Bartolomeo, Di Pietro, Giannini (2016) Optimal monetary policy in a New Keynesian model with heterogeneous expectations

Dear community,

iam working with the above mentioned paper. The authors have compared the gains of commitment over discretion and how heterogenous beliefs affect the gains. They said that " a larger share of non-rational agents involves higher welfare losses…, since the effects of the shocks become more persistence."
They also explained the persistece effect and dispersion effect as reason for the statement above.
The model in the paper contains the hybrid PC.

My question is now:
I simulated the hybrid NKM with bounded rationality. I used alpha as the bounded rationality paramter. In my simulations alpha has the following values: 1, 0.5, 0 and 1.01.
When I compared my results to Bartolomeo et al. I have determined that my loss values do not agree with this statememt. In my case it is as follows: a larger share of non-rational agents involves smaller welfare losses. I get this result in all my 4 simulations. I checked my results , because I was unsure if Iam wrong maybe.

What exactly is the question? You are trying a replication of the paper and cannot reproduce their results?

No, I dont replicte the paper. I am doing my own simluations with different parameters. My question is about the statement of the paper: The loss increases with a larger share of non-rational agents.

My results dont confirm this statement. The explanation of the authors makes sense and that is why Iam confused. They also say " an increase in the share of boundedly rational subjects has an ambiguous effect…it reduces the lead component of the PC"
The lead component is the yellow marked term:

As I understood when the yellow marked term descreases, the loss will increase.
My question is now: Is it possible to have a smaller loss when the share of non-rational agents increase?

That seems like a question one cannot generally answer. It is not obvious how changes in the model affect welfare results. You are the model builder. You need to find out what is going on. If I were you, I would start with a replication and the move step by step to your own model to see at which step the change in welfare occurs.