as common in balance sheet management

Key Rate Duration Measures the effect of a change in the yield curve that is localized at

a particular maturity point

Necessary for development of detailed hedging strategies

Mitigates risk to changes in curve shape

OAS In contrast to the simple “yield curve spread" measurement of bond

premium over a pre-determined cash-flow model, the OAS describes the market premium over a model including two types of volatility: variable interest rates and variable prepayment rates.

Sungard Account ManagementCONFIDENTIAL – FOR INTERNAL USE ONLY

3/20/2011

18

Example Reports: EVE and NII Scenario Dashboard


Example Reports: EVE



1. Repricing Gap







3/20/2011

19

Scenario/Sensitivity Analysis

In addition to establishing and monitoring standard risk limits for the impact of changes in market interest rates, sensitivity analysis to key model assumptions must be performed periodically, e.g. interest rate sensitivity and balance decay functions on non-maturity deposits Sensitivity analysis tests a model’s parameters without relating those

changes to an underlying event or real world outcome

Additional business scenarios, e.g. rapidly growing balance sheet, de-levering balance sheet, economic recession, corporate credit event, need to be periodically modeled Scenario analysis uses the model to predict a possible future

outcome given an event or series of events

Other Matters

The use of IRR models for balance sheet management, e.g. corporate forecasting, strategic analysis, M&A support, adds considerably to a firm’s credibility with regulatory authorities. If a model is used only for IRR management, the question arises as

to its credibility/veracity.

On January 6, 2010, the financial regulators issued an advisory to remind institutions of supervisory expectations regarding soundremind institutions of supervisory expectations regarding sound practices for managing interest rate risk (IRR). “In the current environment of historically low short-term interest

rates, it is important for institutions to have robust processes for measuring and, where necessary, mitigating their exposure to potential increases in interest rates.”

Regulators fear banks going out the curve (and funding short) in the quest for yield. Post credit-crisis, regulators do not want banks immediately increasing risk along another dimension.



1. Repricing Gap







3/20/2011

20

Data Issue Overview: The High-Altitude View

There has been a significant increase in the use of technology to increase efficiencies over the last two decades (you never “arrive”, but are always in the process of “arriving”) Banks have gotten bigger, risk-taking got more “scientific” (models)

Funding liquidity seemed endless (pre-crisis)

Products became more complex

Competition, globally, became more intense

Economies-of-scale seemed to “rule” and “diseconomies” of scale were “inconceivable” in banking

Our experience indicates: Multiple source systems

Legacy data problems

Inability to pull current position balance sheets together

Inability to evaluate, plan and budget over such a large book

Governance by volume and performance, not by risk

Key Assumptions: Income Simulation

1. Starting Balance Sheeta) Data preparation and maintenance is often > 60% of the challengeb) Some institutions are unable to compile complete starting position in a

timely basis (possible red flag)

2. Selection of rate scenarios and driver ratesa) +100, +200, +300, non-parallel, ramps, forward curve, etcb) How many driver rates? How to correlate? Manage?b) How many driver rates? How to correlate? Manage?

3. New business (e.g., roll-rates)a) How will new business “roll” onto books? When? With what

characteristics? How to source these assumptions? Maintain?

4. Repricing attributesa) Important for models not at instrument level

5. Core deposit behavior – earnings and valuation6. Prepayment assumptions

Key Assumptions: EVE

Discount rates utilized

Spreads used, if any Maybe 1) OAS, 2) credit, and/or 3) other

Core deposit average lives and valuation method(s) Core deposit average lives and valuation method(s) Are the same base runoff cash flows used for valuation and for

income simulation?

How are loans of similar type aggregated and evaluated?

How are embedded options valued? How do methods differ across the chart-of-account(s)?

3/20/2011

21

Data Integrity and Assumptions

Documentary support for derived assumptions used in the model: Core deposit assumptions – rate betas and lags (for NII) and decay

rates/durations (for EVE)

Deposit behaviors for risk management should be similar/identical as those for FTP

LOB sponsorship/ownership of pro forma inputs/estimates

Synchronization of pro forma prepayment speeds with historicalSynchronization of pro forma prepayment speeds with historical experience

Data limitations do exist and large, complex institutions will have to strike a balance between a quality process and overload. Sensitivity analysis should be used to test exposures resulting from poor data

quality, data aggregation as well as imperfect model assumptions, e.g. core deposit rate betas.


Data Integrity and Assumptions

"Best in Class" ALM process Frequency: Complete monthly reporting of all risk limits; constant analysis of

model assumptions, data quality and accuracy of output

Data aggregation: As little as possible w/o sacrificing compute efficiency For static scenario models, data granularity can be maximized while data

aggregation (pooling) can be minimized.

For stochastic models, data aggregation needs to be maximized in order to get compute results in a timely manner. Granularity, and hence accuracy, will be p y y, y,sacrificed to some extent. Period (quarterly or annual) analysis using more granular inputs can be used to ballpark errors associated with efficiency in regular process.

Regular LOB involvement in validation of model assumptions and results

Modeling effort that demonstrably drives corporate balance sheet management process



1. Repricing Gap







3/20/2011

22

Policies, Procedures and Limits

Policy comprehensiveness is critical Is it Board approved annually?

Overall policy objectives and IRRM philosophy are well articulated

Outlines major drivers of IRR and liquidity risk for the entity(ies)

Details responsibilities and authorities

Board

ALCO(s)( )

Management processes and procedures

Establishes risk appetite (how?) and tolerance

NII and EVE and exposure of each to major IRR sources (m,y,o,b)

Re-pricing GAP misses many risk factors and is not largely used

Exceptions to policy (process for escalation and remediation)

Relationship to capital, dividend, off-balance sheet, derivatives, investment and accounting policies

Has it been “benchmarked” against peer?


Limit Structures

Limits need to be developed based on a risk appetite analysis, or similar “threshold” exercise

Most large banks place limits to several baseline scenarios for dynamic income simulation NII (over 12 and 24 month horizons) and static Economic-Value of Equity (EVE) ) q y ( )analyses Not uncommon to have ramps, stair-steps or shock-based NII limits

EVE is by nature a shock-based (instantaneous) measure

January 2010 Interagency “advisory” made some suggested changes to the 1996 statement, but is only an advisory


Difference between limits and robust “analysis”

Measuring risk against policy limits is a 1st-order expectation; numerous additional scenarios need to be performed Net interest income

Exposure to yield curve, basis and options (volatility) risk

Exposure to spread risks

Economic value of equity

Exposure to credit spread (should be a MVPE baseline, not EVE)

Not uncommon for a separate model to be run for EVE analysis

Quant, R&D and market-data sourcing (prices and spreads) is a critical element (can have organizational design impacts)

Migration over time to a full-blown OAS method, but most banks are not at this stage yet


3/20/2011

23

Policies, Procedures and Limits

Policy comprehensiveness is critical Is it Board approved annually?

Overall policy objectives and IRRM philosophy are well articulated

Outlines major drivers of IRR and liquidity risk for the entity(ies)

Details responsibilities and authorities

Board

ALCO(s)( )

Management processes and procedures

Establishes risk appetite (how?) and tolerance

NII and EVE and exposure of each to major IRR sources (m,y,o,b)

Re-pricing GAP misses many risk factors and is not largely used

Exceptions to policy (process for escalation and remediation)

Relationship to capital, dividend, off-balance sheet, derivatives, investment and accounting policies

Has it been “benchmarked” against peer?


Limit Structures

Limits need to be developed based on a risk appetite analysis, or similar “threshold” exercise

Most large banks place limits to several baseline scenarios for dynamic income simulation NII (over 12 and 24 month horizons) and static Economic-Value of Equity (EVE) ) q y ( )analyses Not uncommon to have ramps, stair-steps or shock-based NII limits

EVE is by nature a shock-based measure

January 2010 Interagency “advisory” made some suggested changes to the 1996 statement, but is only an advisory


Internal Controls and Model Validation

There is an expectation that Internal or Corporate Audit Staff play a role in the IRR model review process In a decentralized model, various check-points and procedures will

need to be established and flow-through to the annual corporate audit report(ing)

The IRR element plays a significant role in the CAMELS rating process of the banks, and the RFI/C rating for the BHC/FHC.process of the banks, and the RFI/C rating for the BHC/FHC.

Model Validation Various policy statements: OCC 2000-16 and FHFA recent policies

are valuable reference points

Model validations needs to be conducted by quant personnel, independent from modeling crew. The validation needs to be coordinated with audit. Needs to be rigorous. Elements of the Jan 2010 statement need to be incorporated.


3/20/2011

24








Model Architecture: Account Level Detail

1. Ideally, you want to limit the number of accounts you model at the group or pool level and maximize the number of accounts you model at the transaction level.

2. Some accounts you may simply have to model at the


group or pool level using weighted average attributes

Model Architecture

• These constitute the “end-nodes” of your chart-of-account (COA) structure

• Varies from vendor to vendor• Such structure can allow for dynamic

dimensions

e.g., COA “end-nodes”Business Banking:London:CRE:Fixed:Business Banking:London:CRE:Libor3M:Business Banking:Hong Kong:C&I:Libor3MBusiness Banking:New York:Mtg:Treasury12M:

3/20/2011

25

Model Architecture

Model Architecture (continued)

Key Issues: Inputs

Identify model vendor and version number of data transformation engine; note specific modules or add-ons

Identify and assess ETL (extract, transform, and load) process, e.g. BDI, Access, Excel, SQL, etc.

Determine data sources, data owners, availability and control environment

Determine reconciliation process for product balances

Determine extent of data transformation and pooling logic

Select sample of records for testing (will follow through computational engine)p g ( g p g )

Review and test pooling logic

Review data filling routines

Review behavioral overrides in data

Review data record (from source system) granularity (accuracy vs. efficiency)

Review pool record (feed to computational engine) granularity (accuracy vs. efficiency)

Review USD and FX market spot rate and curve inputs (and consistency across models)

Review USD and FX internal spot rate and curve inputs (and consistency across models)

3/20/2011

26

Key Issues: Assumptions and Compute Engine

Identify and review key behavioral assumptions, e.g. prepayments and deposit decay rates; determine source, ownership, review and revision procedures

Review and test model rate generation processes, e.g. stochastic and forward curves

Review and document chart of accounts and key behavioral settings

Select sample of accounts for testing [sampling methodology TBD]

Verify cash flows for sample accounts, including principal and II/IE

Review new business chart of accounts, key behavioral settings

Determine assumption sources and validation methodologies for new business Determine assumption sources and validation methodologies for new business

Determine market value methodologies, e.g. static or stochastic

Review setup and calibration of stochastic rate model used for EAR or EVE calculations

Review EVE discounting (flat or with spread?). Basis for assumptions. Support.

Review key rate and DV01 methodologies

Outputs: Reporting

Review ALCO policy to identify risk limits

Review existing interest rate risk reports

Review GAP report construction, including balance sheet line-item inclusion/exclusion criteria

Review NIM forecast

Review analysis and measurement of mismatch basis yield Review analysis and measurement of mismatch, basis, yield curve, and options risk

Review funding/hedging recommendations

Review economic forecast and recommending balance sheet strategies

Determine if IRR reporting is appropriate given size and complexity of organization’s balance sheet


Governance: Internal Routine and Control

Review model operating procedures

Review back-testing of NIM and synchronization with budget or FP&A forecast

Review analytical reports/studies used to verify model results

Review stress testing

Review “what-if” scenarios

Review business line analysis of model results

Review disclosure of key behavioral assumptions, e.g. prepayments and deposit decay ratesdecay rates

Review model documentation for disclosures around model use and limitations

Review model documentation for discussions of model theory

Review change control procedures

Review recent audit findings for model/modeling process

Review model validation policy

Review previous model validation findings

Determine testing and approval process prior to model deployment


3/20/2011

27








Summary

Governance as much as the model Policies and Limits

Committees and active senior management and board oversight

Routine validations and accountability mechanisms in place

Break down silos. Use the model to drive business, not compliance

Robust stress-testing and sensitivity analysisg y y Non-parallel curve shifts critical in today’s environment

Duration and convexities are critical in today’s environment

Validation, Assumption management and Version control

Resources – does the bank have sufficient internal resources?

Income and value measures

Create a network of experts you can lean on when needed


Q&A?

Any questions or comment?


3/20/2011

28

SunGard BankingIRR Modeling and Management

[NAME]

THANK YOU [email protected]@dcprmiahttp://linkd.in/DC_PRMIA_LI (DC PRMIA LinkedIn Community)

82

Interest rate risk modeling day sun_gard_ambit banking

Economy & Finance

similar threshold

sungard account

balance sheet

current balance

board approved

sungard bankingirr

net interest

net income