Post-Print筆數 : 11 |
Items with full text/Total items : 89327/119107 (75%)
Visitors : 23834265
Online Users : 595
Please use this identifier to cite or link to this item:
|Other Titles: ||A Study of Longevity Risk and Mortality Models|
Longevity Risk;Age-period-cohort Model;Mortality Models;Lee-Carter Model;Computer Simulation
|Issue Date: ||2012-08-30 09:59:31 (UTC+8)|
|Abstract: ||死亡率下降是二十世紀以來的共同趨勢，人口老化為各國必須面對的問題，延壽風險遂成為老年生活規劃必須面對的挑戰。因為壽命延長快速，傳統的保險費率計算會造成年金商品的保費不足，保險公司有破產之虞。過去二十年來有不少因應延壽風險的對策被提出，包括自然避險、長壽債券等，其中隨機生命表因為操作較為直接，是因應對策的主流之一，歐洲各國自1980年代起有學者投入研究。Lee-Carter(LC)模型目前是最受歡迎死亡率模型，不過因為LC模型缺乏「世代」因素的考量，近年有不少加入世代因素的修正模型，彌補LC模型在高齡死亡率的瑕疵，但估計方法及模型仍不少有待改進之處。 本研究計畫引進常用於流行病學的「年齡─時間─世代」(Age-Period-Cohort；APC)模型，將其在世代研究的經驗，套用在修正修正LC的世代模型；另外，本研究也將整理LC以外的死亡率模型，改進現行的死亡率預測。本計畫將分成三年進行：第一年整理APC模型的估計方法，透過電腦模擬及實證分析，評估方法的優劣；第二年將上述APC模型的研究經驗套至LC模型，整合現有的世代修正模型；第三年整理LC以外的其他模型，包括多變量分析、Functional PCA (Principal Component Analysis)等，再與LC修正模型配合，建立適合台灣地區的死亡率模型。|
Mortality improvement has been a common phenomenon in many countries since the beginning of 20th century and the population aging becomes a major concern. Because of the rapid prolonging life expectancy, the traditional calculation based on the period life tables would produce in-sufficient insurance premium for the annuity products. The insufficient premium due to the prolonging life expectancy is usually known as “longevity risk.” Many solutions have been proposed for solving the longevity risk, such as natural hedging and longevity bonds. Among all solutions, using stochastic life table for premium calculation is one of the natural ways of dealing the longevity risk. The Lee-Carter (LC) model (Lee and Carter, 1992) probably is the most popular choice of stochastic mortality models. However, the LC model is lack of the factor of “cohort” and many empirical analyses show that the assumption of the LC fails. By adding the cohort effect, several modifications of the LC model were proposed in the last decade to improve the model fit for the old age groups. Still, there are problems in the modification models remained to be solved, such as the estimation method. In this study, we will introduce the famous age-period-cohort (APC) model in epidemiology and try to improve the cohort-adjusted LC models. In specific, we will compare and evaluate the estimation methods of the APC model and apply the most appropriate method(s) in modifying the LC model. In addition, we will review and study the mortality models, other than the LC model. For the first year of the project, we will focus on the study of the estimation methods of the LC model. Computer simulation and empirical studies will be used to evaluate these methods. The focus of the second year is on improving the cohort-adjusted LC models, based on the study of the APC model from the first year. The third year is to review and summary mortality models other than the LC model, for example, the multivariate analysis method and the functional PCA. We shall compare these methods with the LC model, and then propose a model which can describe the Taiwan mortality well.
|Data Type: ||report|
|Appears in Collections:||[統計學系] 國科會研究計畫|
Files in This Item:
All items in 政大典藏 are protected by copyright, with all rights reserved.