no cointegration - now what

[shyla01]

Hi all, i have a really big problem!

when undertaking my econometric analysis i found all my variables to be I(1) that is stationary after first difference.
my model is as follows: yt = b1 b2 x1 b3 x3 ut (2 independant variables)

the problems:
1) can you estimate a conitegration test for 3 variables using the engle-granger 2 step method or is this only valid in a bivariate model?
2) when conducting the Johansen test my variables were found not to cointegrated!!

is it still possible to continue my study using the first differences of the variables or is this inconsistent?
PLEASE HELP ME IF YOU CAN!!

many thanks!

[random_access]

hi there
if you mean by Engle-Granger test of cointegration is to test the residual of the regression to be stationary i guess you can do that even for multivariate model.
for J-J cointegration you might need to specifiy the lags before running J-J cointegration.
all the best