Running Maximum


I have a model where one of the variables is simply the “running maximum” of another, so:

D_t = max(d0,…,dt)

Is it possible to include such a variable with stoch_simul?

Many Thanks

No, that is not possible, because the max operator will introduce a non-differentiability that will be smoothly approximated.

Many thanks for your reply.

Would it be any different if I wanted:

D_t = max(d0,…,dt-1)

That is to say, apply the max operator to only historical variables that dynare has already solved for.

What I mean is, I appreciate the dyanre solution approach can never deal with a max within the general model, but is there a way to access stored values and then create, in some sense, a parameter to be applied each period?

No that will not work either, because what you suggest would make the model time-dependent. Decision rules would now depend on the number of previous periods.