[Gretl-users] ADF test

[john w]

Hi,
I tried to do ADF tests with and without Test down option. Using always the 
same variable.
Here are the results:

ADF test
Max order =4
Const + trend included

Test down checked

Dickey-Fuller tests for X
sample size 35
unit-root null hypothesis: a = 1

   with constant and trend
   model: (1 - L)y = b0 + b1*t + (a-1)*y(-1) + e
   lag order: 0
   1st-order autocorrelation coeff. for e: -0,044
   estimated value of (a - 1): -0,554223
   test statistic: tau_ct(1) = -2,98904
   p-value 0,1495

P-values based on MacKinnon (JAE, 1996)

Then I tested without Test down option but using lag order used by Test down 
option (i.e. zero). Here are the results:

ADF test
Order= 0
Const + trend included

Dickey-Fuller tests for X
sample size 36
unit-root null hypothesis: a = 1

   with constant and trend
   model: (1 - L)y = b0 + b1*t + (a-1)*y(-1) + e
   1st-order autocorrelation coeff. for e: -0,111
   estimated value of (a - 1): -0,608096
   test statistic: tau_ct(1) = -4,04436
   p-value 0,01589

P-values based on MacKinnon (JAE, 1996)

Test statistics are different i.e. with Test down option is -2,98904, and 
without -4,04436.

Then I tested without Test down option but using one lag. Here are the 
results:

ADF test
Order =1
Const + trend included


Augmented Dickey-Fuller tests, order 1, for X
sample size 35
unit-root null hypothesis: a = 1

   with constant and trend
   model: (1 - L)y = b0 + b1*t + (a-1)*y(-1) + ... + e
   1st-order autocorrelation coeff. for e: -0,044
   estimated value of (a - 1): -0,554223
   test statistic: tau_ct(1) = -2,98904
   asymptotic p-value 0,1352

P-values based on MacKinnon (JAE, 1996)


As we can see, test statistic is the same as with Test down option...but 
used lags are not.
This also can be seen by looking at sample sizes. With Test down optin lag 
order is o, sample size is 35 and test statistic is -2,98904 while without 
Test down sample size is 35, test statistic is -2,98904 but lag order is 1.

Is it something wrong or not?

Thnx

_________________________________________________________________
Be the first to hear what's new at MSN - sign up to our free newsletters! 
http://www.msn.co.uk/newsletters