Identification by Disaggregation
Document Type Article
Published by American Economic Review 75:5 (December 1985), pp. 1165-1167. Copyright © 1985 American Economic Association. Used by permission.
Standard economic theory predicts that the actions of individual participants in competitive markets have negligible effects on market-determined aggregates. Applied researchers, and even some econometric textbooks, incorrectly infer from this that market prices can be modeled as econometrically exogenous with respect to the quantity demanded of an individual consumer. This faulty inference has even led some researchers (for example, Robert Engle, 1978; Nicholas Kiefer, 1984; Roger Waud, 1974) to employ an estimation strategy we call identification by disaggregation (IBD). This procedure attempts to circumvent the simultaneity problem in a macro regression by disaggregating the dependent variable and estimating the relationship for individual agents or sectors. This note provides a simple proof that estimates using disaggregated dependent variables suffer, on average, from the same degree of simultaneity bias as the estimates using aggregate data.