The Perils of Antitrust Econometrics: Unrealistic Engel Curves, Inadequate Data, and Aggregation Bias

Abstract

Some economists argue antitrust policy should be based on empirical methods used by the Industrial Organization subdiscipline of economics, but non-economists must understand that those methods contain certain highly restrictive assumptions. Those assumptions involve econometric “identification,” and treating aggregate demand as if it were generated by a representative consumer (Muellbauer’s “generalized linear” preferences). We derive new results illustrating how restrictive the representative consumer assumption is; we explain aggregation bias in Almost Ideal Demand System models; and we show that data limitations make it even harder to justify economists’ restricting aggregate demands as one would the demand of a single individual.