Case study

AIG Marketing, Inc. (AIGM)

Eliminating leakage in auto physical damage claims adjusting through advanced predictive modeling

Through closed file reviews for liability losses and field vehicle reinspections, the automobile insurance industry has long recognized chronic tendencies to overpayment. Whether from outright fraud or just the inherent complexity of case management, claims leakage has been estimated to cost auto insurers billions of dollars annually.

American International Group, Inc. (AIG) is the world’s leading US-based international insurance and financial services organization, the largest underwriter of commercial and industrial insurance in the United States, and among the top-ranked US life insurers. PA Consulting Group has been developing advanced predictive modeling and data mining solutions to aid the adjuster in determining loss severity and reducing claim overpayments. One problem that the insurer wanted to address was the tendency of overwrites in vehicle repair estimates. Random reinspections, while helpful, were not achieving the desired results; and attempts to target overwrite potential through more traditional predictive methods were not effective.

Through application of advanced Bayesian statistical analysis, PA’s methodology was rapidly able to develop and pilot a predictive model that greatly enhanced the targeting of high-probability overwrites for reinspection, and the appropriate training needed to reduce the incidence of errors for staff, independent adjusters and designated repair shops.

Preliminary results indicate that the model is producing predictive values that will greatly enhance the vehicle reinspection process, improve functionality through online application, and target the field behavior modification areas for the company to lower loss costs significantly through training and other process changes.