People are used to call this extrapolation, it has nothing to do with prediction: One fits the data of the last thirty years to some reasonable model and evaluates the model afterwards outside of the given data range (to the next twenty years). So, how would you call it? It's speculation, extrapolation, fiddling with the unknown. It would  not be speculation if and only if the data must follow the model, i.e. if the model would in fact represent a law. But in order to constitute a law the data must be reproducible and nobody can reproduce the last thirty years.