I have analyzed a table in Rstudio using the VECM function from the vars library.

I believe to have understood from the theory that a VEC model makes use of vectors and the only matrices are coefficients.

[At , Bt] =

[Ca , Cb]

+ SUM(p->lag) GAMMAMATRIX * [dA t-p , dB t-p]

+ PIMATRIX [A t-lag , B t-lag]

+ [Aerror, Berror]

This leaves a single error term per variable. So one vector.

However, when I run the VECM function in rstudio, the resulting data object has as a number of ECT columns equal the r-value in the initial argument, helpfully numbered ECT1, ECT2, etc.

The toLatex function presents these ECTs together in a matrix, creating a regression I don't fully understand (does the ECT matrix get multiplied by the intercept vector?)

This runs counter to everything I thought to know about VEC models.

I have attached both the summary() and toLatex() results of the object to this question.