如何用stata做协整.pdf

《如何用stata做协整.pdf》由会员上传分享，免费在线阅读，更多相关内容在行业资料-天天文库。

1、一、画时间序列趋势原数列：twoway(line变量时间变量,sort)---不平稳一阶差分：twoway(lineD.变量时间变量,sort)---平稳两个或多个一阶单整的时间序列可能存在协整关系。ADF检验：原假设存在单位根，非平稳EXPORT.dfullerlnexport,trendDickey-FullertestforunitrootNumberofobs=17InterpolatedDickey-FullerTest1%Critical5%Critical10%CriticalStatisticValueValueValueZ(t)-0.412-4.3

2、80-3.600-3.240MacKinnonapproximatep-valueforZ(t)=0.9865.dfullerd.lnexport,trendDickey-FullertestforunitrootNumberofobs=16InterpolatedDickey-FullerTest1%Critical5%Critical10%CriticalStatisticValueValueValueZ(t)-3.558-4.380-3.600-3.240MacKinnonapproximatep-valueforZ(t)=0.0335.dfullerd.ln

3、exportDickey-FullertestforunitrootNumberofobs=16InterpolatedDickey-FullerTest1%Critical5%Critical10%CriticalStatisticValueValueValueZ(t)-3.158-3.750-3.000-2.630MacKinnonapproximatep-valueforZ(t)=0.0226IMPORT.dfullerd.lnimportDickey-FullertestforunitrootNumberofobs=16InterpolatedDickey-

4、FullerTest1%Critical5%Critical10%CriticalStatisticValueValueValueZ(t)-2.607-3.750-3.000-2.630MacKinnonapproximatep-valueforZ(t)=0.0915.dfullerlnimportDickey-FullertestforunitrootNumberofobs=17InterpolatedDickey-FullerTest1%Critical5%Critical10%CriticalStatisticValueValueValueZ(t)0.445-

5、3.750-3.000-2.630MacKinnonapproximatep-valueforZ(t)=0.9831EXPENDITURE.dfullerlnexpenditureDickey-FullertestforunitrootNumberofobs=17InterpolatedDickey-FullerTest1%Critical5%Critical10%CriticalStatisticValueValueValueZ(t)0.172-3.750-3.000-2.630MacKinnonapproximatep-valueforZ(t)=0.9706.d

6、fullerd.lnexpenditureDickey-FullertestforunitrootNumberofobs=16InterpolatedDickey-FullerTest1%Critical5%Critical10%CriticalStatisticValueValueValueZ(t)-5.141-3.750-3.000-2.630MacKinnonapproximatep-valueforZ(t)=0.0000二、协整秩检验看是否存在协整关系.vecranklnexpenditurelnexport,trend(trend)Johansentest

7、sforcointegrationTrend:trendNumberofobs=16Sample:1992-2007Lags=25%maximumtracecriticalrankparmsLLeigenvaluestatisticvalue0837.407322.19.448218.1711145.9795320.657512.3038*3.7421247.1314440.13410.vecranklnexpenditurelnimport,trend(nocon)lags(3)JohansentestsforcointegrationTrend:consta