Different numbers in ADF test

[Luv_eviews]

Sorry for this trivial question. But I'd like to know why this happens^^

Below are results from ADF tests in eviews.

Upper one: I myself estimated ADF equation.
Lower one: Result from the built-in ADF test in Eviews.

Everything is the same except for the value of the contant.
Upper one: the constant is 0.048948
Lower one: the constant is 0.070604

Anybody can explain how this occured?
Thank you in advance.

--------------------------------
Dependent Variable: D(LRGDP)
Method: Least Squares
Date: 10/07/12 Time: 18:02
Sample: 1960Q4 2002Q1
Included observations: 166

Variable Coefficient Std. Error t-Statistic Prob.

LRGDP(-1) -0.064894 0.019120 -3.394079 0.0009
D(LRGDP(-1)) 0.240728 0.075082 3.206210 0.0016
D(LRGDP(-2)) 0.204157 0.075419 2.706967 0.0075
C 0.048948 0.012549 3.900502 0.0001
@TREND 0.000504 0.000153 3.285528 0.0012

R-squared 0.175503 Mean dependent var 0.008316
Adjusted R-squared 0.155018 S.D. dependent var 0.008784
S.E. of regression 0.008074 Akaike info criterion -6.770603
Sum squared resid 0.010496 Schwarz criterion -6.676868
Log likelihood 566.9600 Hannan-Quinn criter. -6.732555
F-statistic 8.567631 Durbin-Watson stat 1.988653
Prob(F-statistic) 0.000003

----------------------------------------------
Augmented Dickey-Fuller Test Equation
Dependent Variable: D(LRGDP)
Method: Least Squares
Date: 10/07/12 Time: 17:49
Sample: 1960Q4 2002Q1
Included observations: 166

Variable Coefficient Std. Error t-Statistic Prob.

LRGDP(-1) -0.064894 0.019120 -3.394079 0.0009
D(LRGDP(-1)) 0.240728 0.075082 3.206210 0.0016
D(LRGDP(-2)) 0.204157 0.075419 2.706967 0.0075
C 0.070604 0.019018 3.712564 0.0003
@TREND(1960Q4) 0.000504 0.000153 3.285528 0.0012

R-squared 0.175503 Mean dependent var 0.008316
Adjusted R-squared 0.155018 S.D. dependent var 0.008784
S.E. of regression 0.008074 Akaike info criterion -6.770603
Sum squared resid 0.010496 Schwarz criterion -6.676868
Log likelihood 566.9600 Hannan-Quinn criter. -6.732555
F-statistic 8.567631 Durbin-Watson stat 1.988653
Prob(F-statistic) 0.000003

[EViews Gareth]

I cannot replicate what you are seeing. I just did it a number of times on random data, and always achieved matching results.

Code: Select all

create m 1990 2020 series lrgdp = rnd*100 equation eq1.ls d(lrgdp) lrgdp(-1) d(lrgdp(-1)) d(lrgdp(-2)) c @trend freeze(mytab) lrgdp.uroot(trend,2) show eq1 show mytab

[Luv_eviews]

I'm using Eviews7, Gareth.
And you are right. I did the same thing you did. As you posted, they were perfect match.

Below is what I did. And I attached a screenshot.

The range of lrgdp data is from 1947q1 to 2008q2.
I set the sample period from 1960q4 to 2002q1.

1. Using built-in ADF test
Open 'lrgdp' and view-unitroot test-ADF(level, trend and intercept, lag length 2)
2. Estimating ADF test
Quick-Equation estimation
Set sample from 1960q4 2002q1
Specification: d(lrgdp) lrgdp(-1) d(lrgdp(-1)) d(lrgdp(-2)) c @trend