台灣工程保險之契約條款問題與相關實證研究

本研究係以台灣工程保險契約之保險賠案實證研究為研究主題，首先以近十五年蒐集之國內保險賠案資料為樣本空間，嘗試以風險管理及統計精算相關理論，檢定分析台灣高科技廠房設施安裝工程以及交通工程洪災之損失模型、損失頻率、損失幅度、正規化損失因子等機率分配及風險值。在高科技廠房設施安裝工程以及交通工程洪災之風險值計算部分，本研究更進一步採用變異數與共變異數法、歷史資料法、蒙地卡羅分析法及Riskiness模型法計算。此外，本研究亦利用資訊不對稱模型探討台灣隧道工程是否有無存在資訊不對稱性。根據演算分折之結果，高科技廠房設施工程在損失頻率機率分配部份，高科技廠房設施工程每月每保單損失次數N之母體分配適合於參數λ ̃=0.09701的Poisson分配；在損失幅度分配部分，較適合之機率分配為Lognaormal 機率分配；在損失時間分配部分，較適合之機率分配為Beta機率分配。另正規化損失因子風險值理論模型部分，本研究認為宜採蒙地卡羅模擬法為佳。交通工程洪災損失部分，在損失幅度分配部分，較適合之機率分配為Lognaormal 機率分配；在損失時間分配部分，較適合之機率分配為Weibull機率分配。另正規化損失因子風險值理論模型部分，本研究認為宜採蒙地卡羅模擬法為佳。最後，由實證結果發現，隧道工程保險自負額的高低與隧道工程保險理賠事故呈現條件相關，存在有資訊不對稱的現象

關鍵字

高科技廠房；交通工程；工程保險；風險值；資訊不對稱

並列摘要

This research is based on the empirical research on insurance claims of construction insurance contracts in Taiwan. First, it used the insurance claims data collected in the past 15 years as the sample space, and attempted to verify and analyze Taiwan high-tech facilities based on risk management and statistical actuarial related theories. The probability distribution and risk value of high-tech facility work and traffic engineering flood disaster loss model, loss frequency, loss amplitude, normalized loss factor, etc. In the calculation of risk value of high-tech facilities work and traffic engineering flood disaster, this research further used the variance and covariance method, historical data method, Monte Carlo analysis method and Riskiness model method to calculate. In addition, this research also used the information asymmetry model to explore whether there is information asymmetry in Taiwan tunnel work. According to the result of calculation analysis, high-tech facility work is in the part of loss frequency probability distribution. The matrix distribution of high-tech facility work monthly loss per insurance policy N is suitable for the Poisson distribution of parameter λ ̃=0.09701; In the loss amount distribution, the most suitable probability distribution is Lognaormal probability distribution; in the loss time distribution part, the most suitable probability distribution is Beta probability distribution. In addition, in the part of the theoretical model of the normalized loss factor risk value, this study believed that the Monte Carlo simulation method is better than others. About traffic engineering flood damage loss, in the loss amount part, the most suitable probability allocation is Lognaormal probability allocation; in the loss time ditribution part, the most suitable probability ditribution is Weibull probability. In addition, in the part of the theoretical model of the normalized loss factor risk value, this study believes that the Monte Carlo simulation method is also better than others. Finally, from the empirical results, it was found that the level of tunnel work insurance deductible was related to the conditions of tunnel work insurance claims, and there was a phenomenon of information asymmetry in Taiwan.