Bug in positive/negative IRFs in Dynare++

See https://forum.dynare.org/t/asymmetric-irfs-at-first-order-in-dynare/12246/4 The issue seems to have not yet been solved.

assigned to @sebastien

Using bisection, I could determine that this bug has been introduced by commit b05c7356.

closed via commit 8698b4c5

@sebastien Does this affect all IRFs? Of course at first order, the problem does not arise unless one considers the IRFs to positive and negative shocks. But does this also mean the normal IRFs at higher order were wrong?

My understanding is that it affects all IRFs. But only in Dynare++, not in Dynare proper.

Actually even at first order the IRFs were wrong, because they were also computed using GIRFs.

mentioned in commit 583e2dfc

At first order, due to linearity, the IRFs should still be correct if you are using a sufficient number of replications. You have one shock happening at time t and the other shocks, both in the shocked series and the incorrect baseline should average out to 0.