Skip to main content
main content, press tab to continue

How insurance actuaries can use machine learning to answer CEOs toughest questions

By Serhat Guven | September 2, 2021

Actuaries can use machine learning techniques to make the critical task of identifying changing patterns of claims development far easier.
Insurance Consulting and Technology
Insurer Solutions|InsurTech

In my experience, some of the most important questions insurance company CEOs ask their reserving actuaries are: “How much adverse development are we experiencing?” and “What is driving these results?”

Or roughly translated: “Are any parts of the business heading into trouble?”

Typically, actuaries have gone about answering those questions by analyzing the characteristics of claims data for patterns of adverse development. This process can be time consuming as it requires significant experience and general industry knowledge.

Flipping the problem

Modern analytics (oftentimes just referred to as machine learning) offers a smarter way. We flip the problem: We tell the machine what pattern we are looking for and let the algorithms search the data to uncover those patterns.

Specifically, rather than having an actuary searching for permutations of claims characteristics in swathes of data, you tell the machine what you are looking for — such as an emerging pattern, any outliers or any claims that share characteristics with past cases that have resulted in high costs — and set it off to do the searching.

This may sound counterintuitive, but the approach is relatively straightforward. There are specific considerations such as introducing location-based search functionality using sophisticated geographic smoothing routines, interpreting the results of the automated search and developing an action plan of what to do once you have found what you are seeking.

At the upcoming virtual Casualty Loss Reserve Seminar on September 13, we’ll be demonstrating how we’ve gone about doing just that on an illustrative data set in our Radar advanced analytics software. We will show how the analysis translates into identifying potential adverse development in business segments and geographies that could be addressed by an insurer. If you can’t make the seminar, titled “Smarter segmentation for the lazy insurance actuary,” please contact us for more information.

For hard-pressed reserving actuaries, machine learning could provide a smarter and quicker way to answer those key CEO questions that regularly put actuaries on the spot, such as claims drivers and where they may be brewing trouble for the business.


Global Proposition Leader, Pricing, Product, Claims and Underwriting, WTW
email Email

Contact us