classification ratemaking

Imagine charging every driver in the country the same price for car insurance. The careful 50-year-old in a quiet town would be paying for the risk of a 17-year-old in a fast car in a crash-prone city — and would soon flee to any insurer offering a fairer deal. The careless and the cautious would never sit happily in one price. Classification ratemaking is the practice of sorting risks into groups that are alike, so that each group pays a price reflecting its own expected cost.

While the overall rate level decides how much total premium to collect, classification ratemaking decides how to split that total among customers. It identifies rating variables (age, location, vehicle type, prior claims, building construction, and so on), measures how much each one affects expected loss, and assigns each class a relativity that raises or lowers its price relative to a base. For instance, if young drivers cost on average 1.8 times the base class, their relativity is 1.80; if a low-risk territory costs 0.85 times the base, its relativity is 0.85. A given customer's price multiplies the relevant relativities onto the base rate.

Done well, classification keeps a book stable: each group pays roughly its own way, so low-risk customers stay rather than subsidize high-risk ones and quietly leave (the adverse-selection spiral). Done badly — or constrained by law — and prices become unfair or unstable. This is also where modern pricing has changed most: simple one-variable analyses have given way to generalized linear models that estimate many variables' effects simultaneously, and where regulation forbids certain variables (such as some uses of credit, gender, or postcode), actuaries must classify within those bounds while keeping rates 'not unfairly discriminatory'.