CAJ | 학술논문

针对单个静态利率期限结构模型在拟合收益率曲线时的不足，本文引入组合预测的方法，在绝对误差和与方差和最小准则下，分别建立了静态利率期限结构组合优化模型，并给出了模型的遗传算法求解过程。然后将上海证券交易所2004～2009年的国债每日交易数据分为样本内数据和样本外数据，对多项式样条、指数样条、NS、SV和组合优化模型进行实证比较。结果表明：无论是对于样本内数据的拟合，还是对于样本外数据的预测，组合优化模型的统计特征指标几乎都要优于其他单一模型，并且具有良好的适应性和稳健性，适用于拟合我国国债利率期限结构。
침대단개정태리솔기한결구모형재의합수익솔곡선시적불족，본문인입조합예측적방법，재절대오차화여방차화최소준칙하，분별건립료정태리솔기한결구조합우화모형，병급출료모형적유전산법구해과정。연후장상해증권교역소2004～2009년적국채매일교역수거분위양본내수거화양본외수거，대다항식양조、지수양조、NS、SV화조합우화모형진행실증비교。결과표명：무론시대우양본내수거적의합，환시대우양본외수거적예측，조합우화모형적통계특정지표궤호도요우우기타단일모형，병차구유량호적괄응성화은건성，괄용우의합아국국채리솔기한결구。
For the weakness in fitting the yield curve of certain term structure of interest rates model , this paper introduces the combinatorial forecast method , and develops respectively the combinatorial term structure of inter-est rates models based on the principle of the minimization variance and the minimization absolute error criterion , and presents the process of genetic algorithms for the models .The data of Shanghai Stock Exchange from 2004 to 2009 are divided into in-sample and out-of-sample data for the empirical research .The comparisons of empirical results of the five models including polynomial spline , exponential spline , Nelson-Siegel , Svensson and the com-binatorial optimization show that combinatorial optimization model outperforms other models in most of the statisti -cal indicators , and has good adaptability and robustness .It can be applied to fitting the static term structure of interest rates in the bond market of China .