The RATS Software Forum

Dear Tom,

I need to estimate a VAR model with asymmetric error correction, similar to Equation (6) in Enders and Granger (1998), and to Equation (9) in Enders and Siklos (2001).

I know that @EndersGranger and @EndersSiklos procedures do the estimations for these models. However, different from these models, I want to introduce the dummy variables myself (not estimated by the procedure). I want to estimate the speeds of adjustment to the long run equilibrium of US Treasury bonds (2-, 5-, 10-, and 30-year treasuries) during different time periods, and test the equality of the speeds of adjustments at different time periods. Can I do these estimations with CATS 2.0? If not, what options do I have?

Thank you,

Onem

You can't do that with CATS---it has a number of stability tests, but those are mainly on the cointegrating vector or cointegrating rank, not on the loadings.

If you have four different bond rates, wouldn't you most likely have three cointegrating vectors? That's quite different from the Enders-Siklos model. But basically, you're looking a running a VECM with a fixed set of cointegrating vectors and doing some type of Chow-type test on that. If you're using known break points, you don't have the problem with non-standard distribution.

Thank you very much, Tom!

Best regards,

Onem