Crisis event study

Hi all,

Is it possible to do a crisis event study using Dynare? The example that I have in mind is Mendoza2010 Figure 2 and Akinci-Queralto2022 Figure 6. They simulate the model for long horizon and identify the crises. Then they look at the window around the crises as in the figures. This is somewhat different from impulse response analysis.

Is this something one can do with Dynare? Or one needs to resort to the global method?


  1. The big question is whether the model can be solved in Dynare with sufficient accuracy. Whether you need global methods will depend very much on the model.
  2. The event study exercise can be conducted with any program that allows you to simulate the model, including Dynare. Once you have the simulations, you simply need to identify the events using the same technique you used for the actual data.

Thank you @jpfeifer for your reply!

Regarding your point 2, can I ask if this way would work? What I have in mind is:

  • Specify y0 and shock_matrix for options_.irf periods
    • options_.irf: # of simulations
    • y0 is the steady state
    • shock_matrix: options_.irf by M_.exo_nbr matrix with random realizations from normal distribution
  • Run simult_(M_,options_,y0,oo_.dr,shock_matrix,1); and get the endogenous variables
  • Identify crisis events used for the actual data


Yes, that is the way to proceed.