The RATS Software Forum

Hi,
i am using the above named procedure to analyse some series. here is a sample output.
TESTING THE NULL HYPOTHESIS OF A UNIT ROOT IN LINTRE *
* Using data from 1957:01 to 2008:04 *
* Choosing the optimal lag length for the ADF regression *
* between 0 and 12 lags. *
****************************************************************

Significance of lag12 : 0.82937
Significance of lag11 : 0.30275
Significance of lag10 : 0.00871

****************************************************************
* Augmented Dickey-Fuller t-test with 10 lags: -3.4300 *
* 1% 5% 10% *
* -3.99 -3.43 -3.13 *
* *
* Augmented Dickey-Fuller Z-test with 10 lags: -58.3679 *
* 1% 5% 10% *
* -28.4 -21.3 -18.0 *
* *
* Coefficient and T-Statistic on the Constant: *
* 0.52813 3.5584 *
* Coefficient and T-Statistic on the Linear Trend: *
* 0.00247 3.3984 *
* *
* Joint test of a unit root and no linear trend 5.8832 *
* 1% 5% 10% *
* 8.43 6.34 5.39 *
****************************************************************
my questions are:
1) does it mean at 5%, the test is inconclusive as regards the unit root test?
2) what is the interpretation of the Z test?
thanks for your anticipated response.
omos

my questions are:
1) does it mean at 5%, the test is inconclusive as regards the unit root test?

Yes, but that would also be the case if it were -3.44 or -3.42. All of them are right around the 5% level.

2) what is the interpretation of the Z test?

In this case, it rather strongly rejects the unit root. However, it's a rarely used form since the t-test is considered to be more reliable.

Thank you very much