This paper uses the inverse input demand function framework of Kim [Review of Economics and Statistics, 74 (1992) 546-551] to test for economies of industry and urban size in two US manufacturing sectors of differing technology intensity: farm and garden machinery (SIC 352) and measuring and controlling devices (SIC 382). The inverse input demand framework permits the estimation of the production function jointly with a set of cost shares without the imposition of prior economic restrictions. Tests using plant-level data suggest the presence of population scale (urbanization) economies in the moderate- to low-technology farm and garden machinery sector and industry scale (localization) economies in the higher technology measuring and controlling devices sector. The efficiency and generality of the inverse input demand approach are particularly appropriate for micro-level studies of agglomeration economies where prior assumptions regarding homogeneity and homotheticity are less appropriate. (C) 2001 Elsevier Science B.V, All rights reserved.