Claims Leakage – Plugging the Hole with Predictive Modeling

Part 1 of 4
By Ronald T. Kuehn FCAS, MAAA, CPCU, ARM, FCA, Kim Piersol FCAS, MAAA, & Todd Dashoff ACAS, MAAA, ARM of Huggins Actuarial Services, Inc.
Advanced use of predictive modeling has altered the traditional approach as to when and how claim leakage is measured. One of the main benefits of relying on these models is earlier recognition of potential adverse/favorable development in the expected cost of claims. Earlier recognition can result in loss mitigation processes applied to claims with potential claim leakage before that leakage occurs. Early application of mitigation strategies could reasonably allow for capture and recovery of 25%-50% of adverse development that would have occurred without the use of predictive modeling techniques.
The traditional approach to measuring claim leakage has typically involved review of closed claim files. The approach usually involved identifying a proper claim size sample that was representative of the entire claims population. Often, a questionnaire was then completed for each sample claim that captured specific characteristics about the claim. The results of the questionnaire were then reviewed in detail to identify overall leakage issues as well as specific issues that may contribute to the overall leakage. The issue(s) that resulted in the most leakage then became the subject(s) of revisions to the claims handling process that were communicated to all claims handlers, in order to reduce or eliminate the effect of those issues on leakage in future claims.
Predictive modeling is now being used to better identify claim characteristics that may indicate the potential for adverse development on the ultimate cost of a claim. Rather than relying solely on closed claim data, predictive modeling can be used at the various stages of a claim to identify potential leakage. Some models examine claims at stages as claim occurrence, claim reporting, adjustment (development) period and settlement. Other issues that can be addressed using these models include claim assignment, coverage reviews (primary and excess) claim investigation, negotiations, use of litigation and recoveries.
Another benefit of using predictive modeling in identifying cost drivers is their use in the underwriting process. The identification of cost drivers can then be used in the processes to screen potential new insureds or to produce better pricing on an existing book of business.
The predictive modeling process begins with data collection. As with the more traditional methods, a questionnaire can be used to gather the necessary data. There are vendors who specialize in the gathering and compilation of data needed for use in creating a predictive model. The software developed by these vendors can be made available online. Once the necessary data is compiled on an individual claim basis, a database is created upon which a predictive model is based. As with any model, the volume of data is key to the predictive power of the model. Predictive models can also make use of external data to supplement the claim database. Some examples of external data that can be introduced based on the location of a claim are medical care costs, use of the court systems and income levels of potential claimants. The decision to incorporate external data in the predictive model should take into account both the benefit to be gained as well as the cost of acquiring such information.
The predictive model can be used to help project the future development at various stages of a claim, as well as to identify cost drivers that are correlated with the future development. For example, the development can be measured from the initial reporting of a claim through 12 or 24 months or other increments of time that may start from other than the initial claim report. By identifying factors that are correlated with claims that later exhibit excessive increases, the company can more quickly identify the claims that may potentially exhibit such increases, and monitor the handling of those claims so that the increases are mitigated or do not occur at all. Typically, large claims are excluded from the predictive modeling process, as the potential for development of these claims is often known at the initial reporting of the claim.
The ability to better predict future development and the types of claims where such development is found allows claim adjusters to prioritize case loads based on claims that have the identified characteristics. Loss mitigation processes can then be created to reduce the potential for adverse development. As the claims that have been identified as the source of potential leakage are settled, they can be reviewed to determine whether the characteristics identified by the model actually resulted in savings and the magnitude of the savings on each claim can be measured. This allows for analysis of the accuracy of the model and can suggest areas for revision, either in the selection of the factors used to predict leakage or the need for additional factors to be reviewed, either internal or external. It is important to note that the process is continuous; management must make sure that any processes identified for change are communicated to all affected personnel, as well as making sure that any new causes of leakage are identified as quickly as possible.
In Part 2 of this article we will discuss examples of predictive modeling usage in claims leakage studies in order to mitigate adverse loss development that would have otherwise occurred.
Ronald T. (Rusty) Kuehn is a Fellow of the Casualty Actuarial Society and the Conference of Actuaries in Public Practice, a Member of the American Academy of Actuaries and the International Actuarial Association. In addition, Mr. Kuehn holds the Chartered Property-casualty Underwriter (CPCU) designation and the Associate in Risk Management (ARM) designation.
Mr. Kuehn is a consulting actuary with Huggins Actuarial Services, Inc. In his consulting practice he specializes in medical malpractice, private passenger automobile, workers’ compensation and commercial lines coverage working for insurance carriers, self-insured healthcare systems, self-insured corporations, brokers, municipal bond and mortgage insurance experience, and other types of clients.
During the 33 years prior to joining Huggins Actuarial Services, Inc. Mr. Kuehn was a Partner with Ernst and Young, LLP and a partner of the Hay Group. Ernst & Young, LLP had acquired the insurance actuarial practice of the Hay Group (Huggins Financial Services, Inc.). Mr. Kuehn retired from the partnership of Ernst & Young on October 31, 2003, acquired his practice and personnel from Ernst & Young and joined Huggins Actuarial Services, Inc. on October 31, 2003.
Prior to joining Ernst & Young, LLP and the Hay Group, Mr. Kuehn worked with two major multiple line property-casualty insurance companies. He is experienced in all property-casualty lines of business and distribution systems and has sat on various industry committees.
Mr. Kuehn is professionally active, having served on the Examination Committee and the Committee for Consultants' Interests of the Casualty Actuarial Society, and as past President, Education Chairman and Board Member of the Casualty Actuaries of the Mid-Atlantic Region (CAMAR). In addition, Mr. Kuehn has served on the Insurance Services Office Private Passenger Automobile Subcommittee and the Homeowners Subcommittee. Mr. Kuehn has also served as a Member of the Conference Task Force on Public Policy Debate of the Conference of Actuaries in Public Practice. Mr. Kuehn currently serves on the Board of the Insurance Society of Philadelphia (ISOP), on the Casualty Practice Council of the American Academy of Actuaries and is the chairman of the Casualty Loss Reserve Seminar (CLRS) which is jointly sponsored by the Casualty Actuarial Society (CAS), the American Academy of Actuaries and the Conference of Consulting Actuaries.
Kim E. Piersol is a Fellow of the Casualty Actuarial Society and a Member of the American Academy of Actuaries. Prior to joining Huggins, he served as Senior Vice President & Chief Actuary for Crum & Forester Insurance Companies. Prior to that he served as chief actuary for CNA Insurance Companies, and was a consulting actuary for KPMG LLP, Arthur Anderson LLP, and CFO of AIG Risk Management. He has over thirty-nine years of experience in the actuarial field.
Mr. Piersol has served on the CAS exam committee and participated in presentations at CAS seminars. He has served on the American Academy of Actuaries Environmental Liabilities Work Group, the NAIC Technical Advisory Committee on Catastrophe Reserves, the Workers’ Compensation Reinsurance Bureau Actuarial Committee, and been a director & treasurer of the Professional Liability Underwriting Society (PLUS). He also served on the Professional Liability Actuarial Subcommittee of ISO. Mr. Piersol achieved his Bachelors of Science in Mathematics at Furman University, Greenville, SC. He has also served as a 1st Lieutenant in the United States Army Field Artillery.
Todd Dashoff is an Associate of the Casualty Actuarial Society, a Member of the American Academy of Actuaries and an Associate in Risk Management. He has served as a Manager in the Actuarial Services practice of a big four accounting firm. Prior to that, he was an actuary for a major property-casualty insurance company located in the Mid-Atlantic region of the United States. Mr. Dashoff has over 33 years of experience in the actuarial field. Contact Todd at 610.892.1826 or

Meet The Experts

    Enter name of Insurance Company and press GO button.