ABSTRACT
The conditions for the existence of aggregate production functions are extremely restrictive (see, for example, Fisher, 1969 for a survey and references). Even for linear aggregation of linear models we find that consistency is an exceptional result and, in general, macro parameters depend not only on the corresponding micro parameters but on other parameters at the micro level as weH (see, for instance, Theil1954 and Lütjohann, 1974). For nonlinear models the odds against perfect aggregation are generally worse (see Barker, 1970, 1974). However, even when paying lip service to 'the problem of aggregation' empirical economists often
ll il ii
feel that aggregation will presumably give reasonable approximations so long as elasticities are fairly similar and structural change in output not too large (on the consumption side similar assumptions are often made that income distribution does not change 'too much').
