The RATS Software Forum

Can we use the Granger causality test and Engle-Granger cointegration test for discrete data? Or these tests are only used for time series data?
Many thanks
Best regards

cecedi wrote:Can we use the Granger causality test and Engle-Granger cointegration test for discrete data? Or these tests are only used for time series data?
Many thanks
Best regards

What do you mean by "discrete"? A dummy variable or any bounded variable can't be I(1) so an EG test wouldn't make any sense. You can put dummy shifts into cointegrating vectors if that's what you want to do. CATS is set up to do that. Note that it changes the asymptotic distribution.