Hi, my model has one problem about shocks. I set some temporary shocks on some variable, but the result indicates the shocks did not work. I can not find out what is wrong with the file. Could somebody instruct me with this?

Thanks a lot!

You are using the perfect foresight syntax for setting shocks in a stochastic simulation context. That cannot work.

I see. Thanks, professor Jpfeifer.

Prof Jpfeifer, I have another qustion about this model. After running the model, I find that the figures are not smooth, but have a steep increase/decrease at the end of the period.

I am very confused about the result.

The new program and result are attached below.

Thanks again!!

result.fig (17.4 KB)

myfile2x.mod (5.6 KB)

myfile2x_steadystate.m (2.3 KB)

Why does your model not satisfy the Blanchard-Kahn conditions? Note that your law of motion for capital looks strange. Why is investment dated t+1?

Professor, the original version of my capital motion equation was k= delta*k(-1)+i. But the model diagnostics indicated k is not present. So I changed capital motion equation as k(+1)= delta*k+i(+1).

What surprised me is that even though the problem of B-k condition, the simul works.

- Your previous timing was correct, but incompatible with your
`predetermined_variables k`

-command. If you use the stock at the end of period timing, then that command is wrong. -
`simul`

solves a nonlinear equation system given an initial and a terminal condition. That may have a solution even if the BK conditions are not satisfied. The problem here is that you are providing the steady state as the terminal condition. But if the BK conditions are violated, then the dynamics around that steady state are locally explosive. Your system will not return to steady state, so the solution is wrong.

Thanks, professor. I think I have understood the simul problem. So if I want to avoid the explosive problem, should I set an endval block?

About the timing problem, I set the stock at the beginning of periods, that is why the production function is y = A*(k(-1)^alppha)*(N^(1-alppha)). Is not that means k is predetermined_variable? I am really confused about that (I have been reading the handbook sometimes).

The `predetermined_variables`

shifts variables back by one period to transform the stock at the beginning of period notation into the stock at the end of period notation required by Dynare. But you already used the latter timing, so having the command to shift it back by one additional period is wrong.

I see. Thanks again and again!!