A Lesson in Assumptions: Shock Lapse Experience Report

Prior to RGA completing its first Post Level Term lapse and mortality study with the SOA in 2009, a wide range of assumptions was in use by actuaries with little industry-wide late-duration level term experience available to compare against. This first study analyzed shock lapse and mortality deterioration assumptions and experience by a variety of factors to provide fresh data and insights into these critical areas.

Five years later, RGA did a follow-up study with the SOA to update the findings with new data from the huge spike in level term sales at the end of 1999. This spike in US sales was due to new reserving regulations that went into effect on January 1, 2000. So, when the opportunity presented itself on a similar joint SOA and CIA study on Canadian level-term business, RGA was excited to participate in the project.

The plan was to do a similar report like those completed on the US business. The researchers would add new areas of investigation as well, including whether the premium jump ratio or premium jump amount is more predictive of the lapse rate.

Phase 1 of the study was completed in early 2020 and began with a request for companies to complete a survey about their assumptions on lapses and mortality, along with their premium jumps. These survey results were compiled into a report that allowed for companies to compare their own assumptions to others in the industry. Read the full report for more findings.

Phase 2 involved the summary of the actual lapse and mortality experience and was published on December 2, 2020. The published results revealed how the premium jump affects the lapse rates, as well as how the lapse rates affect the mortality deterioration. The report also shows comparisons between the actual results and the previously submitted assumptions.

The research team spends a lot of time reviewing the lapse skewness, or in other words, in what durational month are the lapses occurring. When looking at the 10-year term product, a large proportion of the lapses in duration 11 occurred in the first three months of the 11th year. If one were to isolate the lapse rates for durations 10 and 11, one could draw the conclusion that a fairly large number of policies pay that full year 11 premium and then lapse at the end of the year. What is illustrated however, is that the most of these policies are only paying one or two months of premium before lapsing. An outcome that could drastically affect the profitability if companies are assuming that full year 11 premium is being paid.

The report also provides a pivot table of the results with many aggregators to view the results as well as the results of the generalized linear model that can be used to help set lapse assumptions. Read the full report to learn more.

Travis Agne, one of the authors of the paper, says, “We feel really good about the results from this report. It should really help both pricing and valuation actuaries set their assumptions with more confidence.”

The authors would like to thank the SOA and CIA staff and volunteers who helped direct this project. This has been an exciting project, and we hope that it provides important information for the industry.