CLEAR (Canadian Loss Experience Automobile Rating)

The CLEAR methodology

• 1. Data normalization

  In order to accurately derive rate group differentials, the data are normalized (i.e., mathematically adjusted) to eliminate the non‐vehicle effects. In CLEAR, non‐vehicle effects are defined as the differences that exist between total premiums and those suggested by CLEAR rate groups. IBC is able to base CLEAR on data that has been adjusted to mathematically remove the effects of such non‐vehicle factors as territory, driving record and vehicle use.

• 2. Statistical estimates

  Using mathematical formulae that link vehicle characteristics with observed insurance claim experience, loss costs are derived for all vehicles. While CLEAR’s goal is to achieve a stable system of rate groups that credibly reflect actual claim experience, a sufficient amount of specific vehicle experience is not always available, especially for newer vehicles. In these cases, supplementary credibility is required to smooth the loss experience. In seeking supplementary credibility, a multiple-level approach is used.

• 3. Annual adjustments

  The estimated loss costs are adjusted each year to reflect the actual observed loss costs for each vehicle/model‐year combination.

• 4. Computation of initial relative loss costs and rate groups

  In order to obtain an initial relative loss cost for each vehicle, the adjusted loss costs are compared to the average loss cost for all vehicles. To ensure that year‐to‐year changes are implemented gradually, maximum year‐to‐year reductions and increases are capped. Based on the capped initial relative loss costs, vehicles are grouped into common rate groups. The process is performed separately for Collision, Direct Compensation – Property Damage (DCPD), Comprehensive and Accident Benefits coverages.

• 5. Final adjustments to rate groups

  Finally, the initial rate groups may be modified to control for rate level neutrality (i.e., any CLEAR table’s average differential will be equal to the average differential of the previous year’s table if both calculations are weighted by the current year’s projected exposures).