Birgit Grodal, Comment on L. H. Summers, The Scientific Illusion in Empirical Macroeconomics. Scandinavian Journal of Economics 93 (2), 155-159, 1991.
Summary
According to Grodal, the reason why sophisticated macro-econometrics has little impact is that the underlying macroeconomic theory is insufficiently developed.
Macroeconomic models are based on unwarranted simplifications. They use one or a few commodities, one or a few representative consumers and producers, perhaps a public sector, and simple institutional arrangements. Yet the conclusions drawn from these models are treated as if they hold in economies with many interacting agents.
There is no basis in economic theory for believing that these models should give a good description of the way an economy with many agents operates.
Only in economies where all consumers have identically homothetic preferences can the demand side can be represented by a representative consumer.
That is true if all income distributions are allowed. If only a fixed income distribution is considered, then it may be possible to represent the demand side by a representative consumer, but only under severe restrictions on the preferences of the individual agents. In this case, the welfare implications of economic policy for the representative consumer can be the opposite of the welfare implications for all the original consumers.
Thus it is not surprising that macroeconomic relations derived from theory are usually rejected empirically. The only way to obtain better empirical macroeconomic models is to develop better macroeconomic theory.
Grodal recommends studying the distribution of agents’ characteristics and using it to derive conclusions about aggregate behavior.