Statistical Estimation of Optimal Portfolios for Dependent Returns of Assets

(1)当金融收益为ARMA-GARCH过程或时变ARCH过程时，提出了适当的重抽样(Bootstrap)方法和最优投资组合权重估计器。此外，还研究了这些估计量的渐近性质。此外，利用经验数据验证了这些方法的实际应用可能性。(2)研究了均值-方差组合以外的各种投资组合及其估计量的渐近性质。特别是对优化下偏矩的“悲观组合”进行了研究。在此基础上，给出了离散时间模型下多期问题的优化算法和养老保险资产负债管理流程。

(1) When the financial returns are the ARMA-GARCH process or the time-varying ARCH process, proper resampling(bootstrap) procedures and the optimal portfolio weight estimators are proposed. Moreover, the asymptotic property of these estimators are investigated. Furthermore, the practical use possibility of these techniques are verified using experience data.(2) Various portfolios other than mean-variance portfolio and the asymptotic property of these estimators are investigated. Especially, investigation of "Pessimistic Portfolio" which optimize the lower partial moment accomplished. Moreover the optimization algorithm in the multi-period problem under a discrete time model and a procedure of ALM(asset and liability management) in pension insurance are proposed.