e(Sigma)

[jean bizi]

Hello
I am running a VAR and getting covariance matrix of the errors instead of correlation matrix using matlist e(Sigma)

matlist e(Sigma)

| pkay pbko pnior pseg psan psik
-------------+---------------------------------------------------------------------------------------------------
pkay | 154.6524
pbko | 43.54131 111.5155
pnior | 50.91939 52.6377 290.7672
pseg | 48.8736 79.47023 58.04128 151.4966
psan | 24.52065 61.51676 27.54746 86.41532 138.0055

[George Ford]

That is not an error. e(Sigma) is the covariance matrix of the errors.

[jean bizi]

Thanks. Yes I agree but I want to get rather the correlation type of the standard errors for a better interpretation of the linear correlation.

[jean bizi]

Does anyone have an idea of how to convert the Cov-matrix into the Correlation matrix?

[George Ford]

matrix A = (1,1,8.1 \ 1,16,18 \ 8.1,18,81 )
mata st_matrix("D",sqrt(diagonal(st_matrix("A"))))
matrix D = diag(D)
matrix Di = inv(D)
matrix R = Di * A * Di
matrix list R

Taken from:

https://blogs.sas.com/content/iml/2010/12/10/converting-between-correlation-and-covariance-matrices.html#:~:text=You%20can%20use%20similar%20operations,reciprocals%20of%20the%20standard%20deviations.

[jean bizi]

Thanks George. I just used a simple command "Correlate" with the appropriate varlist and it seems to work.