Copula Analysis for the Inpatient and Outpatient Claim Data of Cancer Population
| Autor: | LUO, FANG-LIN, 羅方伶 |
|---|---|
| Rok vydání: | 2019 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 107 This study explores the future risk for cancer insurance policies. Because the medical expenses are highest within the first two years after carcinoma, this study researches all outpatient services and inpatient points for all cancer patients within two years. It does not consider the correlation of each project risk because the current commercial insurance pricing method is based on various payment items by which risk is calculated separately. Therefore, it calculates the future risk separately on the individual basis of the original method and the actual basis of the relevant correlation and explores the difference between these two. The five variables of medical points within the first two years after cancer are: "Total points of medical expenses in the first year after cancer," "Total points of medical expenses in the second year after cancer," "Total points of outpatient services within the first two years after cancer," and "Total points of inpatient treatment within the first two years after cancer," "Total points of medical expenses within the first two years after cancer." Furthermore, I calculate statistical variables such as value at risk, conditional tail expectation, and risk measurement values for these five variables and make comparisons among them. Then, I use the Copula model to match the joint distribution of medical expenses for the entire cancer treatment within the first two years. Taking males aged 45 to 54 years that lived for two years as examples, and after matching, the Copula models that I chose are the T Copula and Normal Copula. According to the correlation coefficient analysis, the joint distributions select the optimal Copula models distinctively for different variables. The results of the study show that the risk measurement in the second year after cancer is higher than that in the first, and the risk of inpatient treatment within the first two years is higher than the risk of outpatient services during the same period. After comparing the independent simulation values with the Copula analog values, it is clear that the right tail allocation obtained by the original independent calculation method is more concentrated, while the right tail distribution of the actual correlation condition is more dispersed. Hence, if the original method is used, the future risks may be underestimated, and if the future risk is closer to the end, a greater difference between the two will emerge. Observing the five variables, the error value of the risk measurements is greater than 5% and the more the number of variables the bigger the error value becomes between the calculation method and the actual situation. This means that whether the correlation is considered or not, the result has influence. Therefore, it is reasonable to speculate that it is significant to calculate the Copula model when estimating the future risk. Finally, I hope that the results of this study will be beneficial to insurance companies' assessment of future risk. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|