Optimum Resource Allocation in U.S. Agriculture: Comment

Abstract

Over ten years ago, Tyner and Tweeten presented two imaginative and pervasive articles in this Journal. The first notes past statistical difficulties in obtaining reliable parameter estimates for aggregate U.S. farm production functions with several input variables (Tyner and Tweeten 1965). In proposing an alternative approach, Tyner and Tweeten note that factor elasticities must equal their respective factor shares, given the assumptions of perfectly competitive product and factor markets and market equilibrium.' They also assume that entrepreneurs attempt to organize production such that actual factor shares adjust toward an equilibrium in a geometric distributed lag fashion. Since the parameters of a CobbDouglas production function are factor elasticities, they propose computing elasticities from historical data on factor shares, avoiding the problem of multicollinearity. Estimates of factor elasticities for nine input categories were developed by decade for 1912-61 using data from published and unpublished ERS sources. The second article (1966) built upon the first by incorporating these elasticity estimates into a Cobb-Douglas production function for each decade. This was accomplished by regressing farm output (in constant dollars) on a composite input variable (in constant dollars) defined as the product of each factor raised to its exogenously estimated factor elasticity. A linear model with a zero intercept was used to estimate the constant term for each decade. Having in this manner synthesized a production function, minimum-cost input levels, equilibrium output and input levels, factorsubstitution relationships, and factor-demand and product-supply elasticities were derived, and comparisons were made with the actual situation in each decade. Based on these comparisons, Tyner and Tweeten conclude (a) adjustment to a leastcost input combination to produce the actual average 1952-61 output would have reduced the input dollar volume by $1.9 billion, or 5.6%; (b) adjustment of farm resources to an equilibrium level with all variable resources earning an opportunity-cost return would have entailed a reduction of $4.2 billion ($1947-49), or 12.5% of the actual input volume in the 1952-61 period; (c) labor was in excess supply by two-fifths in the 1952-61 period;2 and (d) capital inputs supplied by the nonfarm sector were not generally used in excess by farmers (1966, pp. 629-30). In summary, their results serve as a guideline for public policy by indicating the magnitude of resource adjustment needed to achieve economic equilibrium, and the economic cost of maintaining nonoptimal resource levels and combinations in agriculture (1966, p. 630). These were, and remain, important conclusions resulting from an ambitious and innovative study-conclusions that added empirical credence to the popular notion of substantially excess labor resources in agriculture. The conceptual, methodological, and policy significance of this work was emphasized by the inclusion of the second article in the A.E.A. Readings in Agricultural Economics. Impressed with the ingenuity of the Tyner and Tweeten approach and its implications for agricultural policy analysis, we, like others, have used their articles in graduate teaching. In so doing, we have discovered an error in the Tyner and Tweeten analysis that substantially affects the policy conclusions presented in the second article. It is the purpose of this comment to present revised results for an article that remains an important source of economic hypotheses and policy prescriptions even a decade later.