Hi Gretl users,
I need advice with regard to Engle-Granger cointegration test. Why the tstats are different for the same series when flag--skip-df is added? I have noticed that for test carried with --test-down flag and without --skip-df flag, gretl uses N-pmax-1 observations (where pmax is max number of lags picked by a user for testing down), while with --skip-df flag it uses N-p-1, where p is number of lags based on information criteria/tstat calculated by Gretl? This affects tstats, pvalues and final test results.
Also, I have noticed that --test-down flag does not allow the choice between MAIC, MBIC and tstat anymore. This flag works fine on adf test thou.
Thanks in advance,
Andrzej