Statistical Model for Medical Expense: Based on Extreme Value Theory

• Main Author: 林君翰
• Other Authors: 張智凱
• Format: Others
• Language: zh-TW
• Published: 2013
• Online Access: http://ndltd.ncl.edu.tw/handle/69kk4a

碩士 === 逢甲大學 === 統計學系統計與精算碩士班 === 101 === This study proposes a loss model based on Extreme Value Theory for the outpatient expense costs in Taiwan Bureau of National Health Insurance. From the National Health Insurance Data from Year 2003 to 2009, a heavy-tail distribution occurs as the average outpatient costs exceed NT 12,000. Empirical results indicate that, the fitted model is consistent with the actual high outpatient costs. Comparing to the actual expense, the Generalized Pareto Distribution (GPD) model reflects the cost down outcome by the Taiwan government’s policy to control the high outpatient costs in 2010. By the proposed models, we discuss the costs in a private insurance contract entitled the payment options between full indemnity and predefined compensation. Accordingly, with the average medical cost of outpatient is approximately 2,000 to 3,000, it can be inferred that the cost of this payment option accounts for 30 to 50% of the average outpatient costs. Our proposed models and empirical results provide important suggestions and economic implications for the risk management of health insurance.