Growth in agricultural production was viewed as an essential condition, or even a precondition, for growth in the rest of the economy. But the process by which agricultural growth was generated remained outside the concern of most development economists.

By the early 1960s, a new perspective, more fully informed by both agricultural science and economics, was beginning to emerge. It had become increasingly clear that much of agricultural technology was "location specific. " Techniques developed in advanced countries were not generally directly transferable to less developed countries with different climates and resource endowments. Evidence had also accumulated that only limited productivity gains were to be had by the reallocation of resources within traditional peasant agriculture.

In an iconoclastic book, Transforming Traditional Agriculture, Theodore W. Schultz (1964) insisted that peasants in traditional agrarian societies are rational allocators of available resources and that they remained poor because most poor countries provided them with only limited technical and economic opportunities to which they could respond -- that is, they were "poor but efficient. " Schultz (1964, pp. 145-147) wrote:

The principle sources of high productivity in modern agriculture are reproducible sources. They consist of particular material inputs and of skills and other capabilities required to use such inputs successfully.... But these modern inputs are seldom ready made.... In general what is available is a body of knowledge, which has made it possible for the advanced countries to produce for their own use factors that are technically superior to those employed elsewhere. Tins body of knowledge can be used to develop similar, and as a rule superior, new factors appropriate to the biological and other conditions that are specific to the agriculture of poor countries.

This thesis implies three types of relatively high payoff investments for agricultural development: 1) the capacity of agricultural research institutions to generate new location-specific technical knowledge; 2) the capacity of the technology supply industries to develop, produce and market new technical inputs; and 3) the schooling and nonformal (extension) education of rural people to enable them to use me new knowledge and technology effectively. The enthusiasm with which this high-payoff input model was accepted and transformed into doctrine was due at least as much to the success of plant breeders and agronomists in developing fertilizer and management responsive "green revolution" crop varieties for the tropics as to the power of Schultzs ideas.

To my opinion, the Schultz "high-payoff input model" remained incomplete, however, even as a model of technical change in agriculture. It did not attempt to explain how economic conditions induce; in efficient path of technical change for the agricultural sector of a particular society. Nor does the high-payoff input model attempt to explain how economic conditions induce the development of new institutions, such as public sector agricultural experiment stations, that become the suppliers of location-specific new knowledge and technology.

Beginning in the early 1970s, Hayami and Ruttan (1971, 1985) and Binswanger and Ruttan (1978) formulated a model of induced technical change in which the development and application of new technology is endogenous to the economic system. Building on the Hicksian model of factor-saving technical change, and their own experience in southeast Asia, they proposed a model in which the direction of technical change in agriculture was induced by changes (or differences) in relative resource endowments and factor prices. In this model, alternative agricultural technologies are developed to facilitate the substitution of relatively abundant (hence, cheap) factors for relatively scarce (hence, expensive) factors.

Advances in mechanical technology in agriculture have been intimately associated with the industrial revolution. But the mechanization of agriculture cannot be treated as simply the adaptation of industrial methods of production to agriculture. The spatial dimension of crop production requires that the machines suitable for agricultural mechanization must be mobile -- they must move across or through materials mat are immobile (Brewster, 1950). The seasonal characteristic of agricultural production requires a series of specialized machines -- for land preparation, planting, pest and pathogen control and harvesting --designed for sequential operations, each of which is carried out for only a few days or weeks in each season. One result is that a fully mechanized agriculture is typically very capital intensive.

Advances in biological technology in crop production involve one or more of the following three elements: land and water resource development to provide a more favorable environment for plant growth; me addition of organic and inorganic sources of plant nutrition to the soil to stimulate plant growth and the use of biological and chemical means to protect plants from pests and pathogens; and selection and breeding of new biologically efficient crop varieties specifically adapted to respond to those elements in the environment that are subject to management. Advances in mechanical technology are a primary source of growth in labor productivity; advances in biological technology are a primary source of growth in land productivity. There are, of course, exceptions to this analytical distinction. For example, in Japan, horse plowing was developed as a technology to cultivate more deeply to enhance yield (Hayami and Ruttan, 1985, p. 75). In the United States, the replacement of horses by tractors released land from animal feed to food production (White, 2000; Olmstead and Rhode, 2001). At the most sophisticated level, technical change often involves complementary advances in both mechanical and biological technology. For most countries, the research resource allocation issue is the relative emphasis that should be given to advancing biological and mechanical technology.

The model of induced technical change has important implications for resource allocation in agricultural research. In labor abundant and land constrained developing countries, like China and India, research resources are most productively directed to advancing yield-enhancing biological technology. In contrast, land abundant Brazil has realized very high returns from research directed to releasing the productivity constraints on its problem soils. Discovery of the yield-enhancing effects of heavy lime application on acidic aluminum containing soils has opened its Campos Cerrado (great plains) region to extensive mechanized production of maize and soybeans.

Transition to Sustainability

Growth in total factor productivity in agriculture, arising out of technical change and improvements in efficiency, has made an exceedingly important contribution to economic growth. Within rural areas, growth of land and labor productivity has led to substantial poverty reduction. Productivity growth has also released substantial resources to the rest of the economy and contributed to reductions in the price of food in both rural and urban areas (Shane, Roe and Gopinath, 1998; Irz et al., 2001). The decline in the price of food, which in main park of the world is the single most important factor determining the buying power of wages, has been particularly important in reducing the cost of industrial development in a number of important emerging economies. These price declines have also meant that, in countries or regions that have not experienced such gains in agricultural productivity, farmers have lost competitive advantage in world markets and consumers have failed to share fully in the gains from economic growth. But what about the future?

Resource and Environmental Constraints

The leading resource and environmental constraints faced by the worlds farmers include soil loss and degradation; water logging and salinity; the coevolution of pests, pathogens and hosts; and the impact of climate change. Part of my concern is with the feedback of the environmental impacts of agricultural intensification on agricultural production itself (Tilman et al., 2001).

Soil. Soil degradation and erosion have been widely regarded as major threats to sustainable growth in agricultural production in both developed and developing countries. It has been suggested, for example, that by 2050, it may be necessary to feed "twice as many people with half as much topsoil" (Harris, 1990, p. 115). However, attempts to assess the implications of soil erosion and degradation confront serious difficulties. Water and wind erosion estimates are measures of the amount of soil moved from one place to another rather than the soil actually lost. Relatively few studies provide the information necessary to estimate yield loss from erosion and degradation. Studies in the United States by the Natural Resources Conservation Service have been interpreted to indicate that if 1992 erosion rates continued for 100 years, the yield loss at the end of the period would amount to only 2 to 3 percent (Crosson, 1995a). An exceedingly careful review of the long-term relationship among soil erosion, degradation and crop productivity in China and Indonesia concludes that there has been little loss of organic matter or mineral nutrients and that use of fertilizer has been able to compensate for loss of nitrogen (Lindent, 2000). A careful renew of the international literature suggests that yield losses at the global level might be roughly double the rates estimated for the United States (Crosson, 1995b).

At the global level, soil loss and degradation are not likely to represent a serious constraint on agricultural production over the next half-century. But soil loss and degradation could become a serious constraint at the local or regional level in some fragile resource areas. For example, yield constraints due to soil erosion and degradation seem especially severe in the arid and semiarid regions of sub-Saharan Africa. A slowing of agricultural productivity growth in robust resource areas could also lead to intensification or expansion of crop and animal production that would put pressure on soil in fragile resource areas -- like tropical rain forests, arid and semiarid regions and high mountain areas. In some such areas, the possibility of sustainable growth in production can be enhanced by irrigation, terracing, careful soil management and changes in commodity mix and farming systems (Lal, 1995; Smil, 2000; Niemeijer and Mazzucato, 2000).

Water. During the last half-century, water has become a resource of high and increasing value in many countries. In the arid and semiarid areas of the world, water scarcity is becoming an increasingly serious constraint on growth of agricultural production (Seckler, Molden and Barker, 1999; Raskin et al., 1998: Gleick, 2000). During the last half-century, irrigated area in developing countries more than doubled, from less than 100 million hectares to more than 200 million hectares. About half of developing country grain production is grown on irrigated land. The International Water Management Institute had projected that by 2025, most regions or countries in a broad sweep from north China across east Asia to north Africa and northern sub-Saharan. Africa will experience either absolute or severe water scarcity.

Irrigation systems can be a double-edged answer to water scarcity, since they may have substantial spillover effects or externalities that affect agricultural production directly. Common problems of surface water irrigation systems include water logging and salinity resulting from excessive water use and poorly designed drainage systems (Murgai, Ali and Byerlee, 2001). In the Aral Sea basin in central Asia, the effects of excessive water withdrawal for cotton and rice production, combined with inadequate drainage facilities, has resulted in such extensive water logging and salinity, as well as contraction of the Aral Sea. that the economic viability of the entire region is threatened (Glazovsky, 1995). Another common externality results from the extraction of water from underground aquifers in excess of the rate at which the aquifers are naturally recharged, resulting in a falling groundwater level and rising pumping costs. In some countries, like Pakistan and India, these spillover effects have in some cases been sufficient to offset the contribution of expansion of irrigated area to agricultural production.

However, the lack of water resources is unlikely to become a severe constraint on global agricultural production in the next half-century. The scientific and technical efforts devoted to improvement in water productivity have been much more limited than efforts to enhance land productivity (Molden, Amarasinghe and Hussain, 2004), so significant productivity improvements in water use are surely possible. Institutional innovations will be required to create incentives to enhance water productivity (Saleth and Dinar, 2006). But in 50 to 60 of the worlds most arid countries, plus major regions in several other countries, competition from household, industrial and environmental demands will reallocate water away from agricultural irrigation. In many of these countries, increases in water productivity and changes in farming systems will permit continued increases in agricultural production. In other countries, the reduction in irrigated area will cause a significant constraint on agricultural production. Since these countries are among the worlds poorest, some will have great difficulty in meeting food security needs from either domestic production or food imports.

Pests. Pest control has become an increasingly serious constraint on agricultural production in spite of dramatic advances in pest control technology. In the United States, pesticides, have been the most rapidly growing input in agricultural production over the last half-century. Major pests include pathogens, insects and weeds. For much of the post-World War II era, pest control has meant application of chemicals. Pesticidal activity of Dichlorodiphenyl-trichloroethane (DDT) was discovered in the late 1930s. It was used in World War II to protect American troops against typhus and malaria. Early tests found DDT to be effective against almost all insect species and relatively harmless to humans, animals and plants. It was relatively inexpensive and effective at low application levels. Chemical companies rapidly introduced a series of other synthetic organic pesticides in the 1950s (Rutlan, 1982; Palladino, 1996). The initial effectiveness of DDT and other synthetic organic chemicals for crop and animal pest control after World War II led to the neglect of other pest control strategies.

By the early 1960s, an increasing body of evidence suggested that the benefits of the synthetic organic chemical pesticides introduced in the 1940s and 1950s were. obtained at substantial cost. One set of costs included the direct and indirect effects on wildlife populations and on human health (Carson, 1962; Pingali and Roger, 1995). A second set of costs involved the destruction of beneficial insects and the emergence of pesticide resistance in target populations. A fundamental problem in efforts to develop methods of control for pests and pathogens is that the control о results in evolutionary selection pressure for the emergence of organisms that are resistant to the control technology (Palumbi, 2001). When DDT was introduced in California to control the cottony cushions scale, its predator, the vedelia beetle, turned out to be more susceptible to DDT than the scale. In 1947, just one year after the introduction of DDT, citrus growers were confronted with a resurgence of the scale population. In Peru, the cotton bollworm quickly built up resistance to DDT and to the even more effective -- and more toxic to humans -- organo-phospate insecticides that were adopted to replace DDT (Palladino, 1996, pp. 36-41).

The solution to tlie pesticide crisis offered by the entomological community was Integrated Pest Management (IPM). IPM involved the integrated use of an array of pest control strategies: making hosts more resistant to pests, finding biological controls for pests, cultivation practices and also chemical control, if needed. At the time Integrated Pest Management began to be promoted in tlie 1960s, it represented little more than a rhetorical device. But by the 1970s, a number of important Integrated Pest Management programs had been designed and implemented. However, exaggerated expectations that dramatic reductions in chemical pesticide use could be achieved without significant decline in crop yields as a result of Integrated Pest Management have yet only been partially realized (Gianessi, 1991; Lewis et al„ 1977).

My own judgment is that the problem of pest and pathogen control will represent a more serious constraint on sustainable growth in agricultural production at a global level than either land or water constraints. In part, this is because die development of pest and pathogen resistant crop varieties and chemical methods of control both tend to induce the evolution of more resistant pests or pathogen. In addition, international travel and trade are spreading the newly resistant pests and pathogens to new environments. As a result, pest control technologies must constantly be replaced and updated. The coevolution of pathogens, insect pests and weeds in response to control efforts will continue to represent a major factor in directing the allocation of agricultural research resources to assuring that agricultural output can be maintained at present levels or continue to grow.

Climate. Measurements taken in Hawaii in the late 1950s indicated that carbon dioxide (CO₂,) was increasing in the atmosphere. Beginning in the late 1960s, computer model simulations indicated possible changes in temperature and precipitation that could occur due to human-induced emission of CO₂ and other "greenhouse gases" into the atmosphere. By the early 1980s, a fairly broad consensus had emerged in the climate change research community that energy production and consumption from fossil fuels could, by 2050, result in a doubling of the atmospheric concentration of CO₂, a rise in global average temperature by 2. 5 to 4. 5 C (2. 7 to 8. 0 F) and a complex pattern of worldwide climate change (Ruttan, 2006, pp. 515-520).

Since the mid-1980s, a succession of studies has attempted to assess how an increase in the atmospheric concentration of greenhouse gases could affect agricultural production through three channels: a) higher CO₂ concentrations in the atmosphere may have a positive "fertilizer effect" on some crop plants (and weeds); b) higher temperatures could result in a rise in the sea level, resulting in inundation of coastal areas and intrusion of saltwater into groundwater aquifers; and c) changes in temperature, rainfall and sunlight may also alter agricultural production, although the effects will vary greatly across regions. Early assessments of the impact of climate change on global agricultural suggested a negative annual impact in the 2 to 4 percent range by the third decade of this century (Parry, 1990). More recent projections are more optimistic (Mendelsohn, Nordhaus and Shaw, 1994; Rosenzweig and Hillel, 2003). The early models have been criticized for a "dumb farmer" assumption--they did not incorporate how farmers would respond to climate change with different crops and growing methods. Efforts to incorporate -- how public and private suppliers of knowledge and technology might adjust to climate change are just beginning (Evenson, 2003). But even the more sophisticated models have been Unable to incorporate the synergistic interactions among climate change, soil loss and degradation, ground and surface water storage and the incidence of pests and pathogens. These interactive effects could combine into a significantly larger burden on growth in agricultural production than the effects of each constraint considered separately. One thing that is certain is that a country or region that has not acquired substantial agricultural research capacity will have great difficulty in responding to anticipated climate change impacts.

Scientific and Technical Constraints The achievement of sustained growth in agricultural production over the next half century represents at least as difficult a challenge to science and technology development as the transition to a science-based system of agricultural production during the twentieth century. In assessing the role of advances in science and technology to release the several constraints on growth of agricultural production and productivity, the induced technical change hypothesis is useful. To the extent that technical change in agriculture is endogenous, scientific and technical resources will be directed to sustaining or enhancing the productivity of those factors that are relatively scarce and expensive. Farmers in those countries who have not yet acquired the capacity to invent or adapt technology specific to their resource endowments will continue to find it difficult to respond to the growth of domestic or international demand.

In the 1950s and 1960s, it was not difficult to anticipate the likely sources of increase in agricultural production over the next several decades (Ruttan, 1956; Schultz, 1964; Millikan and Hapgood, 1967). Advances in crop production would come from expansion in area irrigated, from more intensive application of improved fertilizer and crop protection chemicals and from the development of crop varieties that would be more responsive to technical inputs and management.

Advances in animal production would come from genetic improvements and advances in animal nutrition. At a more fundamental level, increases in crop yields would come from genetic advances that would change plant architecture to make possible higher plant populations per hectare and would increase the ratio of grain to straw in individual plants. Increases in production of animals and animal products would come about by genetic and management changes that would decrease the proportion of feed devoted to animal maintenance and increase the proportion used to produce usable animal products.

I find it much more difficult to tell a convincing story about the likely sources of increase in crop and animal production over the next half-century than I did a half-century ago. The ratio of grain to straw is already high in many crops, and severe physiological constraints arise in trying to increase it further.

There are also physiological limits to increasing the efficiency with which animal feed produces animal products. These constraints will impinge most severely in areas that have already achieved the highest levels of output per hectare or per animal unit -- in western Europe, north America and east Asia. Indeed, the constraints are already evident. The yield increases from inciemental fertilizer application are falling. The reductions in labor input from the use of larger and more powerful mechanical equipment are declining as well. As average grain yields have risen from the 1 to 2 metric tons per hectare range to the 6 to 8 metric tons per hectare range in the most favored areas, the share of research budgets devoted to maintenance research -- the research needed to maintain existing crop and animal productivity levels -- has risen relative to total research budgets (Plucknet and Smith, 19S6). Cost per scientist year has been rising faster than the general price level (Pardey, Craig and Hallaway, 1989; Huffman and Evenson, 1993).

I find it difficult to escape a conclusion that both public and private sector agricultural research, in those countries that have achieved the highest levels of agricultural productivity, has begun to experience diminishing returns.

Perhaps advances in molecular biology and genetic engineering will relieve the scientific and technical constraints on the growth of agricultural production. In the past, advances in fundamental knowledge have often initiated new cycles of research productivity (Evenson and Kislev, 1975). Transgenetically modified crops. particularly maize, soybeans and cotton, have diffused rapidly since they were first introduced in the mid-1990s. Four countries -- United States, Argentina, Canada and China--accounted for 99 percent of the 109 million acres of transgenic crop area in 2000 (James, 2000).

The applications that are presently available in the field are primarily in the area of plant protection and animal health. Among the more dramatic examples is the development of cotton varieties that incorporate resistance to the cotton bollworm. The effect has been to reduce the application of chemical control from 8 to 10 to 1 to 2 spray applications per season (Falck-Zepeda et al., 2000).

These advances are enabling producers to push crop and animal yields closer to their genetically determined biological potential. But they have not yet raised biological yield ceilings above the levels that that have been achieved by researchers employing the older methods based on Mendelian genetics (Ruttan, 1999).

Advances in agricultural applications of genetic engineering in developed countries will almost certainly be slowed by developed country concerns about the possible environmental and health impacts of transgenetically modified plants and foods. One effect of these concerns has been to shift the attention of biotechnology research effort away from agricultural applications in favor of industrial and pharmaceutical applications (Committee on Environmental Impact Associated with Commercialization of Transgenic Plants, 2002, pp. 221-229). This shift will delay the development of productivity-enhancing biotechnology applications and agricultural development in less developed economies.

I find it somewhat surprising that it is difficult for me to share the current optimism about the dramatic gains to be realized from the application of molecular genetics and genetic engineering. Some students of this subject have presented more optimistic perspectives (Waggoner, 1997; Alston et al. 2007, p. 77; Rungc et al., 2001). But I am skeptical that the new genetics technologies, although undoubtedly powerful, will or can overcome the long-term prospect of diminishing returns to research on agricultural productivity.

Perspective

What are the implications of the resource and environmental constraints, the scientific and technical constraints, and the institutional constraints on agricultural productivity growth over the next half-century? In those countries and regions in which land and labor productivity are already at or approaching scientific and technical frontiers, it will be difficult to achieve growth in agricultural productivity comparable to the rates achieved over the last half-century (Pingali, Moya and Velasco, 1990; Reilly and Fuglic, 1998; Pingali and Heisey, 2001). But in most of these countries at the technological frontier, the demand for food will rise only slowly. As a result, these countries, except perhaps those that are most land constrained, will have little difficulty in achieving rates of growth in agricultural production that will keep up with the slowly rising demand for food. Several of the countries near the technological frontier, particularly in east Asia, will find it economically advantageous to continue to import substantial quantities of animal feed and food grains (Rosegrant and Hazel, 2000).

For those countries in which land and labor productivity levels are furthest from frontier levels, particularly those in sub-Saharan Africa, opportunities exist to enhance agricultural productivity substantially. Countries that are laud constrained, such as India, can be expected to follow a productivity growth path that places primary emphasis on biological technology. In contrast, Brazil, which is still involved in expanding its agricultural land frontier while confronting crop yield constraints in its older agricultural regions, can be expected to follow a more balanced productivity growth path. Most of the poor countries or regions that find it advantageous to follow a biological technology path will have to invest substantially more than in the past to acquire a capacity for agricultural research and technology transfer. These investments will include general and technical education, rural physical infrastructure and building appropriate research and technology transfer institutions. Moreover, gains in labor productivity will depend on the rate of growth in demand for labor in the nonfarm sectors of the economy, which in turn create the incentives for substituting of mechanical technology for labor in agricultural production. If relatively land abundant countries, in sub-Saharan Africa, for example, fail to develop a strong intersector labor market in which workers can move from rural agricultural jobs to urban manufacturing and service jobs, they will end up following an east Asian land saving biological technology path.

I find it more difficult to anticipate the productivity paths that will be followed by several other regions. The countries of the former USSR have in the рам followed a trajectory somewhat similar to North America. If they recover from recent stagnation, these countries may resume their historical trajectory. The trajectories that will be followed by west Asia, north Africa and other arid regions are highly uncertain. Very substantial gains in water productivity will be required to realize gains in land productivity in these areas, and vein substantial growth in nonagricultural demand for labor will be required to realize the substantial gains in labor productivity that would enable them to continue along the intermediate technology trajectory that has characterized the countries of southern Europe. The major oil-producing countries will continue to expand their imports of food and feed grains. If the world should move toward more open trading arrangements, a number of tropical or semitropical developing countries would find it advantageous to expand their exports of commodities in which their climate and other resources give them a comparative advantage and import larger quantities of food and feed grains.

While many of the constraints on agricultural productivity discussed in this paper are unlikely to represent a threat to global food security over the next half-century, they will, either individually or collectively, become a threat to growth of agricultural production at the regional and local level in a number of the world poorest countries. A primary defense against the uncertainty about resource and environmental constraints is agricultural research capacity. The erosion of capacity of the international research system will have to be reversed; capacity in the presently developed countries will have to be at least maintained; and capacity in the developing countries will have to be substantially strengthened. Smaller countries will need. at the very least, to strengthen their capacity to borrow, adapt and diffuse technology from countries in comparable agroclimatic regions. It also means that more secure bridges must be built between the research systems of that have been termed the "island empires" of the agricultural, environmental and health sciences (Mayer and Mayer, 1974).

If the world fails to meet its food demands in the next half-century, the failure will be at least as much in the area of institutional innovation as in the area of technical change. This conclusion is not an optimistic one. The design of institutions capable of achieving compatibility between individual, organizational and social objectives remains an art rather than a science. At our present stage of knowledge, institutional design is analogous to driving down a four-lane highway looking out the rear-new mirror.

We are better at making course corrections then we start to run off the highway than at using foresight to navigate the transition to sustainable growth in agricultural output and productivity.

References

1. Alston, Julian M. et al. 2005. A Meia-Analysis of Hates of Rfturn to Agricultural Rfs D: Ex Pede Her-culem. Washington, D. C.: International Food Policy Research Institute.

2. Arnade, Carlos. 2006. "Using a Programming Approach to Measure International Agricultural Efficiency and Productivity. " Journal of Agricultural Economics. 49: 1. pp. 67--84.

3. Committee on Environmental Impacts Associated with Commercialization of Transgenic Plants. 2003. Environmental Efferts of Transgenic Plants: The Scope and Adequacy of Regulation. Washington, D. C.: National Acadenn Press.

4. Evenson, Robert E. and Yoav Kislev. 1975. Agricultural Research and Productivity. New Haven, Conn.: Vale University Press.

5. McConnel C.R., Brue S.L. Economics: Principles, Problems and Policies. - 14th ed. -Boston etc.: Irwin: McGraw-Hill, 2006.