Questions about M_.Correlation_matrix

Dear all,

I am a newcomer to Dynare and I am studying how to specify the variance and covariance of exogenous variables in a model. Dynare Reference Manual provides two ways of specifying the covariance/correlation of two exogenous variables as follows.

var VARIABLE_NAME, VARIABLE_NAME = EXPRESSION;
corr VARIABLE_NAME, VARIABLE_NAME = EXPRESSION;

Accordingly, I tried to replace

var e, u = phi0.0090.009;

in Example1.mod from Resources | Dynare with

corr e, u = phi;

While the variance-covariance matrix M_.Sigma_e of the shocks is the same under the two specifications, the matrix M_.Correlation_matrix are different. Under the first specification, M_.Correlation_matrix is the 2-by-2 identity matrix, while under the second specification, it is [1 phi; phi 1]. I am wondering what exactly M_.Correlation_matrix is. More generally, what are the differences between the two specifications?

Many thanks,
Louis

M_.correlation_matrix stores prespecified correlations that need to be preserved. When you change the standard deviation of shocks, the covariance in M_.Sigma_e needs to be altered accordingly.
In contrast, if you specify the covariance with var e,u there is no correlation to be preserved. Rather, the entry in M_.Sigma_e is estimated directly.

Dear Prof. Pfeifer,

Thank you for your swift reply. It really solves my problem.

Best regards,
Louis