interesting bug in dyn_v3.65

[Bene]

Hi,

either I find an interesting bug in dynare (v3.65), or I'm doing something wrong and I can't see it.

I've have a standard RBC model with three capital assets and three shocks: two Investment-specific shocks and the usual TFP shock.
The shocks are AR(2) processes, and their coefficients come from a previously estimated VAR. In the VAR some coefficients are constrained to 0, so the corresponding terms do not appear in the AR(2) equations. Hereafter you may give a look to the .mod file.

I get the following error (see below), which however disappear if I add an extra variable - and its corresponding equation - to the model. This is interesting because the variable is a dummy affecting no other equation of the model, e.g. I add "dum" which is just the sum of wage and interest rate. Notice that the system is stable and the rank condition is satisfied.
Also , I noticed that if I had an extra term to the process of the shocks I need to add more variables to the model in order not to get the error...

The error is:

??? Subscript indices must either be real positive integers or logicals.

Error in ==> dr1 at 139
b = jacobia_(:,k0);

Error in ==> resol at 56
[dr,info] = dr1(dr,check_flag);

Error in ==> check at 18
[dr, info] = resol(ys_,1);

Error in ==> ict2 at 134
check;

Error in ==> dynare at 26
evalin('base',fname) ;

[SébastienVillemot]

Hi,

Version 3 of Dynare is no longer supported. We won't make any bugfix to it.

You should upgrade to version 4.2, it has more features and is faster.

Best,