PROCEEDINGS May 15, 16, 17, 18, 2005 - Casualty Actuarial Society

ARCHITECTURE FOR RESIDENTIAL PROPERTY INSURANCE RATEMAKING 539$100,000. 31 The actual deductible factor charged in rating, evenfor the hurricane peril, should depend upon the empirical TVIdistribution of the insurer’s book, and indeed the TVI of eachproperty. By design, this is not considered in the experimentaldata set.Rather than resolving “the” proper way to differentiate flatdollar hurricane deductible factors by TVI range, the study settleson an adjustment to a base scenario (that for the 0.5% deductible,which is equivalent a flat $500 deductible for the majority ofunits in the experimental data set). The implied relative loss costfor AOP perils by value range, shown on Exhibit 13, is the ratio ofthe complement of the loss elimination ratios in each range; thecalculation is analogous to formula (15), but relates TVI rangesrather than deductible amounts. Select a relativity, then apply itto the modeled 0.5% deductible factors by zone to produce $500flat deductible factors that vary by both TVI range and zone. Forexample:Low zone, under $75,000:(1 ¡ 25:0%)1:17 ¼(1 ¡ 20:4%) £ 1:23Medium zone, $225,000 and over:(1 ¡ 14:7%)1:26 ¼(1 ¡ 20:4%) £ 1:18and so on. The end result is a reasonable consideration of bothvalue insured and territory loss distributions in the pricing of hurricaneflat dollar deductibles. The calculation could be repeatedfor other flat deductible options.The deductible factors for other wind, where only flat dollardeductibles are offered, are calculated using exactly the sameprocedure and modeled scenario testing, except that factors are31 This is true assuming that the model contains a “static” event set which is applied toevery location. Some models build a “secondary uncertainty” randomization componentinto the analysis, which means the modeled losses for the same scenario on the sameevent set will still differ somewhat every time the model is run.