Aggregation in Economic Research: From Individual to Macro Relations

Our interest in problems of aggregation originates from about seven years ago when we became involved in research in the field of applied microeconomics. To our astonishment a vast majority of researchers in this area took it for granted that their, mostly thoroughly derived, micro models could meaningfully be confronted with per capita data. Nany of them did not even realize - at least they gave no utterance to it - that applying macro data in micro models raises considerable problems. Those who did mention the difficulty, almost always belittled its importance. Fortunately, there are noteworthy exceptions. Thinking about aggregation raises at least two questions: Why or why not aggregate? and How to aggregate and, in particular, to what degree? General answers to these questions can only be given in uninformative wording (as many assertions in economics): one aggregates for the sake of tractability, because of the lack of (individual) data, to avoid or to reduce multicollineartiy, to save degrees of freedom; one abstains from aggregation to avoid loss of information, to avoid aggregation biases and one aggregates such and to such degree as to bypass or reduce the drawbacks mentioned above.