Björn,
I don’t know this model and I don’t have time to search and read the paper by Krause and Lubik… You have eight jumping variables but only seven generalized eigenvalues greater that one. The problem may be related to an error in the timing of the variables. A predetermined variable X at time t have to be written X(-1) in dynare (this variable is choosen in t-1).
If I change , in the production function, n (employment??) by n(-1), dynare solves the model without trouble… But, again, I don’t know this model…
Best, Stéphane.