about cointegration command: coint(method=eg)

[lexjl5]

Hi,

I have one problem about the cointegration command for engle-granger cointegration.

As I have a large pool of price time series and I want to calculate the ADF t-value for each pair-wise in the data-set, I use the command coint(method=eg) in the loop. I found, for most pair-wise, the ADF t-value got from the command is different from the mouse click procedure. I attached two price time series as one sample.

My command is : group g log(_1a) log(_2a)
g.coint(method=eg)

for mouse click: Quick--> Equaiton Estimation--> cointreg --> Dynamic OLS (none) --> Engle-Granger cointegration test

I found both the t-value and the observation in the intermediate results are different.

Why are they different? Or how to write the programme to run as the mouse click procedure.

Your assistance is highly appreciated

many thanks for your help

best wishes

Jianan Li

[EViews Glenn]

Are the lags used the same? The EG uses a default of 1.