小样本协整检验神经网络方法.pdfVIP

13 5 系 统 工 程 理 论 方 法 应 用 V ol. 13 N o . 5 2004 10 SY ST EM S EN GI NEER IN G- T HE O RY M ET HO DO LO G Y A PP L ICA T IO N S O ct . 2004 : 1005-2542( 2004 05-0409-05 杨宝臣, 　张世英 ( , 300072 采用M onte Carlo 模拟方法, 给出了小样本容量下协整 验临界值的响应面方程。结果表明, 小 样本协整 验的临界值不仅与样本容量有关, 还依赖于协整 验中的滞后阶数。进 一步, 以样本容量、变量 个 数、滞后阶数为输入指标, 验临界值为输出指标建立了神经网络模型。神经网络模型比响应面模型具有更 高的精度。 : 协整 验; 响应面方程; 神经网络; 小样本 : O 213 : A Cointegration Test with Small Sample: Comparison of Response Surf ace and Neural Network Approaches YA N G B ao-chen, ZH A N G S hi-y ing ( Scho ol o f Management , T ianjin Univ. , T ianjin 300072, China , Abst ract In this paper w e first developed t he response surf ace equat ions fo r det erminat ion o f crit ical value fo r cointegr at ion t est in sm al l sample u sing M ont e Carlo simulat ion , t he result s indicat e t hat t he crit - ical values f or co int eg rat ion t est are related w it h t he sam ple size, as w ell as t he lagg ed order . A nd t hen , w e est ablished t he neural net w or k model f or cointeg ration test w it h small sample, w her e the sam ple size, num ber of variables , and lagg ed order w er e employ ed as input s and crit ical values as outputs . Evidence show s that the neural net wo rk appr oach is mor e accurate t han t he response surf ace m odel w hich is com- monly used in crit ical v alue determinat io nf or crit ical values . Key words: coint egr at ion t est ; response surf ace m odel; neural net wo rk ; small sample [ 1] [ 24] , ,