How to simulate calibrated DSGE model with data..

[leolee]

Dear all,

I have two questions. It may be simple for experts.

Question 1)

I have an already calibrated DSGE model, and have 100 data point (y, pi, R).

Code: Select all

   # kappa = (1-theta)*(1-theta*beta)/theta;
      
      y      = y(+1) - (1/sigma)*(R - pi(+1)) + g;
      pi      = beta*pi(+1) + kappa*y + z;
      R       = rhoR*R(-1) + (1-rhoR)*(phipi*pi(+1)+phiy*y) + em;

      // process of preference/cost shocks   
      g       = rhog*g(-1) + eg;
      z       = rhoz*z(-1) + ez;


I'd like to get an expectation for R(up to +40 periods) at each 100 data points.

For example,

at t=1, R_1, E_1(R_2), E_1(R_3).. E_1(R_40)
at t=2, R_2, E_2(R_3),.... E_2(R_41)
..
at t=100. R_100, E_100(R_101)... E_100(R_139)

How can I do this with dynare? Could anyone just tell me which part of the manual should I read?


Question 2

How can I get a simulated values for y, pi, R using the model above and the actual data, so that I can compare the second moments of two series? Could anyone just tell me which part of the manual should I read?

Regards,

Leo.

[jpfeifer]

1. You need to define the expectations as a new variable, e.g. the one period ahead expectation as

Code: Select all

E_plus_1=R(+1)

For fourty variables I would recommend using the macro-processor (the doc-folder of Dynare has a separate pdf on this)

2. As you are apparently not estimating your model, you should go for the calib_smoother command.