Measuring the cost-efficiency of basic research investment: Input-output approaches

Restricted research budgets, expanding claims for resources, and steeply escalating costs have increased the demand for methods that establish or predict the efficiency or productivity of expenditures on basic research. Irvine and Martin (I-M) have recently advocated head-to-head efficiency comparisons of research performers using publications and citations. This paper (1) criticizes the I-M approach from the perspective of economics and policy analysis; (2) illustrates the natural production function extension of the I-M approach; (3) demonstrates, by worked out examples, the feasibility of such an extension along with its pitfalls; and (4) argues that combined econometric-bibliometric modeling can be used to improve research allocations with only modest increases in agency “transaction” costs. The paper presents a prototype efficiency analysis using recent NSF data for the behavioral and neural sciences.