We generalize the normalized constant elasticity of substitution (CES) production function by allowing the elasticity of substitution to vary isoelastically with (i) the relative factor share, (ii) the marginal rate of substitution, (iii) the capital–labor ratio, (iv) the capital share, (v) the capital’s rate of return, or (vi) the capital–output ratio. Ensuing isoelastic elasticity of substitution (IEES) functions have intuitively and analytically desirable properties, for example, self-duality. Empirically, for the post-war US economy we robustly reject the CES specification in favor of the IEES alternative. Assuming the IEES production structure we find that the capital–labor elasticity of substitution has remained around 0.8–0.9 from 1948 to the 1980s, followed by a period of secular decline.

A Diewert-flexible (dual) cost function was used to derive a system of conditional factor demand equations for Louisiana rice producers. Generalized Leontief cost and factor share equations were fitted for the 1955-87 period using Zellner's SURE system estimation procedure. The Aitken parameter estimates reveal that: (1) the optimal input mix of rice farmers varies with production scale, (2) the factor-augmenting technical change is labor and chemical saving but seed using, (3) pairwise input substitutions are limited, and (4) factor demands are own-price inelastic. An implication is that Louisiana rice farmers will not appreciably alter their factor utilizations when relative input prices change.

We provide industry-level estimates of the elasticity of substitution (σ) between capital and labor in the United States. We also estimate rates of factor augmentation. Aggregate estimates are produced. Our empirical model comes from the first-order conditions associated with a constant–elasticity of substitution production function. Our data represent 35 industries at roughly the 2-digit SIC level, 1960–2005. We find that aggregate U.S. σ is likely less than 0.620. σ is likely less than unity for a large majority of individual industries. Evidence also suggests that aggregate σ is less than the value-added share-weighted average of industry σ's. Aggregate technical change appears to be net labor–augmenting. This also appears to be true for the large majority of individual industries, but several industries may be characterized by net capital augmentation. When industry-level elasticity estimates are mapped to model sectors, the manufacturing sector σ is lower than that of services; the investment sector σ is lower than that of consumption.