Effective Risk Management with Business Rules and Scorecards
Daryl Pereira 270002AW8D firstname.lastname@example.org | | Tags:  impact09 business rules
0 Comments | 1,643 Visits
This presentation is co-hosted by:
Wee Seong Yeo of ILOG, an IBM Company
Wee Seong starts things off by explaining how enterprise risk management (ERM) has become a major buzzword, and that it's one of the areas where IT spending continues to rise Why? because fraud detection in banking could have prevented if the right risk detection systems had been in place, for example.
The issue is that businesses need to take risks to grow, but somehow these risks need to be managed. For financial institutions, this is a particular problem. When Standard & Poor analyzes financial institutions, they look at the following:
This presentation concentrates on credit risk.
Many financial institutions have all their systems silo-ed with retail banking databases and private banking databases, for example, with many analysts having to generate reports to deal with compliance across each of these systems.
If you can centralize the decisioning platform, you can build effective risk management on top of it.
(Sandeep Gupta from Equifax taks over the presentation.)
Equifax, which manages data on over 300 million consumers, 100 million businesses worldwide and 169 million employee files (salary, duration, role), employs 6,900 employees globally. They operate across multiple industries, including telecom, retail, healthcare and financial services. When a customer interfaces with any of these industries (such as signing up for a new phone line with a telco), Equifax will offer a credit score rating on that person.
Sandeep discusses the technological challenges of ERM. If the rules used to make credit decisions are not transparent and available to business users, then there are problems around compliance. Another issue is if the dictionary in which the rules are expressed are not business-user friendly. If there is an inconsistency of rules, automation will not be complete and a large part of your business process management is manual (expensive, and prone to innacuracy).
Equifax uses ILOG BRMS within their InterConnect solution because:
Equifax pulls information from data sources as wide as Experian, TransUnion, PPS, ID Analytics, TALX, Telecheck, as well as from their own data. All this information needs to be accessed and supplied in a consistent form.
ILOG JRules is used to orchestrate flow processing by deciding data sources to access at the appropriate place during processing. The entire data (rules) dictionary can be expressed in a language familiar to business users. Change management is also simplified with ILOG JRules. Combined with IBM Cognos technology, Equifax can deliver powerful analytical capabilities to combine business rules analysis with data analysis.
They also use scorecards for heavy statistical analysis, such as when evaluating the 2,500 characteristics that exist within an individual's credit file. Using ILOG JRules, Equifax has developed their own statistical grammar which can be exposed to business users who don't need to interact with the underlying code. The rules can then be amended by the customers themselves. For instance, if a bad weather forecast comes out for Black Friday (the massive retail day after Thanksgiving), retailers may offer special credit lines to attract shoppers. The rules governing this credit offer can be issued in a matter of hours.
Scorecards have been a big problem for Equifax due to their complexity. For instance, they allow Equifax to calculate the chances a business will become bankrupt. The models were originally coded in assembly language. To help avoid the possibility of mistakes, the models were coded twice, but the whole process has been simplified by using ILOG JRules.
For more information on Equifax and ILOG see the following session from the Dialog user conference:
For more information on ILOG, an IBM Company's business rule products try the BRMS Resource Center.