Equifax builds highly sophisticated consumer risk scoring solutions

Combining multiple risk scores across large, diverse data-sets to help clients predict customer behavior

Published on 27-Sep-2012

"The combination of IBM SPSS software and our unique fusion method helps our clients assess customer risk in a much more accurate and sophisticated way, which generates real value for their businesses." - Martin O’Connor, Senior Vice President for Analytics, Equifax

Customer:
Equifax

Industry:
Financial Markets, Professional Services

Deployment country:
United States

Solution:
BA - Business Analytics, BA - Predictive Analytics, Smarter Planet

Smarter Planet:
Smarter Risk Management

Overview

Equifax is a global leader in consumer and commercial information solutions, providing businesses of all sizes and consumers with information they can trust. It organizes and assimilates data on more than 500 million consumers and 81 million businesses worldwide, and uses advanced analytics and proprietary technology to create and deliver customized insights that enrich both the performance of businesses and the lives of consumers.

Business need:
Equifax saw an opportunity to provide its clients in the telecommunications, utilities and financial services industries with even more accurate insight into consumer behavior by utilizing multiple data sources and combining multiple scoring models into a single risk score.

Solution:
Using statistical tools incorporated in IBM® SPSS® Modeler software, Equifax applied a newly developed proprietary technique for “ensembling” multiple risk models into a single score. The technique, which Equifax calls “fusion”, weights and combines the significant dimensions of different risk models into a single score that clients can use to make better decisions about their customer relationships.

Results:
In one example, gaining a better understanding of risk enabled one of Equifax’s clients to project an increase of five to ten percent in its total number of customer accounts and annual revenue, which translates to a benefit of tens of millions of dollars. Simplifies the development of new scoring models for individual clients; now that the fusion technique has been documented, a new ensemble model can be deployed in just four to six weeks.

Benefits:
Provides a more nuanced risk model than traditional credit scores, and serves as a better predictor of consumer behavior. Enables Equifax to integrate multiple data sources into its scoring models, providing a more rounded analysis of consumers’ habits. Makes it easy for clients to adopt the new scoring model: the new scores can simply be imported into existing decision management systems.

Case Study

Instrumented: Data on various aspects of consumer behavior is collected from clients’ systems and Equifax data sources.

Interconnected: The data is fed into a sophisticated statistical model that produces a risk score, which is then seamlessly integrated into clients’ systems and processes.

Intelligent: By assessing multiple types of risk in a single score, the solution provides a more rounded analysis of consumer behavior, and serves as a more effective guide to decision-making.

Equifax is a global leader in consumer and commercial information solutions, providing businesses of all sizes and consumers with information they can trust. It organizes and assimilates data on more than 500 million consumers and 81 million businesses worldwide, and uses advanced analytics and proprietary technology to create and deliver customized insights that enrich both the performance of businesses and the lives of consumers.

Headquartered in Atlanta, Equifax operates or has investments in 18 countries, including the US, Canada and the UK, and is a member of Standard & Poor’s (S&P) 500 Index. Its common stock is traded on the New York Stock Exchange (NYSE) under the symbol EFX. For more information, please visit www.equifax.com.

Understanding decisioning models

The data that Equifax uses to create credit scores comes from a variety of sources, including information from its clients’ systems and numerous Equifax data sources. For consumer decisioning, for example, Equifax may combine data on credit-related behavior such as mortgage and loan repayments; incomes; wealth and assets; demographics; utility, pay-TV and telecommunications bill payments. The traditional approach is to feed this data into a statistical model that assesses the most significant factors and produces a score – a single number which the clients of Equifax can use to assess the risk of offering products or services to any individual consumer.

“Our clients expect us to provide the most accurate possible assessment of their customers’ and prospects’ credit-related behavior by incorporating our vast, diverse data assets,” explains Martin O’Connor, Senior Vice President for Analytics at Equifax.

For example, in the US, Equifax manages a database that contains payment history information for more than 290 million consumers, sourced from more than 60 telecoms, pay-TV and utilities companies. The contributors who provide the data want to leverage it to identify and manage financial risks, and to monitor customer relationships. Equifax makes a constant effort to develop new services that use this data to provide deeper and more accurate insights.

Developing a new method

“A couple of years ago we began looking at developing a new method for using data from different sources, and generating a new type of credit score,” comments Thomas Aliff, Vice President of Analytics for Equifax’s US operations. “Traditional credit scores tend to focus on just one aspect of credit risk: the likelihood that an individual or business will fulfill its payment obligations. However, in reality, credit risk is much more multifaceted – factors such as the affordability of credit and the risk of bankruptcy can also be important factors in credit assessments.”

Martin O’Connor adds: “Although we could create multiple scores for all the different aspects, the challenge was to find a way to combine them into a single ‘ensemble’ score that our clients would be able to import into their decision management systems and use to assess their customers. Traditional ensembling methods do not address the issue of incorporating multiple behaviors within one score, so we decided to develop a new method.”

Choosing the right tools

Equifax has been using IBM SPSS Modeler to create credit scoring models for many years, and decided to use the software to help develop the new method.

“The regression algorithms and other mathematical and statistical techniques that SPSS provides are very powerful, and we realized that they could provide a significant head start for us,” says Martin O’Connor. “We were excited by the idea of using an existing tool to solve a wholly new problem, instead of having to purchase or develop new software.”

Creating the power of fusion

A team of expert researchers spent six months developing the new ensembling technique, which Equifax calls “fusion”. In layman’s terms, the fusion approach enables Equifax to assign different weights to each individual score depending on their relevance, and then combine them to create a single overall score. This new approach utilizes traditional statistical building blocks to generate a new adaptive learning system that leverages the data, assesses the most significant factors, and produces a score. For example, the weighting will be different depending on the purpose of the score – for example, if a telecoms company is assessing the risk of offering a free handset to a new customer, it will probably want to take into account the customer’s prior behavior in paying their telephone bills; but it may also be relevant to consider their payment behavior with utilities bills. The fusion method allows greater weight to be placed on the telephone-related behavior, while still allowing the utilities-related behavior to influence the final score.

Thomas Aliff comments: “The fusion method was developed by a team of Ph.D. researchers, and it’s extremely sophisticated; but that doesn’t mean it’s difficult to use in practice. SPSS Modeler makes it easy to apply, so if a client comes to us and asks us to develop a new credit score that meets their specific business needs, we can develop a new model in about four to six weeks.”

Taking advantage of more accurate risk analysis

In tests, the fusion-based scores have shown themselves to be more accurate as a predictor of customer behavior than traditional scores, and a number of Equifax’s clients are already seeing benefits from integrating the scores into their decision management processes.

To take one example, a pay-TV provider wanted to perform a risk assessment of potential customers to determine which products and services to offer them. Using the fusion-based score, the company discovered that a large proportion of these customers had a much lower risk of not meeting their payment obligations than previous scores had indicated, while only a relatively small number saw their risk score increase. This meant that the company was able to make more attractive offers to a larger number of potential customers – projecting an increase of five to ten percent in its total number of customer accounts and annual revenue, which translates to a benefit of tens of millions of dollars.

Similar solutions have also been deployed for other telecoms providers in the mobile and multi-channel sectors, and are delivering similar benefits.

Looking to the future of financial services

In the financial services sector, Equifax has created a scoring system for a bank’s credit card division which uses three separate models to create scores for credit risk, affordability risk and bankruptcy risk, and then uses the fusion method to combine them into a single score.

“Fusion really provides a leading-edge solution for financial services companies,” comments Thomas Aliff. “They have focused on credit risk in the past, but it’s increasingly clear that these other dimensions can also be very significant over a long-term customer relationship. We expect to see more and more banks and other lenders adopting a more rounded approach to consumer risk analysis as they start to acknowledge this.”

Martin O’Connor concludes: “The combination of IBM SPSS software and our unique fusion method helps our clients assess customer risk in a much more accurate and sophisticated way, which generates real value for their businesses. This capability helps to differentiate Equifax from the competition and maintain our leadership position among global credit bureaus.”

About IBM Business Analytics

IBM Business Analytics software delivers actionable insights decision-makers need to achieve better business performance. IBM offers a comprehensive, unified portfolio of business intelligence, predictive and advanced analytics, financial performance and strategy management, governance, risk and compliance and analytic applications.

With IBM software, companies can spot trends, patterns and anomalies, compare “what if” scenarios, predict potential threats and opportunities, identify and manage key business risks and plan, budget and forecast resources. With these deep analytic capabilities our customers around the world can better understand, anticipate and shape business outcomes.

For more information

For further information please visit ibm.com/business-analytics.

Request a call

To request a call or to ask a question, go to ibm.com/business-analytics/contactus. An IBM representative will respond to your inquiry within two business days.

Products and services used

IBM products and services that were used in this case study.

Software:
SPSS Modeler, SPSS Modeler

Legal Information

© Copyright IBM Corporation 2012. IBM Corporation, Route 100, Somers, NY 10589. Produced in the United States of America. September 2012. IBM, the IBM logo, ibm.com, Let’s Build A Smarter Planet, Smarter Planet, the planet icons and SPSS are trademarks of International Business Machines Corp., registered in many jurisdictions worldwide. Other product and service names might be trademarks of IBM or other companies. A current list of IBM trademarks is available on the Web at “Copyright and trademark information” at www.ibm.com/legal/copytrade.shtml. This document is current as of the initial date of publication and may be changed by IBM at any time. Not all offerings are available in every country in which IBM operates. The client examples cited are presented for illustrative purposes only. Actual performance results may vary depending on specific configurations and operating conditions. It is the user’s responsibility to evaluate and verify the operation of any other products or programs with IBM products and programs. THE INFORMATION IN THIS DOCUMENT IS PROVIDED “AS IS” WITHOUT ANY WARRANTY, EXPRESS OR IMPLIED, INCLUDING WITHOUT ANY WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND ANY WARRANTY OR CONDITION OF NON-INFRINGEMENT. IBM products are warranted according to the terms and conditions of the agreements under which they are provided. The client is responsible for ensuring compliance with laws and regulations applicable to it. IBM does not provide legal advice or represent or warrant that its services or products will ensure that the client is in compliance with any law or regulation.