Protected cell growth has proven to be advancing rapidly in all domiciles that have introduced it.Although no reliable statistics are available we estimate there to be more than 2000 cells operatingworldwide as smaller versions <strong>of</strong> property/casualty captives. It is referred to as the growth area for themature captive domiciles.Other new captive concepts to be considered are branch captives and special-purpose financialcaptives.Direct Sale<strong>Hong</strong> <strong>Kong</strong> should work to convince the China Insurance Regulatory Commission to allow <strong>Hong</strong> <strong>Kong</strong>captives to write business directly in China. Currently, corporations must purchase insurance fromlicensed insurance companies. If these corporations want to pass their risks onto its captive they needto use a fronting company and have risk ceded to the (reinsurance) captive either as a participant onthe fronting insurer's reinsurance program or passing through by way <strong>of</strong> retrocession. If the CIRCwould allow <strong>Hong</strong> <strong>Kong</strong> captives to directly write business in China, this would cut down administrativecosts and credit risks.China as a stepping stoneThe demand for alternative risk financing mechanisms is expected to grow rapidly with the utilization <strong>of</strong>risk management in Asia. According to a study by Aon Benfield, only a quarter <strong>of</strong> the Asian Global 500companies currently have a captive, and therefore it is foreseeable that demand for captives willemerge. Labuan and Micronesia are new captive domiciles trying to position themselves for the futuregrowth <strong>of</strong> Asian captives. <strong>Hong</strong> <strong>Kong</strong>, in under going insurance regulatory reform, should treasure this“second chance” to promote itself as an attractive captive domicile. The upcoming reformed <strong>Hong</strong><strong>Kong</strong> Insurance Authority should take a more proactive approach on captive business development.The success and captive experience <strong>of</strong> Mainland Chinese companies will provide incentives for otherAsian companies to consider establishing their captives in <strong>Hong</strong> <strong>Kong</strong>.Ronald T. KozlowskiTowers WatsonJohnny HoTowers WatsonThe <strong>Actuarial</strong> <strong>Society</strong> <strong>of</strong> <strong>Hong</strong> <strong>Kong</strong>, 2202 Tower Two, Lippo Centre, 89 Queensway, <strong>Hong</strong> <strong>Kong</strong>Tel (852) 2147 9420 Fax (852) 2147 2497 Website: www.actuaries.org.hkNote: Views expressed are not necessary those <strong>of</strong> The <strong>Actuarial</strong> <strong>Society</strong> <strong>of</strong> <strong>Hong</strong> <strong>Kong</strong>12
This article was first published in the November <strong>2011</strong> issue <strong>of</strong> the Asia Insurance Review.Why Risk Margin?In a perfect world, Actuaries would be able todetermine best estimate reserves which willexactly be equal to the actual future lossesincurred by an insurer. Unfortunately, in reallife, things are not as simple. There are manyuncertainties and surprises which will causethe best estimate reserves to be higher orlower than the actual future losses incurred.For example, past and future claimfluctuations, future economic or environmentalconditions, errors in data or incorrectprojection models selection.A risk margin on top <strong>of</strong> the best estimatereserves provides for situations where theactual future losses may be higher than thebest estimate reserves. This extra comfortlevel helps maintain the financial soundness <strong>of</strong>insurers and to protect policyholders.Risk Margin DeterminationThere are numerous approaches to determinerisk margin. It may be determined as a fixedmonetary amount or a percentage <strong>of</strong> the bestestimate reserves. In slightly morecomplicated form, the reserving actuarycarries out an alternative set <strong>of</strong> reserveprojections using more conservativeassumptions, such as claim inflation, to arriveat a more conservative reserve estimate. Thedifference between this conservative reserveand the best estimate reserve represents therisk margin.Both approaches are simple to implement, butthey fail to describe the degree <strong>of</strong> prudencequantitatively represented by the risk margin.There is little statistical reasoning or basisbehind the choice <strong>of</strong> the risk marginassumptions.A number <strong>of</strong> methods have been suggested toestimate the risk margin with more solidstatistical basis.Two commonly used methods which can beeasily adapted in Asia are (1) the Mack methodand (2) bootstrapping. In addition, there arethe cost-<strong>of</strong>-capital method integrated in theSwiss Solvency Test and the to-beimplementedSolvency II framework in Europe.Mack methodThe Mack method is a natural extension <strong>of</strong> thechain ladder method commonly used forestimating best estimate reserves. The chainladder method involves using run-<strong>of</strong>f claimsdata to estimate development factors (measure<strong>of</strong> the speed <strong>of</strong> reporting and settling claims),then using the development factor to project theultimate cost <strong>of</strong> claims. Subtracting the paidclaims from the projected ultimate cost <strong>of</strong>claims gives the claims reserves <strong>of</strong> theportfolio. Even without any changes in theclaims management process, claimsdevelopment would still fluctuate over time dueto randomness <strong>of</strong> claims occurrence, claimsseverity and the speed at which the claims arereported and settled. Therefore the estimation<strong>of</strong> development factor is inherently uncertain.In statistical term, a number representing themeasure <strong>of</strong> uncertainty is called the standarddeviation. The uncertainty in claimsdevelopment can naturally be used to infer thestandard deviation <strong>of</strong> the ultimate cost <strong>of</strong> claimsand the reserves.The risk margin may be decided as a certainmultiple <strong>of</strong> the standard deviation <strong>of</strong> thereserves (e.g. 0.5 times, 1 time, 1.5 times, etc).Furthermore, it can be assumed that thereserves follow a particular statisticaldistribution to infer a full distribution <strong>of</strong> thereserves to calculate the level <strong>of</strong> sufficiency atvarious levels <strong>of</strong> reserve. Typically, the range <strong>of</strong>sufficiency level insurers is interested in forreserving purpose is between 75% to 90 %sufficiency. In this case, a normal distributionmay be assumed. For more extremescenarios, e.g. stress testing purpose, thismethod is not recommended.The <strong>Actuarial</strong> <strong>Society</strong> <strong>of</strong> <strong>Hong</strong> <strong>Kong</strong>, 2202 Tower Two, Lippo Centre, 89 Queensway, <strong>Hong</strong> <strong>Kong</strong>Tel (852) 2147 9420 Fax (852) 2147 2497 Website: www.actuaries.org.hkNote: Views expressed are not necessary those <strong>of</strong> The <strong>Actuarial</strong> <strong>Society</strong> <strong>of</strong> <strong>Hong</strong> <strong>Kong</strong>13