(删除变量u)adf检验及模型函数估计

《(删除变量u)adf检验及模型函数估计》由会员上传分享，免费在线阅读，更多相关内容在教育资源-天天文库。

1、删除变量M2/M0)，建立模型：经检验：所有序列（变量）二阶（差分）单整，故可以得出他们之间存在协整关系。一、对所构造的函数形式：lslnm1clngdplnvfruu进行OLS估计得：DependentVariable:LNM1Method:LeastSquaresDate:07/17/11Time:16:07Sample:19932009Includedobservations:17VariableCoefficientStd.Errort-StatisticProb.  C-0.7110690.122437-5.8076510.0001LNGDP0

2、.8897460.03479725.569320.0000LNV0.0252900.0120652.0962220.0600F-0.0045460.000838-5.4245750.0002R0.0116610.0037103.1428340.0094UU0.8197890.07002911.706480.0000R-squared0.999608    Meandependentvar6.379730AdjustedR-squared0.999429    S.D.dependentvar0.800568S.E.ofregression0.019128

3、    Akaikeinfocriterion-4.804743Sumsquaredresid0.004025    Schwarzcriterion-4.510668Loglikelihood46.84032    Hannan-Quinncriter.-4.775512F-statistic5603.099    Durbin-Watsonstat2.559494Prob(F-statistic)0.000000因为变量LnV不显著，将其删除，重新估计：lslnm1clngdpfruuDependentVariable:LNM1Method:Leas

4、tSquaresDate:07/17/11Time:16:10Sample:19932009Includedobservations:17VariableCoefficientStd.Errort-StatisticProb.  C-0.8718970.108072-8.0677730.0000LNGDP0.9497760.02238942.421070.0000F-0.0051480.000892-5.7726120.0001R0.0085100.0038422.2149540.0469UU0.7678080.07417710.351000.0000R

5、-squared0.999451    Meandependentvar6.379730AdjustedR-squared0.999268    S.D.dependentvar0.800568S.E.ofregression0.021665    Akaikeinfocriterion-4.586298Sumsquaredresid0.005633    Schwarzcriterion-4.341236Loglikelihood43.98354    Hannan-Quinncriter.-4.561939F-statistic5458.784   

6、 Durbin-Watsonstat2.600129Prob(F-statistic)0.000000所以：LNM1=-0.871897137169+0.949776234674*LNGDP-0.00514818980363*F+0.00851005559974*R+0.767808226881*UU……(1)；且各个变量均通过显著性水平0.05的检验。二、对回归的残差序列E1进行平稳性检验：NullHypothesis:E1hasaunitrootExogenous:ConstantLagLength:0(AutomaticbasedonSIC,MAX

7、LAG=3)t-Statistic  Prob.*AugmentedDickey-Fullerteststatistic-5.107454 0.0011Testcriticalvalues:1%level-3.9203505%level-3.06558510%level-2.673459其中E1为当前模型的残差序列，由输出结果，可见序列E无单位根，LnM1与各序列协整。短期动态货币需求函数估计:按照Hendry建模理论（年度数据从滞后2期开始，季度数据从滞后8期开始删除不显著变量，本实例中我们均采用年度数据，故从滞后2期开始），首先建立一个能够代表数据生

8、成过程(DGP)的自回归分布滞后模型(ADL)，然后逐步简化，最后得到包含变量间