A Sotr Of Credibility Premium Model For Fuzzy Dependent Risk Variables

In traditional actuarial textbooks, the random variables concerned with the premiummodel are assumed to be independent. However, in many practical applications, there aresome common factor or some mutually connected factors which effect the risk randomvariables, which leads to that individual risks in the insurance are generally more or lessdependent. For instance, the dependent relation between husband and wife in a joint lifeinsurance policy, insurance of the workers work in the same workshop, etc. Obviously, theassumption of independence mentioned above does not coincide with the reality cases,therefore, studying the relationship between risks implies a practical significance. Fuzzyrelation is a common relationship affected by human subjective factors, which has manyadvantages with respect to statistical measurement and treatment. In the past, credibilitypremium and premium principles are considered separately independent in the premiummodel, but there generally are some mutually affection and impact from risks in actuarialpractice, such as the insurance risks in vehicles insurance. Based on the theoretical study ontraditional credibility premium model, we introduce the fuzzy concept into the risks models,and further consider the application of the premium model. Using fuzzy set theory we dealwith the relationship between the variables, and propose a premium model with fuzzy concept.The main considerations of this paper are as follows:1. Using fuzzy set theory to study the credibility premium under the fuzzy environment,to give more reality-fit practical and scientific methods for the calculation of the credibilitypremium.2. For the exponential premium principle, to give an exponential premium principle andwhose calculation method in a fuzzy environment.3. For dependency problems, using a combination of copula function and the fuzzy setsmethods to study the relationship of variables in dependency situation, and give the discussionof dependent variables in the model of credibility premium.