Priority setting thus involves balancing the competing pulls of "problem-driven" research (whose value is apparent in the short run) and "frontier" research that is driven by longer term ambition to shift the scientific frontier, regardless of immediate applications. In effect, this means integrating "demand" factors and "supply" factors in considering alternative investments in research. A set of difficult questions arises: Which research areas have the biggest potential payoffs? Which are most likely to generate successful outcomes? Which are most likely to benefit the poor or to alleviate stresses on marginal environments? Which are most likely to advance the state of knowledge in ways that will subsequently generate useful applications?
A number of approaches have attempted to integrate the supply and demand factors for research.5 Typically, these approaches involve the following ingredients:
1. An assessment of the benefits associated with achieving particular research goals;
2. A weighting of benefits in accordance with social objectives or other desired outcomes;
3. An assessment of the likelihood of success;
4. An estimate of the time at which benefits are likely to be realized;
In some cases, past experience may offer insights into the likely payoffs from different types of research. In such instances, ex post evaluations of research may serve as priority setting tools. Thus, Evenson (p. 99)6 notes that upland rice research has demonstrated few past payoffs. If nothing has changed that would alter the potential for upland rice research, a planner might want to consider the ex post evaluation data before investing funds in further upland rice research.
In most cases, however, priority setting will depend on data-intensive analysis of benefits, costs, and probabilities of success. Among the methods that have been used are scoring methods, expected economic surplus models (including benefit/cost analyses), and programming models. Scoring models simply assign different weights to different criteria, allowing the planner to rank different RPAs accordingly. Expected economic surplus models attempt to quantify the gains to consumers and producers (or subcategories of consumers and producers) from research that will alter the supply and/or demand for commodities. Finally, programming models solve an optimization problem involving the allocation of fixed quantities of scientific manpower and other resources, based on specified assumptions about the relationship between research inputs and expected outputs.
One type of information critical for priority-setting exercises is data on the benefits of research. Benefits may be estimated from crop loss data; from estimates of "yield gaps" between farmers' fields and experimental fields; or from subjective assessments of scientists working in a particular field. The benefit estimates depend on the production gains from "solving" a particular problem, the rate of diffusion of improved technologies, the duration of the gains, and the time lags until the gains are realized. The gains from a new crop variety or a new source of disease resistance will not be permanent. Typically, disease resistance depreciates over time. The duration of effective resistance is not surprsingly important for the calculation of benefits. Benefits also depend on the ways in which markets will respond to the new technology: They are sensitive to the slope of supply and demand curves for the final product. This is particularly critical for assessing the effects of new technologies on different categories or classes of consumers and producers.
A second category of information critical to the priority-setting process relates to the supply of research and, in particular, the probabilities of success and the likely time to success. In many cases, such information can only be obtained from surveys of knowledgeable scientists. These scientists may have overly optimistic assessments of new technologies, but they may also fail to anticipate successes that are near at hand. In general, subjective probability estimates from scientists seem to be relatively reliable, and they are better than any alternative estimates of research time lags and success probabilities.
Methodologies for eliciting scientists' input in priority setting are now well established. Some recent studies that use scientists' estimates to assign research priorities are Mills and Karanja for the Kenya Agricultural Research Institute's wheat program; Mills for sorghum in Kenya; Mutangadura and Norton for the Zimbabwean agricultural sector; and Evenson, Dey, and Hossain for rice in Asia.7-10 This literature is now well established. Although researchers have encountered some difficulties in utilizing data from scientists, the methods employed have become increasingly sophisticated. Moreover, with more experience in priority-setting studies, economists are beginning to have some opportunities to check the validity of scientists' responses. For example, Evenson examines changes over time in the subjective probability estimates of scientists participating in the Rockefeller program on rice biotechnology.11 A group of 15 scientists was surveyed in 1993 and again twelve months later. Although the sample was small, it allowed Evenson to test the hypothesis that scientists set "moving targets" for research completion dates. This hypothesis suggests that scientists will predict today that a project will come to fruition in, say, ten years; but, when they are asked about the same project two years later, they may still say that it will take ten years to achieve success. Evenson found little evidence of moving target problems, however.11 This reinforces the idea that subjective probability estimates may be of adequate quality for priority-setting studies.
From Theory to Practice: A Case Study of the Rockefeller Foundation s Decision to Prioritize Rice Biotechnology
Perhaps the most noteworthy example of research priority setting in recent years was undertaken by the Rockefeller Foundation as part of its decision to concentrate its agricultural investments in the relatively narrow field of biotechnology for rice. Since 1984, the Rockefeller Foundation has spent about $70 million to support a program for rice biotechnology research in the developing world. As Herdt (p. 19)12 notes, "rice was chosen because 90 percent or more of the world's rice is produced and consumed in the developing world, and as a result, gains from technical change in rice will largely accrue there." An added reason for the focus on rice was the sense that public and private research agencies in industrial countries would be unlikely to invest much in rice technologies that would be useful for the developing world. Although there is abundant rice research in Japan, the United States, and a number of European countries, this research may not generate very many direct benefits for developing countries because of differences in climate, photoperiod, indica vs. japonica differences, etc.
Consequently, the Rockefeller program aimed at achieving two objectives: generating technology useful for developing countries, and strengthening the capacity of laboratories and scientists in the developing world to perform rice biotechnology research. To date, a network of 200 senior scientists has been developed, with 300 additional scientific trainees. Evenson (p. 328)11 notes that as of early 1994, the program had supported some 130 projects in 26 countries, including 69 projects in developing countries. More than half of these projects included "biotechnology tool development" as a goal, and more than half specified "yield-enhancement technologies" among their objectives. Disease and insect resistance also accounted for a number of projects, with grain quality technologies and stress resistance technologies accounting for most of the remainder.
It is arguably too soon to see results from the Rockefeller Foundation's investments, but some preliminary results have already been achieved. Participating scientists succeeded in transforming rice in 1988, making it the first of the cereal crops to be transformed. There are now transformed lines containing economically useful traits.
Herdt reports that, in China, a rice variety produced with anther culture at the Shanghai Academy of Agricultural Sciences has incorporated genes for resistance to pathogens and to cold.13 This variety has been field tested on over 3000 hectares (ha) in Anhui and Hubei provinces, with yield improvements of 6-24 percent over the most popular current varieties. Herdt notes a number of additional attainments in rice biotechnology in Asia and predicts (p. 6)13 that "the contributions to rice yield increases from biotechnology in Asia will be on the order of 10 to 25 percent over the next ten years." These are striking benefits, but the yield increases do not by themselves convey the full impact of biotechnology. The incorporation of disease resistance through genetic manipulation can reduce the quantities of pesticides used on rice. The ability to incorporate useful traits from wild relatives raises the potential for a vast broadening of the genetic base for rice and other crops. These gains are not without dangers; many critics of biotechnology worry that hidden flaws will emerge as genetically modified plants become increasingly common. But, for now, the potential for improving human welfare through biotechnology seems to create a strong imperative.
At a less controversial level, biotechnology investments have paid off by improving the speed and efficiency of conventional plant breeding. For example, biotechnology has facilitated the development of molecular markers that allow breeders to easily discern whether a plant possesses traits of interest. Such approaches can improve the productivity of conventional breeding.
Was this article helpful?