I am currently thinking about combining the extended path method (extended_path) with exogenous deterministic variables (varexo_det). It seems to me that this combination should be feasible from a theoretical point of view. However, either I am mistaken, or Dynare doesn't work for this combination.
If I run a small example (the one from here: http://www.dynare.org/phpBB3/viewtopic.php?f=1&t=5951, file attached below) with the same variables, model, etc., but apply extended_path instead of stoch_simul, Dynare or Matlab quits with an error. While calling the _dynamic file in the perfect foresight part of each extended path iteration, Dynare tries to access the values for deterministic exogenous variable (x(it_, 3) in the quote below, or e_det in the example), but it seems that the corresponding matrix (x) does - my guess - not include these series, but only those for the 'regular' stochastic variables (e and u in the example). Formally, Dynare or Matlab returns:
Attempted to access x(2,3); index out of bounds because size(x)=[202,2].
Error in example2_A_dynamic (line 64)
rhs =params(2)*y(2)+params(7)*y(3)+x(it_, 1)+x(it_, 3);
Error in solve_perfect_foresight_model (line 41)
[d1,jacobian] = model_dynamic(z,exo_simul,pfm.params,pfm.steady_state,2);
Error in extended_path (line 203)
[flag,tmp,err] = solve_perfect_foresight_model(endo_simul_1,exo_simul_1,pfm);
I'm very grateful for any help or thoughts in either direction - whether I am mistaken theoretically or Dynare doesn't like this combination (yet). And apologies if I should have missed a similar post.
Best
PS: This quote is based on unstable 4.4.4 (most current version of ZIP file), but the problem also appeared in stable 4.4.3.
