The degree of fineness with which a rate distinguishes risks, whose increase improves selection as long as the market follows and destroys pooling when taken to its limit.
A single rate for everyone pools perfectly and attracts bad risks, since good ones overpay and leave. Segmenting corrects that adverse selection, and this is the technical justification for the practice. But the logic taken to its end reverses: if every policyholder pays exactly their expected loss, there is no transfer between policyholders, hence no insurance in the economic sense, only prepayment. Between the two, each market sets a cursor, and it is set by three forces rather than one. Competition pushes toward segmentation, because a less granular insurer gets its good risks skimmed by a more granular one. Regulation restrains it, by banning certain criteria, sex in European insurance since the Test-Achats ruling, and by watching for disparate impact. And data availability moves the frontier of the possible, telematics and connected devices having made observable behaviors that were not. An insurer segmenting more finely than the market wins, until the day the whole market has caught up and only the riskiest segment is left with no offer.
The Test-Achats ruling of the Court of Justice of the European Union, dated March 1, 2011, banned the use of sex as a rating criterion from December 21, 2012, although it explained up to 15 frequency points among drivers under twenty-five. Insurers rebuilt part of the signal through correlated criteria, vehicle type and mileage, which illustrates how hard it is to suppress information rather than a field.
finesse tarifaire, rating segmentation, granularité du tarif, discrimination tarifaire technique