Predictive Analytics and Casualty Modeling
At the 2016 Advisen Casualty Insights Conference, an panel discussed how predictive analytics are being used in casualty lines of business including workers’ compensation, general liability, and D&O. The panel was:
- Mark Moitoso – Senior Vice President, Analytics Practice Leader – Lockton Companies (moderator)
- Vinny Armentano – Senior Vice President, Business Insurance Claims – Travelers
- Mark Brissman – National Predictive Analytics Practice Leader – AON Risk Solutions
- Hinson Han – SVP, Global Head of Business Solutionsi and Architecture – AIG
- Toby Unwin – Chief Innovation Officer – Premonition
Question: What does predictive analytics mean to you?
Answers:
- Taking an insane amount of data and using that data to show trends in the marketplace identify issues that can impact your business.
- On the claims side, this is used for two things. The first is claims triage which involves matching up the level of skill of the claims handler with the complexity of the claim. On the prescriptive side, it helps show which claims have potential for adverse development and would benefit from additional resources on the claim.
- Predictive analytics allow underwriters and claims handlers to sort through the data and assist in their decision making. It is not intended to replace decision making but to enhance it.
- Predictive analytics are best used to help provide answers to specific questions.
- You start with a hypothesis and use the data to try and validate that hypothesis.
Question: How can we show risk managers that analytics can have a positive impact on their program?
Answer:
- They can show that the claims predictive models do ultimately produce lower claims costs. However, the impact determined by the state of the program going into the predictive analytics. On programs that are running well, the impact won’t be as significant.
Question: Have we been able to validate that predictive analytics are producing better underwriting results?
Answers:
- These models are still evolving so the “proof” is not as strong as they would like. However they feel the trends are positive and that the models need to continue to be refined.