Forward looking vs. predetermined variables

I have a question related to the timing convection in Dynare. I have been working with a search model in which employment evolves as follows:

n_{t} = f(n_{t-1}) + g(u_{t-1},v_{t-1})
where n is employment, u and v denote unemployment and vacancies, respectively. In my understanding, employment is a predetermined variable in my case. Is this correct? If so, do I need do specify employment to be a predetermined variable?

As far as I know (Forward looking & predetermined variables), the following formulation is not identical:
n_{t+1} = f(n_{t}) + g(u_{t},v_{t})


  1. The way you set up the first equation, n_t will be predetermined. You can see that the variables on the right that determine it are all dated t-1. If you use that stock at the end of period timing timing, you do not need to define n as a predetermined_variables. If you were to use this, the timing would be shifted by Dynare, which is not what you want.
  2. You may also want to have a look at End of period convention and decision lag
  3. No, the second formulation is not identical as the left hand side is now E_t(n_{t+1}), which is not the same as n_{t+1} in a stochastic model.