Limited participation models and Dynare

Dear colleagues,

I am working with a LP (limited participation) model in Dynare. The LP mechanism states, that the households have to decide about their nominal money holdings prior to the realization of all time t shocks.

Technically I have:
E_(t-1) f(M(t),B(t),…)],

where M(t) are the previous mentioned money holdings and f() is a nonlinear function of its arguments.
To translate this equation in Dynare, I changed the timing for the money holdings (t -> t-1) in every equation. Afterwards, I shifted the above equation in time to end up having:

E_t [f(M(t),B(t+1),…)].

Unfortunately, this transformation does not work, because the money holdings deviate from their steady state path in the very first period after the shock occurs. Since they are decided on an old information set, which does not include this shock or any time t variable, this deviation is false.

Does anyone of you know how to solve this problem?

Thanks in advance