The long-term care (LTC) insurance industry has shown significant interest in the use of predictive analytics to develop more accurate projection assumptions for the purposes of reserving, rating, and valuation. This article first summarizes how these techniques avoid some of the disadvantages of traditional actual-to-expected (A:E) experience studies. It then dives into new model visualization approaches used to shed light on the “black-box” nature of advanced modeling techniques. These visualizations help better highlight and understand the key drivers of predictions so that they can be understood and validated by all stakeholders. Using the tools discussed, actuaries can allow the machines to do the heavy lifting, leaving more time to review, interpret, and apply actuarial judgment to the results.
This article was published by the Society of Actuaries.