Whitepaper Add Advanced Analytics to Your Artillery to Combat Insurance Fraud Presented on: Jan 2015 Author : Madhur Virmani Asso. GM, Practice-H&I Email Id : MadhurV@hexaware.com Hexaware Technologies. All rights reserved.
Table of Contents 1. Fraud Analytics, a must-have for Insurers 03 2. Driving Loss Reduction with Hexaware s Insurance Fraud Management Solution (ifraud) 04 3. Hexaware s ifraud delivers measureable business outcomes 08 About Hexaware Hexaware is a leading global provider of IT & BPO services and consulting. The Company focuses on key domains such as Banking, Financial Services, Insurance, Travel, Transportation, Logistics, Life Sciences and Healthcare. Our business philosophy, "Your Success is Our Focus," is demonstrated through the success we ensure for our clients. Hexaware focuses on delivering business results and leveraging technology solutions by specializing in Business Intelligence & Analytics, Enterprise Applications, Quality Assurance and Testing, Remote Infrastructure Management Services and Legacy Modernization. Founded in 1990, Hexaware has a well-established global delivery model armed with proven proprietary tools and methodologies, skilled human capital and SEI CMMI-Level 5 certification. For additional information please visit. Insurance Services Hexaware offers a complete range of IT-BPO-Infrastructure services to Life, P&C and Health Insurance, Reinsurance organizations and Brokers worldwide. With 1300+ techno-functional industry certified consultants, Hexaware offers end-to-end solutions and services across the insurance value chain. Our endeavor is to proactively invest in researching and developing advanced solutions to address the emerging business problems. This translates into shortened implementation cycles and creates a toolset that is enhanced continuously to meet new and emerging challenges. For additional information please visit. Author s Bio Madhur Virmani is a Sr. Business Consultant in Insurance Solution Group at Hexaware. Madhur has around 13 years of experience in the Insurance & IT industry across P&C, Life and Health insurance and reinsurance domain with large Insurance and IT organizations. He is an engineering graduate, a Fellow member of Insurance Institute of India and an Associate member of Chartered Insurance Institute London. He carries a rich International experience in Consulting, Business Analysis and designing IT solutions for leading global insurers. He has been a visiting faculty at National Insurance Academy and an examiner at Insurance Institute of India. Madhur can be reached at madhurv@hexaware.com. Hexaware Technologies. All rights reserved. 2
1. Fraud Analytics, a must-have for Insurers Early fraud detection is a key challenge for global insurance organizations. Fraudulent Claims analysts are unable to claims and underwriting leaks pose dual risks to business, eating up margins as well as affecting competitiveness. A recent Insurance Consumer Fraud survey by Accenture flag the right claim at the right found that more than 68 percent of respondents commit fraud because they believe they time to the investigator can get away with it. More disturbing was that 12 percent of adults in the U.S. agreed it is OK to submit claims for items that are not lost or damaged, or for personal injuries that didn t occur. Such attitudes cost the insurance industry billions of dollars each year. Chief Claims Officer, According to the Insurance Information Institute, property and casualty (P&C) insurance AAA Michigan fraud strips an estimated $30 billion from the industry each year losses that must be made up in premiums. The National Insurance Crime Bureau (NICB) estimates that fraud is involved in approximately 10 percent of losses, costing policyholders an estimated $200-$300 a year in additional premiums. Therefore, it is important for insurers to investigate fraud, and have tools and techniques in place to identify fraud early in the process lifecycle and prevent losses that otherwise will create additional financial burdens. Fraudsters have devised multiple ways to compensate their annual premiums and policy deductibles. Examples are: a. Exaggerated estimate of loss b. Inflated value for damaged property c. Car repair estimate including pre-existing damages/escalated repair costs d. Adjuster s report for replacement of damaged car parts that could have been repaired Insurers face a significant challenge to control their claims costs, a major portion of which can be controlled by detecting fraudulent claims early in the life-cycle and avoiding the payment of fraudulent claims. Insurers should consider the possibility that approximately 10 percent of claims reported may be fraudulent. The impact is enormous if all of these claims get paid. This in turn weakens the insurer s financial position and its ability to underwrite large profitable risks, and affects its solvency margins. It also tremendously undermines the insurer s ability to offer competitive rates to its policyholders, which means a higher premium for all. The above challenges present insurers with an opportunity to plug some of the leakages in order to reduce losses and improve margins. Most of the P&C insurers are done with the implementation of a claims management system and hence, at this juncture, the digitization wave which has unleashed the power of analytics has come as a boon to insurers. Insurance companies are turning to fraud analytics solutions to churn structured and unstructured data with a view to improving the fraud detection volumes and quality. Hexaware Technologies. All rights reserved. 3
2. Driving Loss Reduction with Hexaware s Insurance Fraud Management Solution (ifraud) In spite of the industry taking up various steps to combat fraud, few have been successful. The key reason is the possible impact of fraud detection solutions on customer relationships and company reputation. Slow claim payments will directly hurt SLAs and customer satisfaction levels. Moreover, in this new socially connected world, a single false positive is sufficient to bring about long term loss of brand image. There is thus a challenge for insurers to strike a balance between fraud detection and claims handling demands. Moreover, this need to be achieved within today s reality of cost and resource constraints. Hexaware s ifraud offers a comprehensive and streamlined approach to fraud detection which guarantees to increase the detection rate as well as the accuracy. The solution leverages advanced analytics to analyze perpetrators as well as the claims themselves and is helping insurers worldwide to efficiently and accurately detect more fraudulent activities. Hexaware s ifraud uses 10 layers of analytics to detect and prevent fraud across the insurance functional areas via an integrated approach of consolidating information from internal and external, structured and unstructured sources. Each layer produces a meaningful result, however, when combined, increases the quality of results and reduces the probability of false positives. Insurers are able to increase the rate of fraudulent claim detection, identify underwriting irregularity and stop claims and policy leakages. The solution is technology and business-systems agnostic, and seamlessly integrates with core insurance applications independent of the technology platform. It works throughout the claims lifecycle and raises alerts on a real time basis. It can also remotely accept data files and provides the necessary information back in any desired format. It is important to cross - reference claims data with internal policy data, history of claimant, social security number, bankruptcies, judgments, AKAs (false or alternative identities), credit history (for first party claims). I am not aware of anyone doing this effectively Former Claims Director, Nationwide Insurance 10 Layers of Analytics to Detect Highly Complex Frauds Increasing Complexity of Fraud Feature Predictive Analytics Anomaly Detection / Business Rules Text Mining Social Network Analyzer Link Analysis Example Fraud Detected Ditching Past Posting Misrepresentation Staged accidents Vehicle Smuggling Phantom Vehicles VIN switch Suspicious claims from a particular repair shop Fraud in new policies or policies about to expire False location or timing of accident (social media) Inflated damages (various reports) Fraud rings Connections to known fraudster Tweets that do not match with the claim information Confirming true identify of a person Identifying relationships between participants in multiple claims Hexaware Technologies. All rights reserved. 4
Increasing Complexity of Fraud Feature Industry Database Lookup Voice Analysis Telematics Prefill data update detection Image Analysis Example Fraud Detected Fraud rings Duplicate claims Voice recognition Speech modulation Driving behaviour detection Event monitoring Over writing of prefilled information on online portals Multiple changes to description of information Damaged part analysis Under development Claim Intimation Claim Verification Adjudication Settlement Payment Recovery Customer / Prospect / Claim Participant PAS / Claims Application / Online Portal / CRM First Notice of Loss details / Policy Information / Proposal Information Twitter, Facebook (Tweets, links) Structured Data Pattern Detection Business Rules Social Network Analysis Text Mining Industry Database Look-up i ifraud Engine Flagging & Alert Generation Fraud Prevention during / post adjudication of claim Claims Payment Hold Claims Payment Executive / SIU Billing Fraud Voice Reports from Police, Adjuster, Investigator Vehicle driving history, event information Social Network Rules Link Analysis Voice Analysis Telematics Prefill Data Update Detection Investigate Prevention of Underwriting Irregularities Policy Premium Image Analysis Underwriter / CRM / SIU Billing Hexaware Technologies. All rights reserved. 5
How ifraud works: 10 layers of analytics ifraud provides insurers with 10 layers of analytics to uncover hidden complex fraud patterns and rings. The various techniques provide investigators with tools to investigate in the right direction. 1 2 3 4 5 6 7 8 9 10 Predictive model Custom data driven predictive analytics model based on a variety of parameters developed by a team of experienced actuaries, statisticians, data scientists and domain experts. Determines claims which matches known fraudulent patterns and also highlights newer patterns Business rules and anomaly detection - Variety of pre-configured business rules with ability for business users to update and modify rules. Identify anomalies in a large group of similar cases eg., claims cost values having multiple standard deviations outside of the norm of similar type of accidents Text mining - Ability to mine information from unstructured sources such as adjuster reports, police reports and social portals. The engine provides not only an ability to perform keyword search in documents but also an ability to compare approved values of damaged part cost and labor cost with standard customary values set in the claims system or as agreed with a preferred and approved body repair shop Social network analysis - View and analyze relationships and connections of various participants of the claim on social portals as well as with known fraudsters. Investigators can find relationships between a suspicious claimant and other individuals who stand a chance to benefit from the claims, such as insurer s employees or third party business professionals such as independent adjusters, agents and body repair shops Link analysis - Uncover links of participants under various roles in multiple claims. It also enables insurers visually analyze matches and links between disparate individuals sharing the same personal attribute information which is an indication of a fraud ring. This can determine if an individual is using multiple, various versions of personal attribute information Database look-up - This shall look at the internal and external databases such as an internal Special Investigation Unit (SIU) fraudster list or an industry database such as ISO / LexisNexis. It can analyze personal attributes such as name, street address, state/city/zip, date of birth, telephone number, email id, social security number, employer etc to identify likely matches Voice analysis - Employs multiple speech analytics techniques such as phonetic indexing, voice to text transcription to identify claimant emotions, key phrases as well as other call behaviors that are indicative of possible fraudulent activity Telematics - The engine consumes the telematics data to identify underwriting irregularities and suspicious claims. It can analyze the driving behavior based on combination attributes such as daily miles travelled, usual location of travel, and duration of travel, frequency of hard stops / harsh braking, quick bursts of acceleration, sudden cornering, and jackrabbit start etc Prefill data update detection - The engine identifies modifications to the prefilled underwriting and claims information from any internal or external database on a customer, agency portal. It can analyze modification to information such as vehicle details, zip, driving and claims history etc Image analysis - This gives insurers the ability to analyze images related to the claim event for potential anomalies or deviant patterns to identify if the images are in line with the accident description or if they are tampered to falsely represent enhanced damages ifraud assigns an individual fraud score to each of the above techniques. In addition it also assigns an overall score to the claim representing the likelihood of it being a fraud. It also provides a cause analysis to the user for the case being marked as fraud. Some of the other features of ifraud that enable investigators during their decision making process is depicted below: Hexaware Technologies. All rights reserved. 6
Alert management & routing Detection pre, post adjudication & post payment of claims Web & mobile enabled Fraud scoring & confidence rating Dashboard and reports with filtering & sorting Highlighting key information Batch and real time detection Part cost & labor rate comparison FRAUD Dashboard visualization with filtering and sorting This gives the claims manager and investigator a graphical representation of fraudulent claims analysis. The charts and graphs get updated dynamically and can be filtered based on selection criteria such as time period, geography, cause of loss, fraud patterns, claims executive, producer etc. The user can slice and dice the information to perform a deep analysis. The dashboards are role based and can be customized based on users choice. Hexaware Technologies. All rights reserved. 7
Alerts and notifications: It displays the suspicious claims along with the cause for which these claims were flagged. The alerts can be sorted based on the parameters such as Loss date, FNOL date, Reserve amount etc. These alerts can be assigned to investigator s based on business rules and can also be fed back to the claims system. Once the investigation is complete and the claim is not to be paid for being fraudulent, a notification can be sent to other departments such as accounts for non-payment and to underwriting to avoid accepting further policies from the same policy holder. In case the claimant has more than one policy, the investigator can notify the stakeholders of other lines of business area Access through web and hand held device: The solution can be accessed by claims managers and investigators on web and through any hand held device. This ensures that the users have an access to flagged claim from anywhere and at anytime 3. Hexaware s ifraud delivers measureable business outcomes With the use of Hexaware s Insurance fraud management engine, insurers can expect specific benefits, including: Prevention of fraudulent claim settlements resulting in reduced fraudulent claims cost payments by 5 to 10 percent Improved fraudulent claim detection rate by 2 folds with real time and early detection in the process lifecycle Reduce claims cost by initiating recovery for already paid claims Enhanced adjuster / investigator efficiency with improved quality of referral and avoidance of reporting false positives Consistent, error-free selection of profitable individual risks by preventing acceptance of cases with underwriting irregularities Reduction in claims and underwriting leaks to improve profit margins, enabling insurers to adequately price their products for customers A saving of $50~75 per claim for a typical P&C insurer resulting in a ROI of 10:1 with the use of ifraud After the initial cost, I consider 6:1 or 7:1 as reasonable Chief Claims Officer, AAA Michigan For more information To learn more about Hexaware Insurance fraud management solution, please contact corporatemarketing@hexaware.com or visit 1095 Cranbury South River Road, Suite 10, Jamesburg, NJ 08831. Main: 609-409-6950 Fax: 609-409-6910 Disclaimer Contents of this whitepaper are the exclusive property of Hexaware Technologies and may not be reproduced in any form without the prior written consent of Hexaware Technologies. Hexaware Technologies. All rights reserved. 8