Hi everyone,

May I kindly ask about the correlation between components of GDP (simulated in Dynare).

For a simple model with equilibrium condition, `Y = C + I + G`

(and even for complex models), one can deduce from the IRFs whether the simulated series (`C, I, G`

) will be procyclical or countercyclical (with respect to output). However, I have not been able to do the same for correlations between the components of `Y (i.e., C, I, G)`

.

Sometimes simulated `C and I`

are both procyclical (with respect to output) but the correlation between `C and I`

is either positive or negative.

Actually, this also true in actual data. For some economies, there is a positive co-movement between `C and I`

, with both `C and I`

being procyclical (with respect to output). In other economies, there is a negative correlation between `C and I`

, although `C and I`

are both procyclical.

Jespersen & Madsen (2013) have called the correlation between `C and I`

(which can go either ways in actual data) a paradox since theory predicts that `C and I`

should move in opposite directions.

Is this paradox somehow embedded in dynare such that correlation between, for example, simulated `C and I`

is arbitrary, and hence can either be negative or positive?

My instinct tells me it is not arbitrary, but I have not been able to see that from other parts or other results of the model.

Any help to clarify this is very well appreciated. Many thanks!!!