credit risk case study forrester

Upload: marcelo-rojas

Post on 10-Apr-2018

215 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/8/2019 Credit Risk Case Study Forrester

    1/12

    Making Leaders Successul Every Day

    Jaar 5, 2009

    Case Sd: Hp Rea EsaeEabes Cred Rsk PressasWh Bsess Resb Mke Gaer

    r Appca Deepme & Prgram Maageme Pressas

    http://www.forrester.com/
  • 8/8/2019 Credit Risk Case Study Forrester

    2/12

    2009, Forrester Research, Inc. All rights reserved. Unauthorized reproduction is strictly prohibited. Inormation is based on best availableresources. Opinions refect judgment at the time and are subject to change. Forrester, Technographics, Forrester Wave, RoleView, TechRadar,and Total Economic Impact are trademarks o Forrester Research, Inc. All other trademarks are the property o their respective companies. Topurchase reprints o this document, please [email protected]. For additional inormation, go to www.orrester.com.

    Fr Appca Deepme & Prgram Maageme Pressas

    ExECutivE SuMMARy

    Uncertainty in lending regulatory requirements is putting pressure on credit risk modelers to become

    more responsive. Munich-based Hypo Real Estate Group (HRE), an international banking specialist

    or commercial real estate markets and the public sector, implemented a new credit risk management

    approach. Te result: Application development proessionals are no longer a bottleneck, and credit

    risk modelers can respond more quickly to changing regulatory requirements. HREs experiencesuggests that business rules platorms will help many nancial services rms be more responsive and

    more compliant in todays uncertain regulatory environment. Application development proessionals

    can learn rom HREs best practices, which are to: 1) provide risk modelers with a rules-authoring

    environment and 2) create a ormal, collaborative rules lie cycle.

    tABlE oF Cont EntSThe Bet And The Wort O Time For Credit

    Ri Modeler

    Bet Practice: HRE Proided Qant With A

    Rle Athoring Enironment

    Bet Practice: HRE Created A Formal,

    Collaboratie Rle Lie Ccle

    Bet Practice Relt: HRE Achiee Greater

    Agilit In Credit Rating

    WHAt it MEAnS

    The Beneft O Bine Rle Are Not

    Limited To Credit Ri Modeling

    notES & RESouRCES

    Frreser erewed Dad Kag, head Hp

    Rea Esaes cred rsk aacs eam.

    Related Reearch Docmentthe trh Ab Bsess Res Agrhms

    J 24, 2008

    the Frreser Wae: Bsess Res Parms,

    Q2 2008

    Apr 8, 2008

    Jaar 5, 2009

    Case Sd: Hp Rea Esae Eabes Cred RskPressas Wh Bsess Res

    Achee Greaer Ag, trasparec, Ad Cmpace i A uceraRegar Erme

    b Mie Galtieri

    wh Jh R. Rmer, Chares Bre, Mke Gp, ad Was y

    2

    6

    7

    7

    9

    mailto:[email protected]://www.forrester.com/http://www.forrester.com/go?docid=45582&src=47738pdfhttp://www.forrester.com/go?docid=39088&src=47738pdfhttp://www.forrester.com/go?docid=39088&src=47738pdfhttp://www.forrester.com/go?docid=39088&src=47738pdfhttp://www.forrester.com/go?docid=39088&src=47738pdfhttp://www.forrester.com/go?docid=45582&src=47738pdfhttp://www.forrester.com/http://www.forrester.com/mailto:[email protected]
  • 8/8/2019 Credit Risk Case Study Forrester

    3/12

    2009, Frreser Research, ic. Reprdc PrhbedJaar 5, 2009

    Rsk Pressas Wh Bsess Res

    Fr Appca Deepme & Prgram Maageme Pressas

    2

    THE BEsT AND THE WORsT OF TIMEs FOR CREDIT RIsk MODELERs

    Dont blame the credit risk modelers (sometimes called quants) or the current nancial-industry

    meltdown. Te credit risk models that they create are only eective i dealmakers use them, and

    many did not. It is the best o times and the worst o times or quants who are responsible or

    creating the methods and models or rating the risk o huge commercial real estate and public

    inrastructure nancing deals:

    Te good news or quants: It is the best o times or quants because having excellent credit riskmodels is more important than ever in this challenging economic environment.

    Te bad news or quants: It is the worst o times or quants because the pressure is on to bemore responsive to increasing regulatory requirements, corporate risk policies, and new ways o

    rating credit risk.

    HRE I A Microcom O The Problem And show The Wa To A Potential soltion

    One rm subject to these pressures is Hypo Real Estate (HRE) o Munich, Germany.1 HRE nances

    international, commercial real estate projects and, because o its October 2007 acquisition o

    Dublin-based DEPFA Bank (DEPFA), now also nances the public sector and inrastructure

    projects such as roads and bridges throughout the world.2

    David Kang heads HREs credit risk analytics team and has a sta o ve credit risk modelers who

    are responsible or creating and maintaining about 30 models used to rate incoming and existing

    deals. Tese are not the devilishly complex risk models o the investmentcredit risk modelers whom

    Warren Buet described as geeks bearing ormulas.3 Rather, HRE dealmakers use Mr. Kangs

    teams models to determine risk ratings or individual deals. Te more complex models can have as

    many as 40 input actors, while simpler models may have only 10 input actors. Each models output

    is always in the orm o a risk rating such as A+ or AA. A deals risk rating has a huge impact on

    whether or not HRE closes the deal and on determining the terms o the deal.

    HRE Had Two Credit Ri Approache A Third Wa Waiting In The Wing

    Acquiring DEPFA gave HRE the opportunity to create a single, standard approach to credit

    risk modeling or both lines o business.4 Beore proceeding, HRE was able to compare the two

    companies systems:

    HRE deployed models as Java Web applications. HREs previous approach to credit riskmodeling was or quants to create and test each model using Excel and then work with

    application developers to build a Web-based Java application to implement the model. Tis

    traditional approach is widespread at other banks. Although this approach worked, it required

    quants to create methodologies and models in Excel and then communicate those requirements

    to application developers who sometimes had diculty understanding models because o their

  • 8/8/2019 Credit Risk Case Study Forrester

    4/12

    2009, Frreser Research, ic. Reprdc Prhbed Jaar 5, 2009

    Rsk Pressas Wh Bsess Res

    Fr Appca Deepme & Prgram Maageme Pressas

    3

    complexity. Producing new models or making changes to existing models could take as long

    as six months: Developers had to code the models in Java, and quants had to retest them once

    developers had implemented them in the Java application.

    DEPFA deployed models in Excel. Mr. Kangs due diligence o DEPFAs approach to credit riskmodels revealed that DEPFA used Excel spreadsheets not only to develop the models but also to

    deploy them. According to Mr. Kang, it is not uncommon or quants to use Excel to create and

    test models, but using Excel or deployment is unacceptable because German regulators would

    not accept Excel as a proper I environment.

    Te previous HRE approach lacked agility but would meet regulatory requirements which

    DEPFAs more agile approach would not. What HRE needed was the best o each approach: the

    exibility or credit risk modelers to create and test their own models plus the robustness o a solid

    I environment (see Figure 1).

    A possible solution arose in a technology DEPFA had been considering prior to the acquisition:

    DEPFA had planned to replace Excel with Visual Rules, a business rules platorm oered by

    Innovations Sofware echnology.5 Mr. Kangs initial examination o the capabilities o Visual Rules

    led him to an unexpected revelation: I it worked, the business rules approach would not only satisy

    the German regulators requirements; it would also give his group o risk modelers the ability to

    create, maintain, and test the credit risk models directly using the Visual Rules authoring tool.

    I was initially skeptical when Mr. Kang approached them with his ndings; I voiced concerns

    about risk modelers ability to program the models in the Visual Rules environment and about

    who would control deployment to production. But afer seeing a demo and working out a processwhereby I controls deployment to production, I decided to go along with the new business

    rules approach because it enabled the credit risk modelers while also keeping controls in place or

    deployment.

    HRE Replaced Excel And Jaa With Bine Rle

    In April 2008, HRE made the decision to move orward with Visual Rules and started a project to

    implement the existing models o DEPFA in business rules and to run them in parallel against the

    existing Excel spreadsheets. According to Mr. Kang, the new platorm works well: I think a lot

    o banks dont know that something like this exists or dont trust it. In reality, using this business

    rules platorm is very exible and easy. Te new business rules approach satised regulators, risk

    modelers, dealmakers, and I:

    Regulators are satised because the models are auditable. When HRE acquired DEPFA, it wasclear that the German regulators would not accept the use o Excel as a credit risk management

    tool. Te new approach has a well-dened, auditable process or creating models and deploying

    them on I-supported inrastructure. Te German regulators did not consider emailing

    multiple versions o Excel spreadsheets around to be a proper I environment.

  • 8/8/2019 Credit Risk Case Study Forrester

    5/12

  • 8/8/2019 Credit Risk Case Study Forrester

    6/12

    2009, Frreser Research, ic. Reprdc Prhbed Jaar 5, 2009

    Rsk Pressas Wh Bsess Res

    Fr Appca Deepme & Prgram Maageme Pressas

    5

    David Kang is happy because his team can now change direction quickly when necessary.When the regulators come knocking to discuss credit risk models, Mr. Kangs group must

    respond. Te business rules solution gives Mr. Kang and his team o quants the ability to

    respond more quickly to changing regulatory requirements and company policies.

    Figre 2 Cred Rsk Mdeers use the Ahrg t t Creae, Maa, Ad tes Mdes

    Source: Forrester Research, Inc.47738

  • 8/8/2019 Credit Risk Case Study Forrester

    7/12

    2009, Frreser Research, ic. Reprdc PrhbedJaar 5, 2009

    Rsk Pressas Wh Bsess Res

    Fr Appca Deepme & Prgram Maageme Pressas

    6

    Figre 3 Deamakers use the Cred Rsk Rag Web Appca t Ge A Rsk Rag

    Source: Forrester Research, Inc.47738

    Input: quantitative variables

    Output: Credit rating

    BEsT PRACTICE: HRE PROvIDED QuANTs WITH A RuLEs AuTHORING ENvIRONMENT

    An important benet o implementing a business rules platorm is that the risk modelers can use

    the business rule authoring tools to develop, change, and test the credit risk models. As Mr. Kang

    explains, We are the model owners, and we know what models to implement.

    Business rules platorms provide authoring exibility by oering one or more rule authoring

    metaphors, which might include i-then-else statements, owcharts, decision trees, and decisiontables.6 Tese tools enable business users such as HREs quants the ability to develop the rules

    themselves using Visual Rules. Visual Rules provided HRE with:

    A graphical authoring tool in which creating a model is akin to designing a fowchart.Quants are extremely analytical and have a deep knowledge o statistics and nancial

    mathematics. Tat kind o knowledge ofen translates well to the ability to express logic. HREs

    risk modelers were able to recreate some o their models in the new tool within two weeks.

  • 8/8/2019 Credit Risk Case Study Forrester

    8/12

    2009, Frreser Research, ic. Reprdc Prhbed Jaar 5, 2009

    Rsk Pressas Wh Bsess Res

    Fr Appca Deepme & Prgram Maageme Pressas

    7

    Te ability to deploy models to HRE Web servers. Afer the models that the quants createhave been tested, I can deploy them to a production Web application. HRE I worked with

    Innovations Sofware echnologies to automatically generate a Web-based user interace or the

    models. I deploys nished applications on HRE servers, where credit risk analysts can access

    them and then use them to determine risk ratings or potential deals.

    BEsT PRACTICE: HRE CREATED A FORMAL, COLLABORATIvE RuLEs LIFE CyCLE

    Using or not using the right credit risk model can make or break a deal. It is essential that credit

    risk analysts use the most up-to-date credit models. Understanding this, HRE originally built its

    Java Web-based application to have a central and consistent way o developing, testing, deploying,

    and using the models. Business rules implementations require a lie-cycle management process that

    balances the quants authoring reedom with the release management processes needed to protect

    the application environment.

    Deploying bug-ridden rules is more likely in scenarios involving business users who are not amiliar

    with the careul processes that application developers put in place to manage changes and prevent

    disastrous deployments. o prevent deploying buggy rules while maintaining modelers agility, HRE

    created a rule lie-cycle process that:

    Does not encumber the risk modelers. Using the Visual Rules authoring tool, risk modelershave complete independence to create new models, experiment, and test. Tey do this with

    complete knowledge that when they nish their model, they will not then have to explain

    the complexities o the model to a developer beore it is implemented and makes its way into

    production.

    Enables I to retain control o deployment. HRE developed a clearly dened yet simpleprocess in which the credit risk department hands o new models to I or deployment. I likes

    that Visual Rules generates Java code and that I deploys the rules application as a JAR le just

    like any other Java application.7

    BEsT PRACTICE REsuLTs: HRE ACHIEvEs GREATER AGILITy IN CREDIT RATING

    HREs new business rules platorm provides much greater agility by cutting out the extra

    development time needed when involving I. Mr. Kang explained: With the business rulesenvironment, we dont have to involve I at all. We just do it ourselves. I I were to do it,

    developers would keep coming back to us because they dont have the background in modeling.

    Credit Ri Modeler Achiee Greater Agilit, Tranparenc, And Compliance

    Although HRE initially set out to nd the best-o-breed credit risk application, it obtained much

    more: Credit risk modelers can now respond aster to changing regulatory requirements; I is no

    longer the bottleneck. HRE achieved a credit risk application that:

  • 8/8/2019 Credit Risk Case Study Forrester

    9/12

    2009, Frreser Research, ic. Reprdc PrhbedJaar 5, 2009

    Rsk Pressas Wh Bsess Res

    Fr Appca Deepme & Prgram Maageme Pressas

    8

    Increases agility in developing, changing, testing, and deploying credit models. Enabling riskmodelers to develop and change models directly eliminates time-consuming steps such as risk

    modelers translating the model to requirements or an application developer as well as the many

    iterations o development and testing beore I can correctly implement the model.

    Maintains consistent use o the most current models throughout the organization. Excelis a nightmare when it comes to ensuring consistent use o models.HREs Web-based Javaapplication provides a centralized way o accessing the most current models, and it was

    important that HREs business rules implementation maintain that consistency. HREs

    implementation o Visual Rules succeeds in this goal by using the built user-interace generator

    to automatically create a Web-based Java application rom the models rules.

    Improves model transparency or analysis and regulatory compliance. When risk modelers

    explain new models or methods to application developers, some requirements are bound to getlost in translation. Tis method also makes it more dicult or regulators to know what is going

    on once I implements the models in Java or another programming language. Because the

    business rules platorm enables quants to develop rules directly, regulators can be more certain

    that the quant-created models are the ones running in the business.

    Application Deeloper stoc Goe up Becae The Are No Longer The Bottlenec

    Te implementation o the business rules platorm and especially the tool that enables risk

    modelers to program their own models took application developers out o the credit risk

    modeling business. Credit risk modelers can now change their own rules, avoiding the time-

    consuming step o translating the requirements to application developers who must then implement

    the model in Java.

    Although I no longer codes the models, it still plays a crucial role in ensuring that:

    Te models are deployed on the Web inrastructure. Giving I control o the productionenvironment helps ensure that the rules lie-cycle process is ollowed which prevents

    developers, whether rom the business or I, rom deploying models that have not been

    properly tested.

    Te quants have the support they need when using the new tool. I also has the role o

    supporting the tool and the rules runtime environment. Tis can involve installing or makingavailable sofware updates, ensuring that server versions are in sync with the tools, and elding

    user questions or routing them to the vendor.

  • 8/8/2019 Credit Risk Case Study Forrester

    10/12

  • 8/8/2019 Credit Risk Case Study Forrester

    11/12

    2009, Frreser Research, ic. Reprdc PrhbedJaar 5, 2009

    Rsk Pressas Wh Bsess Res

    Fr Appca Deepme & Prgram Maageme Pressas

    10

    8 A recent Forrester report gives advice on how to achieve success in implementing business rules that will

    empower your business users and in avoiding insidious pitalls. It recommends that organizations ollow

    these our best practices: 1) assemble a team that knows business rules implementation; 2) establish a

    methodology to nd your rules; 3) design with business users in mind; and 4) include business users inyour rules lie-cycle process. See the October 3, 2008, Best Practices In Implementing Business Rules

    report.

    http://www.forrester.com/go?docid=46339&src=47738pdfhttp://www.forrester.com/go?docid=46339&src=47738pdf
  • 8/8/2019 Credit Risk Case Study Forrester

    12/12

    Forrester Research, Inc. (Nasdaq: FORR)

    is an independent research company

    that provides pragmatic and orward-

    thinking advice to global leaders in

    business and technology. Forrester

    works with proessionals in 19 key roles

    at major companies providing

    proprietary research, consumer insight,

    consulting, events, and peer-to-peerexecutive programs. For more than 25

    years, Forrester has been making IT,

    marketing, and technology industry

    leaders successul every day. For more

    inormation, visit www.orrester.com.

    Australia

    Brazil

    Canada

    Denmark

    France

    Germany

    Hong Kong

    India

    Israel

    Japan

    Korea

    The Netherlands

    Switzerland

    United Kingdom

    United States

    Headquarters

    Forrester Research, Inc.

    400 Technology Square

    Cambridge, MA 02139 USA

    Tel: +1 617.613.6000

    Fax: +1 617.613.5000

    Email: [email protected]

    Nasdaq symbol: FORR

    www.orrester.com

    M a k g l e a d e r s S c c e s s E e r D a

    For a complete list of worldwide locations,

    visit www.forrester.com/about.

    Research and Sales Ofces

    4773

    For inormation on hard-copy or electronic reprints, please contact Client Support

    at +1 866.367.7378, +1 617.613.5730, or [email protected].

    We oer quantity discounts and special pricing or academic and nonproft institutions.

    mailto:[email protected]:[email protected]://www.forrester.com/