As discussed in the previous chapter, system dynamics models its target systems as indivisible wholes and does not take into account the fact that for the social scientist target systems usually consist of individual persons, groups, classes, subpopulations and so on. Social scientists will therefore be interested in modelling approaches on several levels – an aggregate level and at least one lower level. The first approach that tried to solve this problem was the classical microsimulation approach. It has been used to predict the individual and group effects of aggregate political measures that often apply differently to different persons. For instance, a tax formula that imposes taxes only on persons with incomes above a certain threshold might be changed by moving this threshold. If we want to calculate the gross effect on the total tax revenue, a simulation on the macro level cannot help. We must instead go back to the individual cases, calculate their taxes due before and after the tax revision, and reaggregate the tax revenue.
Another example can be taken from demography. Changing age structures of a population can be simulated on a macro level – see the discussion on page 53 in Chapter 3. We would have several level variables with the sizes of a number of sex/age groups to which we would apply age-group-specific death rates, and we would calculate births from the sizes of the female age groups between 15 and 45 years of age, applying age-dependent fertility rates. If we were only interested in the age structure of a population, this deterministic macro model might be sufficient because, with a very large