JOHMLE—Johansen ML Testing/Estimation for Cointegration

[TomDoan]

@JOHMLE does basic tests for cointegration using the Johansen ML technique. This is a very small subset of what the CATS package of procedures can do.

Detailed description

[abi]

Hi Tom,

Is it possible to use Johansen cointegration test for variables that are integrated with different orders of stationary?. I have five variables that fourth of them are I(1) and another one is I(0).
Thanks,

[TomDoan]

An I(0) variable is cointegrated "with itself". If you understand that, then yes. Analyzing that combination of variables, however, is more what CATS is designed to do. (For instance, it has specific tests for stationarity).

[abi]

Thank you Tom for reply

Does it mean an i(0) variable adds one rank to the cointegration space? however, i would be appreciate if you explain more.

[TomDoan]

Yes. X1 is stationary (i.e. I(0)) then X(t)beta is I(0) if beta=||1,0,...,0|| which means that that qualifies as a cointegrating vector.

[abi]

Thanks again,

Tom! why RATS doesn't report Std. Err. or t statics for normalized cointegrating vectors?

[TomDoan]

@JOHMLE isn't intended to be a substitute for CATS. Once you get more than one cointegrating vector, if you want to make any sense of the results, you need something more than basic normalizations (either some type of loading restriction or some type of construction restrictions on the CI's, or both) which is what CATS is designed to help you with.