ABSTRACT

As noted by Capalbo and Vo in chapter 3, econometric productivity (or production function) studies have historically been of a singleoutput, several-input type. The early studies employed the primal approach, fitting (aggregate) output to several variable inputs. Later studies employed the dual approach by fitting the profit function or input demand functions derived from underlying cost or profit functions. 1

Many of these earlier studies have serious deficiencies as frameworks within which to learn about agricultural productivity. First, the singleoutput methods assume that technology is separable between inputs and outputs. This implies that decisions on optimal relative output (input) quantities are unaffected by input (output) prices (Lau, 1972; Christensen, 1975). For example, the relative quantity of corn to soybeans supplied is unaffected by the price of fertilizer. More importantly, production is assumed to be either nonjoint in inputs or joint in a special way. When applied to agriculture, nonjointness in inputs ignores technical interdependencies among outputs and interdependencies induced by allocable fixed factors (Christensen, 1975; Lau, 1972; Shum-

way, Pope, and Nash, 1984). If both crop and livestock outputs are supplied, production is almost certainly joint in inputs. A potentially fertile area of agricultural productivity research seems to be the disaggregation of outputs so that the relative and absolute supply response to exogenous variables-for example, prices, allocable fixed factors, and public sector agricultural research-can be quantified.