Dealing with I(1) variables

Hi Guys,
I have put my query in the .pdf file attached.

2-country model.pdf (44.7 KB)

Just two things before I look at the PDF file.

  1. what is the origin of nonstationarity in the model?
  2. Don’t use Dynare version 3 with I(1) variables, there is still a bug. Version 4 is OK as far as I know

I will answer later to the detailled questions




you asked that
"what is the origin of nonstationarity in the model? "

The model I have in the .pdf file is a simplified version of the 2-country 2-money 2-good exchange economy that is a simplified version of the model I actually I intend to simulate. But I did not want to bother you with non essential details.

Consumption and nominal money growth processes are exogenous and I(0). While the intertemporal Euler and money demand equations have the levels of consumption, money and price I have scaled the variables to make them stationary. However I need to back out the price levels in both countries and some of the equations do not need to be approximated.

I have been using Harald Uhligs toolkit for years and my own (relatively poorly programmed) version of LQ methods in RATS. The beauty of your Dynare program is you can write the equations as you would normally write them in a paper and your code takes care of the rest.

The queries I have made in the pdf file are more about how to extract simulated variables after the stoch_simul command. I have included a subsection of the relevant Dynare code and I am not quite sure I am programming waht I require correctly. I cannot tell from your documentation. I taught that you or a member of your team would probably spot the error of my ways immediately and point me in the right direction.

Any help is appreciated and i do understand if you cannot devote much time to my query.

Thanks for your time.


Hi Maurice,

I don’t understand entirely what you want to do that and I didn’t verify if equations (7) to (12) will work on their own. But, assuming they do

  1. in Dynare version 3, you can retrieve the simulated variables in Matlab vectors with the same name as the variables
  2. the exogenous shocks are in a matrix called ex_, the shocks are in the columns, in alphabetical order of their name.
  3. In order to cumulate the I(1) variables from their first differences, you need Matlab code that can be put in the *.mod file after stoch_simul. But matlab doesn’t know about leads and lags and you can’t use syntax as c(-1).
    Do for example, c1 = cumsum(dc1)+c0, where c0 is the initial value of c

Kind regards


Thank you for the advice.