dfuller lgdp, trend lags(12) regress /*p-value is high, accept the null, has a unit root*/
regress dlgdp date l.lgdp l(1/12).dlgdp
test date l.lgdp /*n=142 10%5.47 5%6.49 1%8.73 since 1.27 below all, accept, time trend is not relevant*/
dfuller lgdp, lags(12) /*high, accept, has a unit root*/
regress dlgdp l.lgdp l(1/12).dlgdp
test _cons l.lgdp /*142 10%3.86 5%4.71 1%6.70 since 7.10 higher than all, reject, drift is relevant*/ 请问此时我检验出来drift是相关的应该怎么办呢?我又做了一步检验如下
dfuller lgdp, drift lags(12) regress /*0.067 stationary*/ 请问我是否可以直接得出结论数据是稳定的?还是说这是一个带drift的不稳定序列我需要做一阶差分?另外一阶差分结果如下:
dfuller dlgdp, trend lags(12) /*0.0035 significant at 1%, reject, first difference is stationary*/
第二组是电力季度消费量:
dfuller lele, trend lags(8) regress /*p-value is high, accept the null, has a unit root*/
regress dlele date l.lele l(1/8).dlele
test date l.lele /*142 10%5.47 5%6.49 1%8.73 since 4.68 below all, accept, time trend is not relevant*/
dfuller lele, lags(8) /*0.0302, reject, has not a unit root, stationary*/ 此时我可以直接得出数据稳定的结论么?如果可以 请问我这一步稳定和第一步稳定对我后续做协整检验和格林杰因果检验有什么不同的影响么?
第三组是季度人口:
dfuller lpop, trend lags(12) regress /*p-value is high, accept the null, has a unit root*/
regress dlpop date l.lpop l(1/12).dlpop
test date l.lpop /*142 10%5.47 5%6.49 1%8.73 since 1.74 below all, accept, time trend is not relevant*/
dfuller lpop, lags(12) /*0.5782, accept, has a unit root*/
regress dlpop l.lpop l(1/12).dlpop
test _cons l.lpop /*142 10%3.86 5%4.71 1%6.70 since 1.99 below all, accapt drift is not relevant*/
dfuller lpop, lags(12) noconstant /*1.321 higher than all conventional critical values. accept and has a unit root*/
dfuller dlpop, trend lags(12) /*0.4054 insignificant at 10%, accept, first difference is not stationary*/
这是不是说源数据和一阶差分都不稳定?可是二阶差分就没有经济意义了啊,是否意味着这个变量我不能用呢?
最后,我想将以上三组数据加上另一个I(1)稳定的数据一起做协整检验和格林杰因果检验,请问是否可行?麻烦各位了


雷达卡



京公网安备 11010802022788号







