The Identification of Structure at the Sectoral Level: a Reformulation of the Hirschman-Rasmussen Key Sector Indices

The Hirschman-Rasmussen key sector identification indices are reformulated to correct for their sensitivity to large multiplier values in the Leontief inverse matrix. This leads to four results: (1) the generalized key sector indices; (2) the generalized coefficients of variation; (3) a likelihood ratio test, obtained from computing the likelihood function of each sample of Cauchy-distributed indices; and (4) the elasticity of intermediate sectoral supply response. Empirical analysis reveals that the methods proposed identify more accurately the structure of Washington State, while not capturing the spread of the linkages associated with that structure.