Hi DSGE solvers

I’m trying make SOE DSGE model. But, I cannot solve NaN/Inf problem…

I need your advice that how can I fix this.

Hi DSGE solvers

I’m trying make SOE DSGE model. But, I cannot solve NaN/Inf problem…

I need your advice that how can I fix this.

Please try a calibrated version of the model first. See Remark 2 (Using stoch_simul before Estimation) in Pfeifer(2013): “A Guide to Specifying Observation Equations for the Estimation of DSGE Models” sites.google.com/site/pfeiferecon/Pfeifer_2013_Observation_Equations.pdf.

Focus on equation 17 and the model-local variables involved in it. If I understand it correctly, you have

```
#rsss=1/(ebs*beta);
#rlss=(rsss*ebs)/ebbarl;
```

But ebs is an exogenous variable that is mean 0, so you are dividing by 0 and rsss will be infinity. rlss then has infinity divided by infinity, resulting in NaN

Thank you very much, Proffessor !

I made a easy mistake. And, fixed this.

But, I have new problem that colinear variables/equations.

I read much Q’s in this forum, but I don’t get it well.

How can I fix this problem for example?

many thanks,

Cross-Country Co-movement in Long-Term Interest Rates Appendix.pdf (325 KB)

Cross-country co-movement in long-term interest rates a DSGE approach.pdf (675 KB)

sot2.mod (12.8 KB)

That is hard to tell. Usually, there is one equation missing that needs to replace one of the redundant equations entered in the mod-file. Alternatively, there may still be a (timing) error in your model equations.

Hi Professor Pfeifer.

Why dinare divides by the mean value, and not by the current value of exogenous variable?

Thank you very much in advance.

Best,

Leonid

@Leonid_65 What do you mean? Dynare does not divide by anything unless you specify something.

I mean your answer "Focus on equation 17 and the model-local variables involved in it. If I understand it correctly, you have

#rsss=1/(ebs*beta); #rlss=(rsss*ebs)/ebbarl;

But ebs is an exogenous variable that is mean 0, so you are dividing by 0 and rsss will be infinity. rlss then has infinity divided by infinity, resulting in NaN"

Best, Leonid.

For solving the model with perturbation, Dynare evaluates the derivatives at the deterministic steady state. That is where the problem pops up. You cannot divide by a mean 0 variable.