I am trying to solve a simple, two-country model with incomplete markets - i.e. a single bond. I am not sure what I am doing wrong but Dynare cannot solve the model - BK conditions not satisfied. I believe the model is well specified and the steady-state is solved. What is interesting is that if I set the bond-clearing condition as: B1(-1) + B2(-1) = 0 ; (which I believe is the correct way to set it since bonds is a predetermined variable) the model fails the BK conditions. If I set this to B1+ B2 = 0 ; the model is solved but I am a bit skeptical about the results.

Shooting from the hip, I would argue that the second one is the correct market clearing condition. On some level they are equivalent, of course. But at time t, a feasible allocation must satisfy

While

needs to hold as well at time t, it is a condition on past choices that does not constrain the choice of B1 and B2, because the values were decided on in the past.

Thanks a lot! To avoid confusion, I have replaced B2 with -B1 and then the equilibrium condition becomes redundant. It is nevertheless good to have these discussions to enhance awareness on Dynare.

If I may ask, how would you interpret the extreme sensitivity of the results to the elasticity of substitution (parameter theta_1) especially at values 0.35-0.45 (i know this is a very important parameter in IBC models). Note that at these low values the model attains a unit root. Is that the reason behind these big differences in the results, despite small movements in the parameter? Also, why is there a unit root? Assuming the model is well specified, is that a pathological case? Note that at low values of theta, omega (the degree of home-bias) gets pretty close to 1 (0.98-0.99…)