Euler equation in heterogeneous expectations model


i want to simulate the Branch & McGough (2009) model for my master thesis. There are rational agents with fraction alpha and “adaptive” agents with fraction (1-alpha) in the model. In order to simulate the model with an optimal targeting rule i derived (not in the code here), i need to have the consumption of the rational agents (cr) as a further variable. In order to do so, i included their euler equation as follows:

//Phillips curve
pi = alpha*beta*pi(+1)+(1-alpha)*beta*(theta^2)*pi(-1)+kappa*y+cps;  

y = (alpha*y(+1)+(1-alpha)*(theta^2)*y(-1))-sigma*(i-alpha*pi(+1)-(1-alpha)*(theta^2)*pi(-1));

//Consumption euler of rational agents
cr =cr(+1)-sigma*(i - pi(+1));
//Taylor rule
i = phi_pi*pi + phi_y*y;

//Cost push shock
cps = rho_cps * cps(-1) + e_cps;

However, dynare returns that there are eigenvalues greater than 1 in absolute value. Without the euler equation the simulation works just fine. I probably have to miss something very obvious here. Can anyone help with that, please?

What exactly is the error message. Introducing an Euler equation introduces a forward-looking equation. For that equation, you will need an unstable root (eigenvalue bigger than 1) to rule out indeterminacy. If that is the problem, there is typically a timing problem somewhere in the model.