Exam 8: Class relativities need a common comparison basis
A raw segment loss ratio can reflect several correlated rating variables and different premium adequacy. Classification ratemaking must account for the intended rating structure. A univariate comparison is not automatically a new relativity.
Worked example or practice scenario
Suppose one vehicle class has more high-mileage drivers than another. Its higher observed claim frequency may partly reflect mileage mix. A multivariate model can separate effects under its specification, but confounding, sparse data and changed mix still need review.
Try this next
Compare raw summaries with an adjusted model result and explain the difference. Identify the benchmark category, normalisation and exposure basis. Check the prospective effect when inserting a relativity into the full rating algorithm.
Reading sources
ActNet editorial guide · October 1, 2026 · Original illustrative examples.