Estimating the Herfindahl Index from Concentration Ratio Data

Abstract High concentration ratio (C)—Herfindahl (H) index correlations found in United States studies may be biased upward by using estimated H. For New Zealand data, correlations between C and actual H are lower; the H—C relationship appears to be nonlinear. Guided by income distribution studies, a new family of H estimators—derived from H's upper and lower bounds rather than generalized assumptions about firm size distributions—is proposed. The methodology for calculating the bounds from concentration ratio data is described, and the efficacy of various estimators is tested.