国债市场利率期限结构建模—-负指数立方L1平滑样条

通过对比国内外利率期限结构静态估计模型的优劣, 分析节点数目变化和定位改进B样条函数对利率期限结构静态估计的误差, 构建最小化定价误差的节点组合布局搜索程序, 并引入负指数平滑立方L₁样条优化模型, 将误差函数最小化结构从平方和最小化转化为误差距离最小化, 权衡拟合误差绝对距离最小化与贴现函数波动性约束, 克服B样条函数对节点数目与定位的人工干预和放宽对贴现函数的二阶平滑要求, 保留B样条函数刻画中长期利率波动趋势的优势, 增强对短期利率波动结构突变的估计和预测能力, 提高定价精确度和缓解利率期限结构曲线的过度波动问题.

Modelling term structure of interest rate in T-bill market based on negative exponential cubic smooth L1-spline

Based on the comparision of basic static estimate methods of term structure of interest rate (TSIR), we improved B-spline function estimate method, which involved optimization on estimation programmes, node numbers choice, and node placement design. To overcome the subjective effect of B-spline node distribution and C² smoothness condition of discount function, we introduced negative exponential smoothness cubic L₁-spline optimization technology with minimum constraint function of estimation error from quadratic sum to absolute value and minimum volatility of discount function, to increase the estimation reliability and prediction ability of short-term interest rate's volatility structure mutation, improve the advantage on depicting the long-term interest rate volatility trend, and reduce the excessive volatility of discount function.