engle-granger test with deterministic regresors
Posted: Wed Mar 21, 2018 9:33 am
Hi,
I have a problem with the p-value for Engle-Granger cointegration test with user-defined deterministic regressors, when we perform it using 2 different procedures offered by Eviews.One procedure is to do this test from the windows of estimation results (after FMOLS for example) and the other is to do it from the windows of the group of variables.
Using my data, with the first procedure, I obtain:
Cointegration Test - Engle-Granger
Specification: Y1 Y2 Y3 Y4 Y5 C T1 T2 D1
Cointegrating equation deterministics: C T1 T2 D1
Null hypothesis: Series are not cointegrated
Automatic lags specification based on Schwarz criterion (maxlag=7)
Value Prob.*
Engle-Granger tau-statistic -4.936035 0.0396
Engle-Granger z-statistic -28.80904 0.0420
With the second procedure, I obtain:
Series: Y1 Y2 Y3 Y4 Y5
Null hypothesis: Series are not cointegrated
Cointegrating equation deterministics: C T1 T2 D1
Automatic lags specification based on Schwarz criterion (maxlag=7)
Dependent tau-statistic Prob.* z-statistic Prob.*
Y1 -4.936035 0.2154 -28.80904 0.2377
We obtain exactly the same values for the statistics (in red), but the p-values (in green) are different.
Even if a message indicates in the 2 cases that “p-values do not account for user-specified deterministic regressors”, I think that they have to be the same in these 2 procedures. Please, do you have an explanation for this problem or is it a bug?
Thank you in advance.
I have a problem with the p-value for Engle-Granger cointegration test with user-defined deterministic regressors, when we perform it using 2 different procedures offered by Eviews.One procedure is to do this test from the windows of estimation results (after FMOLS for example) and the other is to do it from the windows of the group of variables.
Using my data, with the first procedure, I obtain:
Cointegration Test - Engle-Granger
Specification: Y1 Y2 Y3 Y4 Y5 C T1 T2 D1
Cointegrating equation deterministics: C T1 T2 D1
Null hypothesis: Series are not cointegrated
Automatic lags specification based on Schwarz criterion (maxlag=7)
Value Prob.*
Engle-Granger tau-statistic -4.936035 0.0396
Engle-Granger z-statistic -28.80904 0.0420
With the second procedure, I obtain:
Series: Y1 Y2 Y3 Y4 Y5
Null hypothesis: Series are not cointegrated
Cointegrating equation deterministics: C T1 T2 D1
Automatic lags specification based on Schwarz criterion (maxlag=7)
Dependent tau-statistic Prob.* z-statistic Prob.*
Y1 -4.936035 0.2154 -28.80904 0.2377
We obtain exactly the same values for the statistics (in red), but the p-values (in green) are different.
Even if a message indicates in the 2 cases that “p-values do not account for user-specified deterministic regressors”, I think that they have to be the same in these 2 procedures. Please, do you have an explanation for this problem or is it a bug?
Thank you in advance.