Correlated shocks

Stochastic simulations
1

Dear all,

BKK_1992.mod (3.3 KB) I am working with a standard BKK model (international RBC model) with 2 countries and 2 agents in each country. I have 4 idiosyncratic shocks to productivity : one for each agent in each country (code is attached)

I want to specify negative correlations between these shocks but I get this error:

MATRIX OF COVARIANCE OF EXOGENOUS SHOCKS
Variables eps_z1h eps_z2h eps_z1f eps_z2f
eps_z1h 1.000000 -0.500000 0.250000 0.250000
eps_z2h -0.500000 1.000000 0.250000 0.250000
eps_z1f 0.250000 0.250000 1.000000 -0.500000
eps_z2f 0.250000 0.250000 -0.500000 1.000000BKK_1992.mod (3.3 KB)

"Error using chol
Matrix must be positive definite.

Error in simult (line 77)
chol_S = chol(DynareModel.Sigma_e(i_exo_var,i_exo_var))"

Thank you very much.
Emilio

2

The problem is not the negative correlations, but the fact that your correlation matrix has a zero eigenvalue, i.e. linear dependence in the columns.

3

Perfect, Thank you very much Johannes.

Emilio