Third-order approximation generates wrong IRFs

Stochastic simulations
1

nk_price_wage_stickiness.mod (6.2 KB)
Dear All,

When I solve a New Keynesian DSGE model with price and nominal wage stickiness, it generates correct IRFs to a monetary policy shock up to a second-order approximation. However, when I use a third-order approximation, it generates wrong predictions. For example, output rises in response to the real interest rate increase.

Does anyone know why that might happen?

(I’m using a third-order approximation to generate IRFs to a technology uncertainty shock. I considered a monetary policy shock to check whether IRFs in the simulated model make sense.)

I have attached the code, whose basic structure and equations can also be found on Eric Sim’s page.

I appreciate any comments and help.
Best,

2

Dear All,

I’m still struggling to understand why the third-order approximation doesn’t work in this model. Any help is appreciated.

Thank you,

3

Decision rules at higher order are not certainty equivalent. The standard deviation of the shocks you are considering makes no sense. They are orders of magnitude off.

4

That was helpful. Issue solved. Thanks!