Blanchard Kahn conditions are not satisfied: indeterminacy d

I replied at Error: Blanchard & Kahn conditions are not satisfied: no stable equilibrium - #2 by jpfeifer

Hi Prof. Pfeifer

I have encountered the same problem “Blanchard & Kahn conditions are not satisfied: indeterminacy.”

I tried to run a RBC model with capital adjustment cost in the spirit of Aguiar and Gopinath (2007). I introduced investment-specific technology shocks and omitted trend productivity shocks from that paper. Here’s the file with complete derivation of model equations and steady states.
RBC model_capital_adj.pdf (573.5 KB).

And you can find the mod file here;
my_rbc_cap_adj.mod (2.9 KB)

Please note that I followed your code on Aguiar and Gopinath 2007. I’d appreciate your feedback.

Best regards

Why does your steady state depend on investment adjustment costs? They should drop out in steady state.

Hi Prof. Pfeifer

Thanks for your feedback. I managed to drop the capital adjustment cost from the steady state by redefining it. In addition, I had an issue in the resource constraint. With these two corrections, the model now works. The mod file is here
my_rbc_cap_adj3.mod (2.8 KB)

Then, I augment the model with capacity utilization. But, when running the model I faced an error saying that "the steady state file cannot compute the steady state.

I see residuals in the equation of FOC for capacity utilization only. For your reference, I upload the mod file here.
my_rbc_cap_adj_util.mod (3.0 KB)
Could you please help me find the sources of errors?


How do you make sure that capacity utilization really is 1 in steady state? Usually, that involves setting a parameter.

DOEL_WithoutE.mod (2.7 KB)

Dear Prof. Pfeifer

I have encounteresd problems with the Blanchard & Kahn conditions, receiving the error message `There are 6 eigenvalue(s) larger than 1 in modulus
for 5 forward-looking variable(s)'.
I am working on a Small Open New Keynesian Model, where I leave monetary policy aside in a first analysis. Thus, there is no Taylor rule. The Steady-State seems so be correct and there are no residuals in the static equations. However, there must be a problem with the time index in (at least) one equation. Even after hours of double-checking, I was, however, not able to spot a mistake. I would therefore greatly appreciate if you could take a look and help me find the source of error.

Thanks a lot.

What makes sure in your model that b does not explode?

Dear Prof. Pfeifer

Thank you so much for your reply - it is highly appreciated. I once again thought about the constraint of the government and redefined it as in line 65 in the Dynare code, i.e., that the debt level of the government must equal the difference of the subsidies and the tax level in each period. Dynare is giving me IRFs now. However, I find them a bit strange, especially for the technology shock (where a positive, exogenous increase in technology decreases output). Do you see any obvious mistakes?


DOEL_WithoutE.mod (3.3 KB)

Sorry, I don’t see anything immediately suspicious. Usually, the way to check is to drop features to see where the issue appears.

Hi Prof. Pfeifer

I solved the issue. Then I augmented the model and ran it successfully. The mod file is available here.
TFP_IST_trend2.mod (5.7 KB)

Then I proceed to do exp() substitution. Unfortunately, the steady state file could not compute the steady states. The file is here:
TFP_IST_trend_exp.mod (6.2 KB)
Could you please guide me to spot the mistakes?


Why would you even want to to an exp()-substitution? Append auxiliary equations and variables.

The purpose of this exercise is to compare the results from log linearized model derived by hand and see if I could do the log-linearization correctly.

But for such a check, you typically don’t need a full substitution, only the most important variables.

Thanks! But I have log-linearized all model equations. Will this log-linearized model be consistent with partial exp() substitution model? Could you please give me a bit more hints in this regard?

What do you mean with consistent? I still don’t understand what you are trying to do in the end.

First, I want to stoch_simul the exp() substitution model. Then I want to do the same with the log-linearized (by hand) model and compare the results with that of the exp() model. According to my understanding, the results should be similar.

Which results are you talking about that you want to compare?

Theoretical moments, variance decomposition and IRFs etc.

Then again,


I’ll do. Thank you!