The NK-DSGE Model has an odd IRF

General DSGE Modeling
1

I build a basic NK model, but the IRF is odd in time 1, it’s not smooth.

If I change it to a RBC model, or change the production function Y = A * K^ alpha * N^(1-alpha) to Y = A * N, the IRFs are OK.

I’m just stuck with this problem for days, and can’t find a solution by myself, need some help, thank you!

NK.mod (1.1 KB)

00
00588×990 24.4 KB

01
01718×636 35.5 KB

02
02718×636 30.9 KB

2

I had the same question.My TRF is also not smooth.Waiting for answers.

3

untitled.fig (142.4 KB)

4

Hope for your help.
@jpfeifer
Thanks！

5

@DAXIA Your results look very strange. Is this a first order approximation?
@jordima I don’t immediately spot anything strange. Are you sure this is not just regular economic behavior without investment adjustment costs?

6

Thanks for your reply. @jpfeifer

I’m confused about why it only occurs in time t+1, and then it returns to normal?

What’s the economic implication?

7

It seems because most adjustment is done after that period due to investment and hours jumping.

8

Thank you very much!

It’s interesting that almost all of the basic NK models are built without capital, maybe they came up with the same problem.

So in order to get a normal IRF, we must augment the model to a medium scale DSGE model?

9

Pretty much every model fit to the data that includes capital has some form of adjustments costs. The missing empirical hump-shape is often given as the reason for that.

10

Thanks for your kind reply, it really helps.

11

Hi Jordima!
May I ask did you solve this problem (NK model with capital stock but having an undesirable IRF)?

I have made a similar model just now and confronted the same problems. And I am also wondering why most of the NK models without capital stock and any forms of investment.

I added investment adjustment costs into the model and got an undesirable IRF.

12

@jpfeifer
Dear Professor Jpfeifer,
I have added investment adjustment costs into NK model, while it didn’t make any difference. May I ask how to improve this situation?

NK1k11.mod (2.1 KB)

3
31071×393 12.7 KB

Regards,
Modesty

13

I think your FOC is wrong. There should be a linear term showing up.

14

Thank you for your reply, Professor Jpfeifer!

May I ask do you mean I need to log-linearize all the equations and then put them into dynare?

15

No, I am saying your nonlinear FOC has wrong terms/exponents. Check them again.

16

Professor Jpfeifer, I have found a mistake in the dynare code. It should be rk(+1) instead of rk. And this is the new result.

I think, at least the result is more reasonable. That interest rate rises causes the decrease of output and inflation. What do you think, may I ask?

But I wonder why IRFs are still not so smooth…
4
NK1k11.mod (2.1 KB)

@jpfeifer Professor Jpfeifer, may I ask do you think it could be more improved?

17

There must still be a mistake somewhere as changing the investment adjustment cost parameter does nothing at all.

18

@jpfeifer
3

Professor Jpfeifer, it seems that I have solved this problem.

Your reminding is very important to me. I tried to change the objective function of intermediate firms, and the IRFs turns much smoother, and they conform with the expectation.

There are two points, I think.

  1. If a slow convergence speed is needed, then letting the household contain C-gamma*C(-1) might be useful.

  2. To let the intermediate firm as the sector of accumulating capital stock (and there is an adjustment cost of investment) might be useful for making IRFs smooth