Hi all,

I have a question. When I am building DSGE models, I always find that if I use talor rules monentary policy like rn- log(rn_ss)= rho*( rn(-1) - log(rn_ss) ) + rho_pi*(pi_H(-1)-log(pi_H_ss)) + e_rn ;

, the mod runs well. But when I use **PEG policy: de = 0**;, Dynare tell me “All endogenous are constant or non stationary, not displaying correlations and auto-correlations” or “BK condition is not verified”. Can someone tell me why it will happen? Thanks.

Without seeing the files it is hard to tell.

Of course, pardon me professor. Here is my modmodel.mod (12.7 KB) . I am trying to build a two-coutry model with financial accelerators. What I found is that, when I let coutry M follow talor ruler monentary policy and country H follow PEG policy, dynare tells me “All endogenous are constant or non stationary, not displaying correlations and auto-correlations”. But when I do the opposite, letting country H follow talor rule while letting M follow PEG, it works. (besides, both countries follow talor rules policy also works)

So I wonder why this will happen.

Can you please provide the version that generates the message you quote?

Professor, this is the one that generates the message I quote. model.mod (12.7 KB) . When I let country H follow talor rule while letting M follow PEG, it works. But when I let coutry M follow talor ruler monentary policy and country H follow PEG policy, it has the above problem.

Your only shock does not even enter the model. No wonder, nothing moves.

Sorry, professor, I am comfused.

What you mention seems likely to happen in small open economy models. I have learned that in two-country models, the countries is symmetrical. So in my comprehension, no matter which country use PEG policy ( and the other one use talor rules), the mod should give the same result, but actually it isn’t. (maybe there is some equations in mod are asymmetric ?).

***By the way, is this problem a feature of two-country models, or just something wrong with my model? Because I find some models in some preofessors’ webpage have the same problem.

Again, thank you for your kind answer very much.

You have

```
shocks;
var e_rn_H; stderr 0.1;
% var e_rn_M; stderr 0.1;
% var e_a_H; stderr 0.1;
% var e_a_M; stderr 0.1;
% var e_g_H; stderr 0.1;
% var e_g_M; stderr 0.1;
end;
```

so `e_rn_H`

is the only shock with non-zero variance. But in the model

```
//54-55. Monetary policy;
% rn_H - log(rn_H_ss) = rho_rn_H*( rn_H(-1) - log(rn_H_ss) )
% + rho_pi_HH*(pi_HH(-1)-log(pi_HH_ss)) + e_rn_H ;
de_HM = 0;
rn_M - log(rn_M_ss) = rho_rn_M*( rn_M(-1) - log(rn_M_ss) )
+ rho_pi_MM*(pi_MM(-1)-log(pi_MM_ss)) + e_rn_M ;
```

so the only equation containing that shock is commented out. Thus, the shock does not appear in the model at all. If you had used Dynare 4.6.1 you would have gotten a notification.

Of course… Forgive me professor, it’s a stupid mistake The next time I will check it more carefully.