case study: rule-based technology in the mortgage domain (special focus on automated underwriting)

61
Business Rules Forum 2007 Gil Ronen Jeff Adler Case Study: Rule-Based Technologies in the Mortgage Domain (special focus on Automated Underwriting) [email protected] http://www.linkedin.com/in/gilronen

Upload: gil-ronen

Post on 18-Dec-2014

733 views

Category:

Technology


3 download

DESCRIPTION

Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting). Presentation at Business Rules Forum. Written with Jeff Adler.

TRANSCRIPT

Page 1: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

Business Rules Forum 2007

Gil Ronen

Jeff Adler

Case Study:

Rule-Based Technologies

in the Mortgage Domain

(special focus on Automated Underwriting)

[email protected] http://www.linkedin.com/in/gilronen

Page 2: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

2

Roadmap

Overview of Mortgage Value Chain and Life Cycle

Origination

Processing

Servicing

Lessons Learned

Questions and Answers

Mortgage industry deconstructed into a set of processes/tasks

represented using business rules then folded back into a

coherent structure across the mortgage value chain

Page 3: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

3

Sample Clients

Countrywide

Saxon

Wells Fargo

Chase

Aegis Mortgage

ACT Mortgage

GMAC

HomeBanc

NetBank

AmNet (Wachovia)

Ownit

Fifth Third Bank

First Greensboro

First Preference

Ford Motor Credit

Ocwen

Equifax

American Express

Complete Size and Product Spectrum

United Mortgage

Option One

First Horizon

PHH

UBS

Realty Mortgage

Paul Financial

Taylor Bean Whitaker

Citibank

Page 4: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

Mortgage Value Chain

Mortgage Life Cycle

Page 5: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

5

Mortgage Lifecycle

Page 6: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

6

Mortgage Origination Timeline

Page 7: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

7

Key Accounting Dates

Page 8: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

8

Value and Risk across the Lifecycle

Best - Fit Deal Structuring

Mortgage Suitability

Risk - Based

Pricing

Cross - Sell

Origination

Borrower Credit

Risk

Pipeline Risk Management

Fraud Detection

Processing

Loss Mitigation

Servicing

Page 9: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

9

The Secondary Mortgage Market

Depository

Institutions

Non-

Depository

Institutions

Hold in Portfolio

Sell Loan Investor

Secondary

Market

Conduit

Sell Whole Loans

Pool Loans into

Mortgage Backed

Securities

Issue bonds

backed by LoansSell Bonds

Sell MBS

Agencies

Pension Funds

Life Insurance Companies

Commercial Banks

Thrifts

FNMA

FNMA

FHLMC

GNMA

Private Investment Banks

Commercial Banks

Mortgage Banks

Mortgage Brokers

Loan

Originators

Page 10: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

Origination

Page 11: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

11

Optimize Borrower Experience

During Sales and Origination Lenders have one shot at capturing a

customer.

Moving knowledge about products & UW criteria to POS guarantees

quick turnaround time, accuracy and information at critical juncture.

Business rules define the offerings, facilitate the interaction.

Right PriceRight

Product(s)

Customer

Profile Preferences

Assets and Debts

Credit

Market

Conditions Risk-based pricing

Profitability

Credit

Point of

Contact

Touch Points

Business Channels

Interface Mode

Page 12: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

12

Adding Value to Customer Interaction

Interact with customers consistently and profitably through all delivery channels and touch points

Identify the best products to meet borrower needs and objectives (e.g. “Build Equity”, “Lowest Payment”)

Offer a competitive price point and pricing options

Determine conditions needed to process and close the loan

Identify opportunities to cross-sell products that would be of interest/benefit to the borrower

Business Rules Applications

Manage products and pricing across touch-points

Deal Structuring – Rules can describe strategies to configure a loan similar to the best Loan Officers.

Mortgage Suitability

Cross-sell rules

Page 13: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

13

Alternative Loan Configurations

Requested Loan

Purchase $ 540 , 000

Loan Amount $ 440 , 000

LTV 82 %

Goal Lowest Payment

Alternative

Configurations

75 - 7 - 18 ( $ 415 / 25 K )

80 - 0 - 20 ( No MI )

85 - 0 - 15

80 - 10 - 10

IO No 82 % with IO = “Y”

Combo No

LTV 90 %

Page 14: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

14

Bringing Intelligence to the Sales process

Control product offerings Low down payment loans – objective-based

Low Income borrowers – FHA loans

Relationship customers – bank accounts, HELOCs

Special properties – construction loans, co-ops

Strategies for creating deals and offering features Mortgage Insurance

Interest Only

Combo loans

Cross-sell opportunities Home equity seconds

Life Insurance

Mortgage Protection insurance

Page 15: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

15

Alternate Scenarios

Alt. Scenario Description When Created

Interest Only (IO)

Interest Only Objective = Minimize Monthly Payment AND IO

= No

Non-IO Create scenario with non Interest Only

Objective != Minimize Monthly Payment AND IO

= Yes

LTV/COMBO Change LTV to 80-20 If requested LTV [85-100]

Loan Amount Modify Loan Amount Based on price hits [60K, 500K]

Stated Doc Change doc type to Stated Primary Borrower is self-employed, Full Doc

requested

Page 16: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

16

Alternative Deal Offerings

Page 17: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

17

Best Fit – Minimize Monthly Payment

Strategies for structuring a

deal with goal to minimize

monthly payment

Represents different lines of

reasoning

Minimize

Monthly

Payment

Minimize

Rate

Increase Down

Payment

Lower Loan

Amount

Select Variable

Rate Product

Increase Points

Purchase:

Do not Finance

Closing CostsPay Closing

CostsRefinance:

Do not roll-in

Closing Costs

Select Fixed Rate

Productwith

Longer Term

Page 18: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

18

Power to the People

Page 19: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

19

Mortgage Suitability

Suitability is often defined as “a subjective evaluation by the lender whether a product is best for that borrower”

Suitability is a hot topic in a market where foreclosures are spiking due to customers being given loans that they were eligible for but that they cannot repay due to volatility of rates, insufficient income, etc.

Inversely, innovation in the market has allowed for the proliferation of exotic products that are suitable under certain circumstances:

Pay Option ARMs allow flexible payments and are suitable for customers who are paid seasonally or are professional investors – they are not for average salaried customers with low sophistication.

The solution is not to reduce flexibility of product offerings but to look at Suitability.

Lack of borrower sophistication and poor risk management in the secondary markets are key factors behind the suitability movement

Page 20: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

20

Business Rules and Suitability

Introduce a new class of Suitability Rules to measure the degree of suitability of a deal offering

Concepts to include

Additional functionality to better communicate offerings of riskier products – explain the benefits and risks

Model borrower-centric what-if scenarios

Expanded modeling of borrower goals and objectives

Evaluate long and shorter-term needs

Risk assessment benefit messages and disclosures

Page 21: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

21

Cross-Sell

Become a sole-source financial service provider working for the best-interests of customers

Build customer relationships

Know customer needs and objectives

Identify cross-sell services that will help customers achieve their goals

Point-of-Sale: Cross-sell decision support tools support point-of-sale customer relationship building by identifying products that will service the customer.

Offering sensible solutions to customer needs can clinch a deal

Servicing Portfolio: Reach out to existing customers and expand their relationship with you (Home Equity, Credit Cards, Insurance…)

Page 22: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

22

Cross-Sell Decisioning

Enterprise Decisioning Platform

CRM

Data Base

Enterprise

Data

Business

Inquiry Business

Rule

Repository

Cross-Sell

Decisioning

Engine

Decision points often benefit from access to enterprise information

Page 23: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

Processing

Page 24: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

24

Processing Value and Risk

Borrower Credit Risk

Automated Underwriting – guidelines

Risk-Based Pricing – loan characteristic risk evaluation

Compensating factors – weighing positives and negatives

Pipeline Risk Management

Handling fallout – avoid losing loans before closing

Variation in interest rates - hedging

Fraud Detection

Reduce lender’s risk exposure to assist in loss mitigation

Income/Employment; identity theft; occupancy and property fraud

Page 25: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

Underwriting

Page 26: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

26

Automated Underwriting and Pricing

Business rules are most famously known for their use in

qualifying and pricing loans

These systems evoke the best that business rules have to

offer:

Reduce operational costs and streamline business process

Agility in face of rapidly changing market conditions

Real-time distribution across channels (guides, pricing, products)

Stability of rules throughout application time-line (application date)

and business process workflow (same rules for Pre Qual and UW)

Page 27: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

27

Product Guidelines

Purchase and Rate / Term Refinance

Property Type

Maximum

LTV

(w/o Sec.

Fin.)

Maximum

LTV

(w/ Sec.

Fin.)

Maximum

CLTV

(w/ Sec.

Fin.)

Maximum

HTLTV

(w/ Sec.

Fin.)

Credit

Score

1-2 Unit Primary Residence (O/O)

$417,000 to $650,000 90% 90% 95% 95% 620

95% 95% 95% 95% 680

80% 80% 100%1 100%1 680

$650,001 to 1,000,000 70% 70% 70% 70% 620

80% 80% 95% 95% 680

$1,000,001 to $1,500,000 75% 75% 80% 80% 700

$1,500,001 to $2,000,000 70% 70% 70% 70% 700

Page 28: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

28

Decision Processes

Strict Decisioning

IF Loan Credit Score < Guideline Credit Score

THEN Loan Application is Ineligible

Interpretive Decisioning

Difference Between Loan and Guideline Credit Score values

Greater than or equal to 0: Pass the Guideline

Less than 0 but greater than -10: Near-Miss – Not Sure what to do

Less than -10: Fail the Guideline

Loan Guideline Delta Result

700 680 20 Pass

675 680 -5 Near-Miss

640 680 -40 Fail

Page 29: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

29

Compensating Factors

Guidelines are not true measurement of composite

borrower risk. They are heuristics designed to provide

an understandable structure for assessing loan

applications. Invoke other areas of borrower strength to

override the minor transgression of a guideline.

Exceptions work well in the limited context when single

factors are “Near-miss”. Consider the case when 2 or more

factors are near-miss:

LTV = 71 (Max 70) and DTI = 46 (Max 45)

Example: LTV within 5% of Guideline, will accept if

borrower meets 2 of the following

Credit Score at least 20 points above guideline

Borrowers show 6 months reserves when 3 or fewer are needed

DTI is at least 10 points lower than guideline

No 30-day lates last 24 months

Page 30: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

30

Comprehensive Risk Assessment

Goal: Assess the strengths and weaknesses of the entire

case

How are CFs created?

Observe manual exception handling

Loan Performance data – not always seasoned, interest

environement changes

Policy – what investors will accept

Page 31: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

31

Risk Scoring

A holistic perspective on borrower risk would look to balance all

risk factors in a loan application and determine the level of risk.

Higher Risk - The degree to which a factor exceeds a guideline

Lower Risk – The degree to which a factor is within a guideline

A composite risk score can be invoked in place of

compensating factors to provide a better measure of risk

Page 32: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

32

Layered Decisioning

CF Analysis

Approved=Y?

Refer=N

Exceptions=N

AUS Decision =

Approved

Y

Ineligible=N?

AUS Decision =

Ineligible

Y

Refer=Y?

AUS Decision =

Refer

Y

N NScore Exceeds

ReferThreshold?

Y

N

AUS Decision =

Approved w/CF

N

Page 33: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

Pricing

Page 34: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

34

Pricing

Page 35: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

35

Price-Rate Relationship

0.25 0.5 1.0 1.5

0.5

Rate

1.5

2.0

(0.5,0.25)

(1, 0.5)

(1.5,0.88)

(2,1.25)

1

Price

0.5

1.0

(-1,- 0.5)

Buy up

Buy down

Rate Adj Price Adj

- 0.5 - 1.0

0 0

0.25 0.5

0.5 1

0.88 1.5

1.25 2

Page 36: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

Fraud Detection

Page 37: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

37

Types of Fraud

It is believed that 10-15% of all Loan Applications contain misrepresentations

There are 5 sets of data that are targets for fraudulent activity

Income/Employment: Inflated Income, Falsified Employment

Credit: False identity

Property: Falsified appraisal data, wrong property type

Occupancy: Intent to occupy, investor listed as primary residence

Assets: Falsified bank statements, gifts, concessions

Page 38: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

38

Objectives for Fraud

Falsification of data to purchase a property

Try to improve the 1003 data to ensure approval

Buy investment properties as Primary Residence to get better

terms

Misrepresentation of facts to profit from a mortgage

transaction

Flipping properties

Collusion among parties (Broker, appraiser, lender…)

Skimming Equity

Investor property with High LTV. Keep rent, do not make

payments

Page 39: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

39

Actual Case of Repeated Fraud

09/06 – John Jones takes a purchase primary loan (SISA) from

Lender for residence in Woodland Hills, CA

10/06 – Carol Jones (wife) takes a purchase primary (NIVA) loan from

Lender for residence in Pasadena, CA

01/07 – John Jones takes a purchase primary (SISA) loan from

Lender for residence in Walton, KY

01/07 – Carol Jones takes a purchase investment (NIVA) loan from

Lender for property in Florence, KY and states she owns other

investment props in CA and IL jointly with spouse (no reference on

his applications)

07/07 – John Jones applies for a purchase primary (SIVA) loan from

Lender for beach-front condo in Naples, FL

Page 40: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

40

Using Business Rules to Detect Fraud

Consistency of Data on Loan Application

Compare property state to employment or residency

Analysis of REO Schedule

Validating income levels

Comparison of new applications with CRM/Application

databases

Look for loans with same applicants with/without spouse

Look for trends in data

Use business rules to control workflow and integration with

third-party verification services

TALX – Income and Employment verification

Page 41: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

Servicing Risk and

Portfolio Management

Page 42: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

42

Loss Mitigation

Identify borrowers at risk of default and pursue appropriate

loss mitigation strategies designed to preserve

homeownership

Identify Loans within a servicer’s portfolio that may be at risk

for problems when their rates reset

Appropriate Strategies to Save Home Ownership

Loan Restructuring / Modification – ARM to fixed, 40 year term

Payment Deferral – tacking missed payments

Principal/Interest Rate Reduction

Other Strategies focus on helping borrower off load the

property prior to foreclosure

Page 43: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

Technology

Page 44: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

44

What Technologies Are Used?

Domain Ontology – Mortgage data is described more and more using MISMO

XML – Standard payload for system data exchange

Web Services – Standard protocols, service descriptions, …

Data Mining – Statistical regression, etc. to capture rules from unstructured data (e.g. actual manual exceptions, loan performance data)

Knowledge Acquisition/Business Analysis – Capture domain expertise (e.g. Sales, UW principles)

BPMS – Management of Business Process flow

Rules – Encapsulated business logic

Rule Engine – Encapsulated algorithm for handling Business Rules

BRMS – Business Rule Management

Usual Suspects: Systems Analysis, UML, DB...

Page 45: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

45

What is MISMO?

Mortgage Industry Standards Maintenance Organization.

It’s mission is to develop, promote, and maintain voluntary electronic commerce standards for the mortgage industry.

Established in 1999 by the Mortgage Bankers Association, MISMO encourages participation from all sectors of the industry.

MISMO defines xml structures to coordinate transfer of information across the mortgage value chain. For example: Requesting an underwriting decision from an automated AUS

Ordering Title from a title company

Sharing loan closing data

Ordering or re-certifying credit from a credit reporting agency

Transferring loan information to a mortgage servicing company

Obtaining mortgage insurance coverage and monthly PMI

Page 46: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

46

MISMO Influence

Page 47: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

47

MISMO Credit Liability

<CREDIT_LIABILITY

CreditLiabilityID="CRLiab0002" BorrowerID="Coborrower"

CreditFileID="CRFilEFX02 CRFilTUC03 CRFilXPN02"

_AccountIdentifier="541712485999999"

_AccountOpenedDate="1999-02“

_AccountOwnershipType="AuthorizedUser“

_AccountReportedDate="2002-05“

_AccountStatusDate="2001-11“ _AccountStatusType="Open“

_AccountType="Revolving“

_CreditLimitAmount="16200" _DerogatoryDataIndicator="Y“

_HighCreditAmount="16200" _LastActivityDate="2001-11“

_MonthlyPaymentAmount="232“ _MonthsReviewedCount="39“

_PastDueAmount="1773" _TermsDescription="$232/M“

_TermsSourceType="Provided" _UnpaidBalanceAmount="11637“

CreditBusinessType="Banking“ CreditLoanType="CreditCard">

</CREDIT_LIABILITY>

Page 48: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

48

Interoperability – Specification Initiatives

Page 49: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

49

Rule Engines

Rules provide business logic encapsulation

Rule engines encapsulate rule processing algorithms

Most use RETE therefore engines are becoming a

commodity (MindBox ArtEnterprise, Blaze Advisor, ILOG

JRules,…)

RETE is an algorithm providing asymptotic performance – as

the number of rules increases the execution time stays the

same

Page 50: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

50

Sample Business Logic – Gift Funds

Earnest Money and Gifts

If the LTV is 80% or less, 100% of the down payment can come

from a gift. This is only acceptable for those loans that are fully

or alternatively documented

If the LTV is 80% or greater, 5% of the down payment can come

from a gift. This is only acceptable for those loans that are fully

or alternatively documented

Page 51: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

51

Employment History

Borrower

Employer

Previous

Employer

Most recent

CurrentEmployment

YearsOnJob

CurrentEmployment

MonthsOnJob

CurrentEmployment

TimeInLine

OfWorkYears

EmploymentBorrower

SelfEmployed

Indicator

EmploymentCurrent

Indicator = YES

Check for 2

Continuous Years

with Current

Employer

Check Multiple

Consecutive

Employment records for

2 Continuous Years In

Same Line of Work

CurrentEmployment

YearsOnJob

CurrentEmployment

MonthsOnJob

CurrentEmployment

TimeInLine

OfWorkYears

EmploymentBorrower

SelfEmployed

Indicator

EmploymentCurrent

Indicator = NO

2 Consecutive

Years of Self

Employment

Page 52: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

52

Decision Tree vs. Decision Table

Employer

Relocation?

Distance >

35 Miles?

YES Current

WFHM

Customer?

No

Current WF

Customer?

No

YES

Code = CIG

YES

Code =

WFHM

No

Code =XX

Code = WF

YES

No

1 2 3 4 5 6 7

Employer Relocation

Moving Distance > 35 Miles YES

Current WFHM Customer YES YES

Current WF Customer YES NO NO YES NO

Agreement Code Assignment CIG WFHM WF XX WFHM WF XX

NONO

NO

Variable

RULE

NOYES

Page 53: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

53

Comp Factors Model for DTI

DTIDelta

-1.5

-1

-0.5

0

0.5

1

1.5

-30 -20 -10 0 10 20 30

DTIActual

0

0.2

0.4

0.6

0.8

1

1.2

0 20 40 60 80

DTI Score is based on:

DTI Delta – the difference between guideline and actual DTI

-Dynamic range from -20 to 20

-Values below 0 are positive / above 0 are negative

DTI Actual – the Debt-to-Income percentage

-Dynamic range from 20 to 50

-Values closer to 20 are positive / closer to 50 are neutral

Page 54: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

54

CF Scoring Models

Generates an overall score Indicates whether positive compensating factors outweigh any exception

violations

Maximizes the value of available historical data and incorporate industry best-practices May incorporate various scoring methods

Weighted-sum, neural network, fuzzy logic model, decision trees, etc.

-1.5

-1

-0.5

0

0.5

1

1.5

-30 -20 -10 0 10 20 30

-1.5

-1

-0.5

0

0.5

1

1.5

-40 -20 0 20 40 60

S

0

0.2

0.4

0.6

0.8

1

1.2

0 20 40 60 80

Page 55: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

55

Automated Decisioning Transactions

Pricing Only Transactions

Multiple deals (vary by price point) with

pricing, no stipulations

Best Fit Transactions

Multiple deals (vary by product) with

pricing, no stipulations

Underwriting Transactions

Single deal with stipulations, no

pricing

Stated 1003

Stated Credit

Full 1003

Stated Credit

Full 1003

Full Credit

UBS Conduit Lender Live/X2 Quick Qual Quick Price Clayton Everbank

UWFF

UWFS

POSS

BFFS

BFSS

BFFF POFF

POFS

Correspondent

Wholesale

Retail

In Production

Future

UWSS

Page 56: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

56

Mortgage Decisioning

Correspondent Customer Direct

Wholesale

Retail Affiliate

Other

Knowledge Repository

Pricing DB

Price/Rate Sheets

Price Determination/Database Price Feeds

Channels

Decisioning

Users

Rule Engine

POS/LOS/

Website/

Portal

Underwriter

Broker/Loan Officer

Borrower

Data Feeds

Due Diligence

Credit

Servicing

To support the various decisioning processes being called by

different channels, different interfaces, different users, using

multiple data feeds, at different points in the business

process the decisioning platform is created as a versatile

Service-Oriented Architecture

Page 57: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

Final Words

Page 58: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

58

ROI

Significant ROI for Business Rule Implementation

Reduced processing costs

Increased productivity

Automated workflow

Direct origination opportunities

More consistent decisioning across channels and business lines

Page 59: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

59

Automated Decisioning in Mortgage-ROI

Up Markets – Increased Productivity/Volume

Down Markets – Lower Costs/Higher Margins

Decisioning – Consistent, sophisticated and fast processing of large and often incomplete data or data of suboptimal quality

Best Fit – Mimic behavior of best Sales staff anywhere in process

Underwriting – moving UW to POS, guarantee deal for brokers and exit strategy for Originators

Secondary Markets – No cost for re-UW to any set of Guides

Comp Factors – replace expensive manual processing

Pricing – add any adjustor rule easily

Knowledge Hub – centralize decision-making and maintenance of knowledge

Page 60: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

60

Optimizing Value / Minimizing Risk

Across the value-chain: Understanding goals, values, pain

points, guiding principles, culture, standards, terminology

and finding common ground.

Loan Officers, Brokers, Underwriters, Processors, Due

Diligence officers, Secondary Markets, Traders, Securitizers

and Investors accessing the same Knowledge Base.

institutional centralization, industry standards, technology

standards, flexible and powerful architecture, centralized and

extensive/expressive knowledge repository.

Page 61: Case Study: Rule-based Technology in the Mortgage Domain (special focus on Automated Underwriting)

Questions & Answers

Contact:

[email protected]

http://www.linkedin.com/in/gilronen