随机情景生成模型的参数估计

The Parameter Estimation of Stochastic Scenario Generation Model

随机规划模型由于其自身的独特优势,在金融机构及个人的长期金融资产负债管理中的应用日益广泛,而对未来不确定性的合理刻画(通常被称为情景生成)是其成功应用的关键。随机微分方程是一种生成情景的重要方法,为了提高情景生成的质量,需要对模型参数进行精确的估计。与常用的一些模型参数估计方法相比,综合参数估计法是一种更具优势的方法。

Stochastic programming model （SPM） is widely used in asset and liability management by many financial institutions and individual investors for its special merits. It＇s an essential step to describe future uncertainty （often named scenario generation） accurately to ensure its successful utilization. Stochastic differential equation is an important way to generate scenarios. Model parameters need to be estimated accurately for getting more representative scenarios. This paper briefly introduces some usual ways for estimating parameters and offers a better method named integrated parameter estimation （IPE） in detail. Then it empiricaUy compares the effects of ML, GMM and IPE and offers the direction for further study of parameter estimation.