The RATS Software Forum

Dear All,
How do we estimate the error correction model from the output of the EndersSiklos procedure (as shown in Enders W., Siklos P. L.,(2001) "Cointegration and threshold adjustment", Journal of Business & Economic Statistics, 19(2), 166-176). Compare to enders and granger proc, we can retrieve z_plus and z_minus but in enderssiklos proc, retrieving the xsplit(1) and xsplit(2) does not work. I really hope you can assist me on this matter. Your assistance is highly appreciated. Thank you.

I'm not sure what you're looking for. The Enders-Granger procedure does all the regressions with actual data (y, its lag and its differences), while Enders-Siklos does everything in terms of residuals from a first stage Engle-Granger regression, so the mapping from that to the original data series is a bit tedious.

Hi Tom,

I guess I have the same question. In the example for Enders and Siklos (2001), the paper replication code programs everything up to Table 7 in their article.

I am interested in the programming to get the final error-correction models (Equations 20 and 21) on page 174.