CIESIN Reproduced, with permission, from: Crosson, P. 1989. Climate change and mid-latitudes agriculture: Perspectives on consequences and policy responses. Climatic Change 15: 51-73.


PIERRE CROSSON Senior Fellow, Resources for the Future, 1616 P Street, N.W., Washington, DC 20036, U.S.A.

Abstract. Because of population growth, economic development, and technological change, world and mid-latitudes agriculture will look very different than they do today by the time 2xCO2 climate change begins to have major impact. It does not appear that impact would seriously restrain the growth of world agricultural capacity. However, significant shifts in regional comparative advantage in agriculture would be likely. Because the consequences of 2xCO2 climate change for agriculture would vary among countries - some suffering losses, others seeing themselves as potential winners - these consequences could impede international agreements to control climate change. However, even countries gaining agricultural advantage from climate change will need changes in policy to capture the gains. And policies to lessen the costs to the losers will be essential. If global warming continues beyond that associated with 2xCO2, all countries in time would be losers.


I take as a point of departure the consensus among atmospheric scientists and modelers that continuation of present trends in emissions of CO2 and other radiatively active trace gases will gradually increase global average temperatures of the troposphere 1.5-5.5 deg.C by sometime in the second third of the 21st century, the increase being greater in the polar regions and less at the equator (Bolin et al., 1986). Given these temperature increases and related changes in global and regional climates, I ask two questions: (1) what policy issues will the consequences of these changes for mid-latitudes agriculture raise for mid-latitudes governments? (2) what strategies should these governments be considering for dealing with the consequences?

Identification of Policy Issues

Underlying Assumptions

The major climate consequences of the prospective 1.5-5.5 deg.C increase in global average temperature will occur beyond the lifetime of most people now in policy positions in mid-latitudes governments. To say that these consequences will pose, or should pose, policy issues for these people is to assume that they accept, or should accept, some responsibility for the welfare of not only the next generation but also of generations not yet born. It is not obvious that the assumption is generally valid. The tendency of elected officials to heavily discount events occurring beyond the next election is notorious. Even those who in principle accept a responsibility toward the future may believe that with prudent tending on their part, economic growth will provide generations to come with the knowledge and other wherewithal needed to deal with whatever problems our generation may impose on them.

Such attitudes undoubtedly exist among policy people in mid-latitudes (and all other) governments, but it is not obvious that they predominate. If any event, if one is to fruitfully discuss policy issues presented by climate change one must assume that policy people share, or can be persuaded to share, one's view that such issues are, or soon will be, upon us.

I make a second assumption: that the concepts of social benefits and costs provide useful guidelines in seeking to identify policy issues raised by climate change. Questions have been raised (Ausubel, 1983) about the usefulness of economic analysis for addressing these issues, and benefit-cost concepts are very much part of economics. Three principal objections have been raised. One is that many of the costs and benefits of climate change are unpriced, e.g. habitat values lost or gained, hence they cannot be adequately captured in a cost-benefit calculation. Another objection is that the discounting of future events which is integral to cost-benefit analysis will reduce most of the consequences of climate change to insignificance because they occur 50 to 100 years or even further in the future. Yet no rational person of our generation will accept, for example, that the consequences of a 6 m rise in sea level would be insignificant because it would occur 150-200 years from now.

The third objection, related to the first two, is based on awareness that we face vast uncertainty about the long-term consequences of climate change but that some of the consequences, e.g. sea level rise, could verge on the catastrophic in some important regions of the world, e.g. Bangladesh. A strict application of benefit-cost analysis to a situation combining such high uncertainty with the potential for such severe, perhaps irreversible, damage would be a child's game, not to be taken seriously by serious people.

I fully accept these objections to the strict application of benefit-cost analysis to consequences of climate change. Nevertheless, whatever policies mid-latitudes governments adopt to deal with climate change, whether to avert it or adapt to it, will cost something. In the strategic thinking needed to sort out the more promising policies, governments must consider the costs of the alternatives, even while recognizing that attempts to quantify all the costs would be a waste of time and resources. Similarly, in this strategic thinking governments must consider the benefits, in damages averted, of alternative policies, again recognizing the pointlessness of seeking quantification of the benefits. It is in this broad qualitative sense that I urge the usefulness, indeed the essentialness, of the concepts of benefits and costs in strategic thinking about policy responses to climate change.

What Makes A Policy Issue ?

I assert that the consequences of climate change for agriculture will raise policy issues for mid-latitudes governments when two conditions are met: (1) the consequences threaten to impose high social costs; (2) within the existing policy regime, the responses to the threat of individuals and institutions appear unlikely to bring the costs within socially acceptable limits. New policies, therefore, must be considered.

The costs which excite concern include those measured by market prices, e.g., the value of lost agricultural output, but they also include any losses of valuable but unpriced resources, for example, wetlands as habitat and regulators of floods, landscapes of esthetic value, the health of people not in or likely to be in the labor force, genetic diversity of plant and animal species, and land and water qualities with unpriced recreational value.

The unpriced costs may also include a variety of perceived national security losses. Governments now largely self-sufficient in production of food and fiber may view climate change threats to self-sufficiency as posing socially unacceptable costs. Part of the rationale for the European Economic Community's protectionist agricultural policies is maintenance of sufficient agricultural capacity to avoid the food shortages the continent suffered in two world wars. Agricultural capacity also is an element in the geopolitical balance of power. Countries for which climate change threatens a significant loss of capacity, therefore, may view this as posing socially unacceptable costs in the form of weakened geopolitical position.

Unpriced costs may also threaten to emerge out of the distribution of the consequences of climate change within generations, if the distribution is perceived to violate widely shared canons of fairness. In this case, for example, climate change which favors one region of a country (or of the world?) but disadvantages another creates a possible need for policies to prevent this from happening, or to compensate the losers if prevention promises to be too costly.

Impending high costs of climate change are a necessary condition for the emergence of policy issues, but they are not sufficient. The sufficient condition is that governments have reason to believe that in the absence of new policies, individual and institutional responses to the threat of higher costs will be less than socially optimal. This is of critical importance in strategic thinking about policy, particularly in weighing policies to slow climate change against those for adapting to it, and in sorting out alternative policies to encourage adaptation. People and institutions will surely seek adaptive responses to climate change, even if no new policies are adopted. In general, the lower the social cost of these responses, using our broad definition of social cost, the less the need for new policies. The resources available to governments to develop and implement new policies are always limited. It is essential, therefore, that governments carefully distinguish those climate change consequences which people and institutions can satisfactorily manage within the existing policy regime from those which would be manageable only with new policies.

Consequences of Climate Change for Mid-Latitudes Agriculture

The mid-latitudes are here defined broadly as those parts of the earth between the tropics and 60deg. north and south latitudes. Among principal agricultural producers this includes all of the United States, China, New Zealand, Pakistan and South Africa, virtually all of Argentina and Europe west of the Ural Mountains, most of the agriculturally important parts of Canada and the Soviet Union, and agriculturally significant regions of Australia and India. By this definition, most of the world's present agricultural capacity lies within the mid-latitudes.

The Background of Change

By the time global average temperatures have risen 1.5-5.5 deg.C, with related changes in precipitation, windiness, storm frequency and intensity, mid-latitudes agriculture and the world of which it is a part will look quite different than they do now. Such is the pace of change in world population, income, and agricultural technology that this will be true even in the unlikely event that warming on this scale occurs within the next two or three decades. Any discussion of the consequences of climate change for agriculture, therefore, must seek to discern, however dimly, where these background trends in world population, income and technology will have taken us several decades hence.

Growth of Demand. World population and income growth are important because they are the main determinants of the growth of world demand for food and fiber. The present population of the world is about 5 billion. World Bank projections (World Bank, 1984) indicate roughly 10 billion by the last quarter of the 21st century, with most of the growth occurring by 2050. Over 80% of the growth will be in the developing countries of Asia, Africa and Latin America.

Since 1960 world per capita income has been growing at an average annual rate of 3.0-3.5% (World Bank, 1984). Nordhaus and Yohe (1983) used a projection of little more than 1% annually in the 21st century. Crosson (1986a) assumed an annual rate of 2% in this period, noting that this was conservative by experience since 1960, and citing evidence (Goeller and Zucker, 1984) indicating that supplies of energy and other natural resources should be adequate to support such growth.

This quick look at the numbers suggests that population growth alone could increase world demand for food and fiber by 60 to 80% from the mid-1980s to the middle of the next century. Income growth in the developed countries will not add much to this because people in those countries already are well fed. In the developing countries, however, present levels of per capita income and nutrition are so low that a significant proportion (20-30%) of any additional income is spent on food. Per capita income growth in these countries, therefore, could add substantially to the growth of world demand for food and fiber. Indeed, the contribution could roughly match that of world population growth. Under this set of assumptions, world demand for food and fiber in 2050 could easily be 2 to 2.5 times the current level.

Growth of Supply. If prices of food and fiber are not to rise, the world will have to deploy in agriculture a combination of more, and more productive, resources such that the supply of food and fiber increases in step with the increase in demand. The required rate of increase in supply would be 1.0 to 1.4% annually.

By the standard of the last 35 years this does not appear to be a particularly formidable challenge. Over that period world production of food and fiber grew roughly by 2.5% annually, and the trend of prices, after adjustment for inflation, was down, indicating that the growth of supply generally outpaced the growth of demand (Crosson, 1986c). All major regions of the world, except Africa, shared this experience of rising agricultural production (Crosson, 1986c).

If history is a reliable guide, this performance suggests that, climate change apart, world agricultural capacity could accommodate the prospective increase in demand with ease. But history may not be a reliable guide, for two reasons. One is that the expansion in capacity in recent decades undoubtedly entailed a variety of environmental costs not reflected in world market prices of food and fiber. The evidence on this score, although anecdotal, is compelling, e.g. evidence about the rate and environmental consequences of tropical deforestation, and the vast expansion in the use of pesticides known to have high potential for environmental damage (Eckholm, 1976). There is no way of knowing whether these environmental costs increased enough to offset the decline in economic costs reflected in world market prices. But there is not much doubt that the combination of economic and environmental costs did not decline as much as prices.

The second reason is that the quantities of good land and water available to support additional agricultural production are more limited now than they were 35 years ago, especially in the developing countries. The evidence on this also is anecdotal, but it appears persuasive. Roughly 20 percent of the growth in crop production in the developing countries over the last few decades was owed to additional land, the rest being accounted for by higher yields per hectare (World Bank, 1982). It is plausible to believe that the land brought into production in this period was that best suited to that purpose. Latin America and Africa, but not Asia, contain substantial amounts of arable land in low intensity uses, but much of it is in remote areas, e.g. the Amazon, or faces other severe obstacles to more intensive use, e.g. the tsetse fly problem and river blindness in Africa (Crosson, 1986a). Bringing this additional land into production likely would be more costly than in the past.

Concern often is expressed (e.g. Brown et al., 1984) about the effects of soil erosion on the quality of the land. This concern is about land now in production. Significant erosion-induced losses of productivity on this land would weaken the ability of world agriculture to respond to the increases in demand implied by world population and income growth.

The best data about the amount of soil erosion and its productivity consequences are for the United States (Stocking, 1984). Analysis of these data indicates that continuation of present rates of cropland erosion for 100 years would reduce yields 5-10% below what they otherwise would be (Crosson, 1986b). Compared to the technology-based yield increases now in prospect (discussed below) this erosion-induced loss of yield is trivial. Soil erosion is not a serious threat to agricultural capacity in the United States.

Little is known about soil erosion and its productivity effects in other countries (El Swaify et al., 1982 on amounts of erosion; Rijsberman and Wolman, 1984 on it sproductivity consequences). Clearly there are places, especially in the developing countries, where erosion is high and its productivity consequences severe. But the evidence is very limited. Comprehensive estimates do not exist.

The quantity and quality of water available to support further expansion of world agriculture must also be more limited now than it was several decades ago. Over 160 million hectares of land were irrigated in the developing countries in 1982, an increase of over 60% from 1960 (World Bank, 1982). In the United States irrigated land increased 6 million hectares, 45% in this period. As it is plausible to believe that the expansion in agricultural land over the last several decades was on the better land, so it is probable that the growth of irrigation made use of lower cost sources of water. This is known to be the case in the United States (Frederick with Hanson, 1982). Moreover, the increase in irrigation has been accompanied by increasing problems of water quality, particularly rising salinity concentrations in irrigation return flows (Frederick with Hanson, 1982, for the U.S.; Eckholm, 1976, for developing countries). The potential for additional irrigation in the developing countries is considerable (Food and Agriculture Organization, 1979), but the direct cost of realizing the potential likely will be higher than in the past. Moreover, increased demand for water in industrial and municipal uses and for protection of water related ecological values will increase the opportunity cost of water in agriculture.

In the absence of continued advances in agricultural technology and institutions, prospective increases in world demand for food and fiber would put intense pressure on world supplies of land and water, with consequent sharp increases in both economic and environmental costs of agricultural production. Fortunately, the prospects for developing new land and water-saving technologies are good. The ground for expecting the needed institutional development appears somewhat less solid, but nonetheless gives some basis for optimism. Space does not permit development of this argument, but it relies heavily on the world's experience in developing and deploying the Green Revolution technologies, on studies of the potential for further technological advance (Office of Technology Assessment, 1985; English et al., 1984), and on the argument (the induced innovation hypothesis) developed most fully by Hayami and Ruttan (1985), that the emergence of rising scarcity of land and water resources induces the technical and institutional innovations necessary to offset the effects of the scarcity on production costs.*

The induced innovation hypothesis is subject to an important qualification. The inducement mechanism assumes that the agricultural research establishment is guided by rising prices of land and water to develop technologies to conserve these resources. To the extent that prices fail to signal emerging scarcity, the inducement mechanism falters. In countries where markets are the principal instrument for allocating resources, land markets work reasonably well in transmitting signals of land scarcity.** But water markets are rare, water typically being allocated by systems of 'rights' managed by governmental institutions. Often water is not priced at all, and where it is the price seldom bears any relationship to the scarcity value of the resource. For this reason the evidence of emerging scarcity of water is less clear than it is for land and the inducement to develop water-conserving technology, therefore, is weaker. How seriously this institutional defect may inhibit the expansion of world agricultural capacity is uncertain. I believe it will be an obstacle, but that as the scarcity of water mounts, governments will find ways around it. Not, however, without costly lags.

Comparative Advantage. A region's comparative advantage in the world agricultural economy depends in part on its endowment of land and water resources, but also, and increasingly, on the size and quality of its agricultural establishment, from farmers to the public and private research and other institutions which serve them. Not least among these are those institutions where agricultural and macroeconomic policies are made. For the last 40 years the combination of natural, human and institutional resources has particularly favored the United States, Canada, and Australia, and more recently the European Economic Community.

Consequences of climate change apart, the configuration of global agricultural comparative advantage in the middle of the 21st century is highly uncertain. There of course is no guarantee that the present configuration will persist. Indeed several countries - Brazil, Argentina, India and the People's Republic of China- already show signs of shaping a new configuration. These countries have land and water resources sufficient to make them major challengers for position in the world agricultural economy. What they have so far lacked are the investments in human capital and research capacity necessary to compete effectively in the science based agriculture which already characterizes the present and increasingly will characterize the future. India, at least, already has begun to make these investments (Plucknett and Smith, 1982). Whether the others will also remains to be seen, but as their economies grow and internal markets for food and fiber rise, they increasingly will have both the resources and the incentive to do as India has begun to do.

Food is so basic to national welfare, and uncertainties about world political alignments sufficiently great, that it probably is fair to say that, net exporters apart, most countries would like to be more self-sufficient in food than they are. And many probably already are more self-sufficient than they would be by a strict economic calculation of comparative advantage. Japan is the major example of this. Apart from the countries mentioned above, which clearly have the potential to become major if not dominant actors in the world agricultural economy, many other countries may seek greater food self-sufficiency over the decades ahead. The economic cost of doing this should become more bearable as their economies grow and their stock of human and other kinds of capital increases.

The implication of this scenario is decreasing international specialization in agricultural production. World trade in agricultural commodities would continue to grow in absolute amount, but it would decline relative to world agricultural production. Comparative advantage would be more diffused, and some countries now wielding little of it may emerge as principal actors. In this scenario, however, no country would have as large a share of world agricultural production and trade as the U.S. does today.

Summary. In the absence of climate change, the combination of population and income growth will make the world economy in mid-21st century 6.0 to 6.5 times as big as it is now. Over 80% of the population growth and 35-40% of the income growth will be in the presently developing countries of Asia, Africa and Latin America. These countries, which presently account for roughly one-fourth of the world's income will by then account for roughly one-third.

In this scenario, world production of food and fiber will be 2.0 to 2.5 times the present level by the middle of the next century, growing more slowly than the world economy because of biological limits on the amount of food people can consume. Virtually all the increase in output will be owed to advances in science and technology embodied in human and other forms of capital. Accordingly, the relative contribution of land and water resources to total agricultural production will be substantially less than it is today.

World agricultural trade will be higher in absolute amount but lower relative to world production. Production will be more diffused geographically than it is today, with relatively more production than at present in the developing countries of Asia, Latin America, and probably Africa. Most production still would be in the midlatitudes as here defined, but the percentage would be less because of the growing importance of Brazil and other tropical producers.

World prices of food and fiber at the farm gate will be lower than today, reflecting declines in economic costs of production. Environmental costs may be higher, however, especially in developing countries where the greater part of the production increase will occur. Damage from agricultural chemicals probably will not be much higher, and may be lower than at present as past management relies more on plant resistance and other forms of biological controls, and nitrogen is used more efficiently and more of it is provided by biological fixation. The more important sources of higher environmental costs would be loss of species and other habitat values and degradation of soil and water quality following conversion of forest land to crops and pasture, and the loss of ecological values resulting from the spread of large irrigation systems. How much environmental costs might rise, if they do, is quite uncertain, much depending on the sensitivity of policy people and the agricultural research establishment to emerging signals of environmental stress. However, because environmental costs are not priced, the signals of stress are not always clear, so technical and institutional responses to the stress may be delayed and off-target when they arrive. This is the main reason for believing that the prospective increase in world agricultural production may result in higher environmental costs.

Climate Change and Its Consequences

Global warming of 1.5-5.5deg.C and related changes in regional climates would alter this scenario of what the world will look like at mid-21st century. Present understanding of climate change impacts, however, gives no persuasive reasons for believing that world population or income growth will be significantly affected. Schelling (1983, p. 481) speculated that climate change resulting from a doubling of CO2 would not lower world living standards by more than a few percent. If he is right, then the income effect of climate change on the prospective scale would be trivial in a world where per capita income is higher by 260 percent. (Two percent compounded annually from 1985 to 2050). The population effect of climate change would be similarly small, to the extent that population growth is a function of income growth. I see no reason to believe that climate change on the prospective scale would have significant other direct or indirect effects on world population growth.

If total world population and income growth are little affected by the prospective climate change then the effect on growth in world demand for food and fiber also should be small. It does not necessarily follow, however, that the effect would be similarly small on the world's ability to expand agricultural capacity in step with demand. Should the expansion fall short, economic costs likely, and environmental costs surely, would rise. Even if aggregate world capacity grows equally with demand, climate change likely will alter the configuration of world comparative advantage in ways that pose both international and domestic policy issues for midlatitude governments. These possibilities must be considered further.

Consequences for World Agricultural Capacity. Climate change would most directly affect world agricultural capacity by changing global and regional temperature and water regimes and length of growing seasons. Changes in seasonal patterns of precipitation are likely as are changes in the probability of extreme weather events, e.g. drought or consecutive days of high temperatures. In general, a warmer world will be a wetter world, but specific regions, some of them important in mid-latitudes agriculture, might become drier. Firm judgments about this are impossible because of the well known failure of the general circulation models to agree about the regional climate implications of global warming. Nonetheless, some "what if" speculation can be fruitful in thinking about these issues.

Manabe and Wetherald (1986) of NOAA's Geophysical Fluid Dynamics Laboratory (GFDL) at Princeton University used their global climatic model to study the changes in global and regional climates that would occur if atmospheric concentrations of CO2 were to rise from 300 ppmv to 600 ppmv. (Current concentrations are about 340 ppmv). They found that the increased temperature resulting from the higher concentration resulted in significantly less summer moisture in continental North America, western Europe and Sibria. The reason was reduced precipitation and greater evaporation forced by the higher temperature.

Drawing on research by Langbein et al (1949) and Stockton and Boggess (1979), Revelle and Waggoner (1983) examined the effects of higher temperatures and reduced precipitation on water supply in the American west. Considering the 7 major river systems in the region, they concluded that a 2deg.C increase in temperature coupled with a 10% decrease in precipitation would reduce average annual runoff of surface water by 53%. Even with no increase in demand for water, the ratio of demand to supply would rise from 0.43 to 0.90. Revelle and Waggoner do not consider future demands for water in the region, but by the middle of the next century they surely will be greater than at present.

The decline in precipitation depicted by Revelle and Waggoner would imply diminished supplies also in aquifers dependent for replenishment on infiltration of precipitation and run-off.

Revelle and Waggoner assert that results similar to those in the western United States can be expected in many water-short regions elsewhere in the world. Humid regions, however, evidently would not be seriously affected. Stockton and Boggess found that the assumed 2deg.C temperature increase and 10 percent precipitation decrease would have little effect on runoff in the part of the United States east of the 100th meridian, which bisects the Great Plains states.

Several investigators have found that in some regions higher temperatures will diminish water supplies even if annual precipitation is increased. Mather and Feddema (1986) used results from the global climatic models of GFDL and the Goddard Institute for Space Studies (GISS) at Columbia University to examine the consequences for 'water budgets' of climate change induced by a doubling of the atmospheric concentration of CO2. They considered the consequences in 12 large regions: south central Canada, the Argentine Pampas, the Soviet Ukraine, north central Siberia, southeast China, west central Africa, southern Africa, southeast Australia, the Pacific northwest and upper midwest of the United States, and a region including the state of Texas and north central Mexico. They summarized their findings as follows:

(1) Both models showed increases in temperature, hence in potential evapotranspiration in all 12 regions;

(2) The GFDL model showed precipitation up in 8 of the 12 regions and down in 4;

(3) With the GISS model precipitation increased in 10 regions and declined in 2;

(4) Because the temperature increases generally outweighed the precipitation increases, most regions showed an increase in the annual water deficit (the difference between potential and actual evapotranspiration), and a decrease in summer soil storage.

In a study of the Sacramento basin of California, which provides over 30% of the runoff for the whole state and virtually all the water for agriculture in the state's Central Valley, Gleick (1986) examined the consequences for runoff of alternative combinations of increased temperature and more or less precipitation than at present. The assumed temperature increases were 2 and 4 deg.C and the precipitation assumptions were +20, +10, 0, -10, -20% from present levels. Compared with the present, all ten combinations showed reductions in summer runoff and in soil moisture in the agricultural parts of the basin. Winter runoff increased in 7 of the 10 combinations, as more precipitation occurred as rain, less as snow. Gleick concluded (p. 217) that "The most profound effect of such climatic changes may be major alterations in regional hydrological cycles and changes in regional water availability".

These findings suggesting that global warming will cause increased evapotranspiration and reduced runoff and summer soil moisture are for parts of the midlatitudes. Parry and Carter (1987) assumed a doubling of atmospheric CO2 and used the GISS model to study consequences for climate in cool temperate and cold regions. They found that, unlike in the lower portions of the mid-latitudes, precipitation would increase more than enough to offset the increase in evapotranspiration caused by higher temperatures. In cool temperate and cold regions, therefore, the water balance would become more favorable to agriculture.

The key question for the mid-latitudes is how higher temperatures and altered precipitation regimes would affect crop yield and other measures of agricultural productivity, account being taken also of the productivity effects of increased atmospheric concentrations of CO2. Where CO2 is the principal factor limiting plant growth, increasing it tends to increase growth. The effect is generally much stronger in C3 plants, (most of the world's crop species, but not corn and sorghum, Warrick et al. 1986) than in C4 plants. Few studies of the productivity consequences of climate change examine both the CO2 enhancement effects and the effects of temperature and precipitation changes. The few which have (e.g., Rosenberg, 1981 and 1982; Waggoner, 1983) note that while scientific theory and laboratory experiments concur that increased atmospheric CO2 increases plant productivity, especially in C3 species of plants, evidence for this under field conditions so far is lacking. Rosenberg (1982) concludes, nonetheless, that as CO2 concentrations rise, the productivity enhancement effect should begin to take hold. For C3 plants the effect could be substantial. Warrick et al. (1986) cite experimental evidence that doubling atmospheric CO2 from the present 340 ppmv could increase yields of C3 plants 10-50%. At least 95% of the earth's biomass is of the C3 type. Moreover, 12 of the 15 major crops in the world, including wheat and soybeans but not maize or sorghum, are C3 (Waggoner,1983).

Rosenberg (1982) does not estimate the yield effect of increased atmospheric CO2, nor the combined effect with higher temperature and changed precipitation and evapotranspiration. He is critical, however, of the way evapotranspiration is parameterized in various versions of the Manabe-Wetherald model (1969, 1980, 1986) which show increased dryness in middle North America. Rosenberg (1982) cites experimental evidence that evapotranspiration from alfalfa in Nebraska often exceeds the amount that can be explained by net solar radiation. The additional factor is heat carried by the wind from the south and southwest. Since global warming would reduce the equator-to-poles temperature gradient, it could also reduce windiness. Further, some model runs indicate that these sources of heat carried by the wind to the American midwest might receive more rainfall with global warming, which would reduce the energy carried by wind into the region.

Waggoner (1983) concluded that an increase in atmospheric CO2 from the present 340 ppmv to 400 ppmv would increase yields of field crops in the American Cornbelt by some 5%, but that the associated warmer and drier climates (drawing on Manabe and Wetherald, 1980), would reduce yields by 3 to 12%, depending on the crop and the part of the region.

Bolin, Jager and Döös (1986) found that studies of the crop yield effects of climate change generally show that with no change in precipitation, a warming of 2 deg.C might reduce average yields of maize and wheat in the mid-latitudes of North America and Western Europe by 10 +/- 7%, assuming no change in cultivars, technology or management. The yield declines would be less than this where warming is associated with increased precipitation and more where precipitation declines. These findings are confined to wheat and maize in middle North America and Western Europe because most of the research on yield effects of climate change have focused on these crops in these areas (Warrick et al., 1986).

The studies referred to consider the consequences of CO2-induced climate change on crop yields. But the developmental processes which produce rising CO2 emissions also produce increasing amounts of chlorofluorocarbons and nitrous oxides, which can diminish stratospheric ozone, and hydrocarbons which contribute to increasing tropospheric ozone. The increase in ultraviolet radiation consequent upon reduced stratospheric ozone could adversely affect crop productivity, although the matter has been little studied (Teramura, 1986). And rising concentrations of tropospheric ozone also are a threat to crop yields (Kopp and Krupnick, 1987).

The global warming now predicted could raise the sea level 1 meter or more within the next century. Most of the world's agricultural capacity would not be affected by this, although capacity in Bangladesh and Egypt could be (Titus, 1986). Coastal wetlands, however, would be severely affected. Titus (1986) asserts that a 1- to 2-m sea level rise would destroy 50-80% of U.S. coastal wetlands, and that this probably is representative of global effects. While not of major importance in the world agricultural economy, coastal wetlands provide important habitat and other environmental values.

The discussion suggests that for world agricultural capacity a principal, if not the principal, impact of climate change would be on hydrological regimes. The finding that warming in parts of the mid-latitudes generally would result in reduced summer runoff and soil moisture, in some circumstances even when precipitation is increased, appears highly significant. But caveats abound. The general uncertainty about impacts of warming on regional climates is only one. Another is that hydrological impacts of climate change have been little studied and are poorly understood (Beran, 1986; Gleick, 1986). Yet another stems from Rosenberg's point (1982) that models of regional climate change and studies of consequent water balances may overestimate the negative impact of warming on water supply because of unrealistic calculations of evapotranspiration.

As far as they go, the studies of climate change impacts on crop yields do not suggest a major threat to global agricultural capacity by mid-21st century. Even if yield loss were at the high end of the range noted by Bolin et al. (1986) - 17% - the loss would look small compared with the doubling or more of yields which even modest (by historical standards) advances in technology and management would bring about. Moreover, the studies of yield impact typically do not include the positive yield effects of increased atmospheric CO2; and they implicitly ignore some factors which could offset some of the effect of higher temperatures on evapotranspiration. Finally, virtually all the studies of yield loss are focused on two crops wheat and maize - in two regions - the American middle west and western Europe. Even if the yield effects on these crops and regions were highly and conclusively negative, this would not be enough to support conclusions about consequences for global agricultural capacity. In any case, for reasons given earlier, the relative importance of these regions in world agriculture may decline by the middle of the next century, quite apart from impacts of climate change.

On balance, the available evidence does not indicate that the global warming now expected by mid-21st century will pose a major threat to the world's ability to expand agricultural capacity in step with demand. The implication is that on a global scale, climate change will not add significantly to the combined economic and environmental costs of agricultural production. However, a major caveat applies: the evidence supporting this conclusion is thin. Consequently, the conclusion must be considered highly tentative. It is useful as a guide to current thinking about the agricultural consequences of global warming, but it is not a sufficient base for specific policies to deal with the consequences.

Regional Distribution of Capacity. Even if global agricultural capacity is not seriously affected, capacity in some regions may be severely impaired. The inability of the general circulation models to agree on the regional climate impacts of global warming is a major obstacle to identifying the winning and losing regions. As far as they go, however, the models suggest a northward shift in the world's grain belt of several hundred kilometers per DC increase in temperature (Bolin et al., 1986; Ciborowski and Abrahamson, 1986). If this is correct, then global warming of 1.55.5 deg.C over the next 50-75 years could shift the world's grain belt 500-1500 km to the north by the middle of the next century. The larger shift would eliminate much of the wheat-maize-soybean producing capacity of the United States. It would increase capacity to produce these crops in Canada and the Soviet Union, to the extent that their capacity now is limited by temperature. This is an important qualification. In Canada soils may be more limiting than temperature, and in both countries the number of days with adequate light may be more limiting than temperature.

Temperature gradients are key to defining the geographical extent of agriculture in the mid- and high latitudes. In the tropics precipitation is key (Warrick et al., 1986). It defines the growing season, and also defines gradients along which crop varieties, planting dates, and management practices differ widely. Shifts in these gradients resulting from global warming could have strongly negative productivity effects in regions adapted to the present regime (Warrick et al., 1986). The consequences of global warming for precipitation in the tropics, however, are highly uncertain.

Significant shifts in agricultural comparative advantage could occur along a narrow latitudinal band within the mid-latitudes. This is illustrated by the finding of Stockton and Boggess (1979) that an assumed 2 deg.C temperature increase and 10% precipitation decrease in the United States would sharply reduce runoff west of the 100th meridian but not east of it. This situation would greatly increase the cost of irrigated agriculture in the west relative to rainfed agriculture in the east. Although the northerly shift in the grain belt would reduce total U.S. agricultural capacity, the reduction would be more pronounced in the west. Whether similar shifts in comparative advantage would occur elsewhere within the mid-latitudes is uncertain. But why should the United States be unique in this respect?

Policy Issues for Mid-Latitudes Governments

Identifying the Issues

Economic and Environmental Costs. If barriers to trade in agricultural commodities become no more limiting than in the without climate change scenario, then the effects of climate change on the economic and environmental costs of agricultural production should raise no major policy issues for mid-latitudes governments. This is the implication of the conclusion stated above that on presently available evidence prospective climate change should not impede the expansion of world agricultural capacity in step with demand. So long as climate change does not induce an increase in trade barriers, mid-latitudes countries would be able to import commodities they could no longer economically produce at about the same economic and environmental costs as in the without climate change scenario.

Trade in Agricultural Commodities. The assumption of no strengthening in barriers to trade in agricultural commodities is essential to this assessment. Countries gaining comparative advantage through climate change would have no interest in seeing trade barriers strengthened. On the contrary, they could be expected to become advocates of freer trade. If protectionist sentiment rises it most likely would be among those mid-latitudes countries facing the loss of comparative advantage. Agricultural interests in those countries probably would pressure their governments to raise barriers against imports from countries favored by climate change.

There is an irony here. Under the conditions depicted, mid-latitudes countries losing comparative advantage would be able to avoid rising economic and environmental costs of agricultural production only if they could import freely from emerging lower cost producers. But it is precisely in the disadvantaged countries that agricultural protectionism likely would strengthen. One likes to assume that in responding to climate change, mid-latitudes governments will seek policies which serve the public interest. The issue described here is one - only one - that would put that assumption to the test.

National Security. By permitting rising imports, mid-latitudes countries losing comparative advantage could avoid higher costs of agricultural commodities. But governments in these countries might see this situation as posing unacceptably high social costs in other forms. The shift would mean a loss of self-sufficiency in agricultural production, and this may be seen as a threat to national security. The greater the shift the more serious the threat likely would appear.

The reduction in agricultural capacity may also be viewed as a threat to the geopolitical position of some mid-latitudes governments. The U.S. would appear to be the country most exposed in this respect. Over the last 30 years the U.S. emerged as by far the dominant country in world agricultural trade, and the supplier of last resort when world agricultural output was depressed, as in 1973-74. This surely gave the U.S. leverage in its dealings with other countries. Should climate change weaken or eliminate that leverage, in particular if it should simultaneously strengthen the position in world agriculture of the U.S.'s chief rival in world affairs, the Soviet Union, then the U.S. government might view climate change as posing serious policy issues on this score.

Regional Decline. The decline in agricultural capacity in some mid-latitudes countries implies a decline also in economic activity linked to agriculture in those countries. Some aspects of this climate-induced decline likely will pose policy issues, but others should not. It is important that mid-latitudes governments be alert to the difference. The key is relating the time scale of the climate-induced changes to the time scales of the decisions affected by the changes (Clark, 1985). The climate-induced changes depicted here will evolve over a period of 50-75 years. This is a long time for many of the economic decisions that will be induced by the changes. People's decisions to stay in agriculture and agriculturally related activities, to invest in new equipment, more animals, a new barn, a fertilizer or seed supply business, are made on time scales of perhaps 5 to 10 or 20 years. As the climate-induced decline in agriculture gradually becomes evident, these decisions can be adjusted accordingly with little if any economic loss. It would be critical to this, however, that mid-latitudes governments maintain macro-economic policies to assure that human and other resources forced out of agriculture by climate change could find employment in other sectors of the economy. Given this condition, many of the regional adjustments forced by climate-induced declines in agriculture could be made at acceptable social costs. New policies would not be needed.

Longer-lived investments in roads, public utilities, schools and other community infrastructure could be more exposed. Even these, however, may find low cost adjustment opportunities. Community infrastructure investments generally are not specialized in the service of agriculture. If other economic activity expands as agriculture declines, these investments can be protected. The growth in the last several decades of millions of non-farm jobs in rural areas of the United States suggests this is not idle speculation. Again, the viability of this adjustment mechanism would require a generally vibrant national economy. Given this, new policies would not likely be needed.

Water Resources. Investments in development of water resources for use in agriculture are on a time scale similar to that of the postulated climate change. And the earlier discussion suggested that at least in arid and semi-arid parts of the midlatitudes, climate change consequences for water resources could be severe. Total runoff could be sharply reduced and its seasonal pattern radically changed. This suggests that countries considering investments in large water resource projects, especially if they are to provide flood control and power generation in addition to irrigation, should even now bring the possibility of climate change into their calculations.

But climate change consequences for water resources likely will raise important policy issues apart from those concerning the viability of long term investments. In some parts of the mid-latitudes, e.g. the American west, competition for water is mounting, and in some cases already is intense. Irrigated agriculture is by far the major consumer of water in these areas, but population and economic growth are rapidly increasing demand for industrial and municipal uses. These same processes also increase the recreational value of water. Moreover, as these various demands exert rising pressure on the available supply it becomes increasingly clear that unpriced ecological values of water - water and wetlands as habitat for a variety of plants and animals - are under a rising threat.

Because water markets generally are weak or non-existent, and important ecological values of water are unpriced, water management everywhere is a concern of public policy. The increasing competition for water implied by continued population and economic growth would pose mounting challenges to these policies in mid-latitudes countries even in the absence of climate-induced reductions in water supply. Such reductions, especially if they were on the scale suggested by Revelle and Waggoner (1983) for the American west, would greatly complicate an already complex set of policy issues.

The time scale for the possible climate-induced reduction in water supply is comparable to that for some of the water management decisions mid-latitudes governments might have to consider. The processes driving regional population and economic growth have powerful momentum. If growth is to be slowed, or redirected to regions not threatened by climate-induced reductions in water supply, then it is not too soon for governments to start thinking about policies to that end. The same can be said about investments in regional and municipal water supply systems, which may easily have time scales of 50 to 75 years.

Thinking Strategically About Policy

In principle, three broad strategies can be defined; (1) avert further global warming or hold any increase to a level that would pose no serious problems; (2) slow the rate of warming to give countries more time to devise strategies for dealing with the consequences; (3) accept whatever warming occurs and concentrate on adaptive strategies.

On the assumptions underlying this discussion, mid-latitudes governments will evaluate the attractiveness of these alternative strategies by weighing their relative social costs and benefits to them. Governments which see their countries winning, or not losing, from unrestrained global warming will have strong incentive to opt for the third strategy, for themselves and for the world community. Other governments will prefer one of the other two strategies. But neither strategy may be available to them. Global warming will not be slowed (strategy 2) or averted (strategy 1) unless all countries which contribute significant amounts of CO2 and other greenhouse gases agree to one or the other strategy. Such agreement would be difficult to achieve. Countries with large supplies of coal or other fossil fuels and mounting energy demands, and who believe they will not lose, perhaps will even gain, from climate change, will see strategies 1 and 2, particularly 1, as subjecting them to high costs and few, perhaps even negative, benefits. Moreover, control of CO2 emissions would deal with only a part of the problem. Other greenhouse gases also would have to be brought under control, and for some of them the costs of control could be high. Sources of methane, for example, are numerous and widely dispersed around the world. And apart from methane, some of them produce important social values, e.g. rice paddies for food and wetlands for habitat and flood control. The possible loss of these values would have to be added to the direct costs of controlling methane from these sources.

The judgment here that climate change over the next 50-75 years would not seriously impede the expansion of world agricultural capacity, and that some countries likely would gain comparative advantage in agriculture, implies that concerns about agriculture would not promote international agreements to adopt strategies 1 or 2. So far as agriculture is concerned, the adaptionist strategy would dominate, in countries threatened by loss of agricultural comparative advantage as well as those who see themselves as winners (or not losers).

But the prospects for strategies 1 and 2 are not as poor as this indicates, for two reasons. First, the assessment for world agriculture is for the middle of the next century, assuming global warming by then of 1.5-5.5 deg.C. Even if that correctly depicts the warming trend, governments cannot now be sure that it does, and must allow for a positive probability that the increase by mid-21st century will be significantly more than 5.5 deg.C. Even if they completely discount this probability, they must assume that the world will not come to an end in 2050. Unrestrained warming of 1.5-5.5 deg.C between now and 2050 would suggest the possibility of global temperature increases substantially greater than 1.5-5.5 deg.C by 2100. Even allowing for continued advances in agricultural technology and institutions, the end-of-century assessment for world agriculture could look significantly less benign than that for mid-century, and the likelihood of any countries emerging as better off because of unrestrained climate change significantly diminished.

The second reason for not giving up on strategies 1 and 2 is that the implications of 1.5-5.5deg.C warming by mid-21st century for non-agricultural economic activity, human health, and environmental values not related to agriculture might appear considerably more threatening than those for agriculture. Countries will have to consider the totality of climate change consequences in evaluating strategies 1, 2, and 3, not just those for agriculture.

The world community almost surely would have more difficulty getting agreement on strategy 1 - avert further global warming or hold it to non-damaging levels - than on strategy 2. (Strategy 3 would not require international agreement). Strategy 1 either would require countries rich in fossil fuel resources, some of which are poor countries, e.g., the People's Republic of China, to make important sacrifices in future economic growth, or to quickly find economical alternative energy sources, some of which, e.g. nuclear power, have high current economic costs and excite environmental concerns. Moreover, all countries currently, or potentially, emitting significant quantities of other greenhouse gases would have to control these within globally non-threatening limits. Strategy 2 - slow but not halt future warming and seek low cost ways of adapting to the consequences - might also imply some sacrifice of economic growth or accelerated development of non-fossil energy and control of non-CO2 greenhouse gases, but the costs of these measures probably would be substantially less than for strategy 1. Strategy 2, however, implies costs of adaptation which are not entailed by strategy 1.

Clearly, governments will make little headway in evaluating the relative merits of the three strategies until they have some basis for judging the scale of the social costs of each and the related scale of benefits in terms of social costs averted.* For purposes of this discussion it is assumed that an international consensus will soon emerge that strategy 2 is the most promising of the three. Within this strategy, the rationale for policies to slow global warming is the effect of warming on the entire world economy and society, not just the effect on agriculture. For agriculture (or any other single sector of the economy or society) the question is what lines of policy are governments likely to find most promising in adapting to the consequences of climate change, whatever may be done to slow it down. The focus here is on mid-latitudes agriculture, but much of the discussion should apply to other regions as well.

Elements of an Adaptionist Strategy for Agriculture

The initial condition for fruitful thinking about policies for adapting agriculture to climate change is recognition by governments that change is likely, and on a scale outside the limits of recorded human experience. Henceforth, this recognition should explicitly inform all policies related to the long-run future of agriculture. Just as these policies now build in assumptions about trends in technological change, land use, regional development and so on, from now on they should include the assumption of climate change. When government agencies lay plans for long-term investments in water supply facilities, in regional infrastructure, in research capacity for agricultural science and technology, and in anything else related to the long-run future of agriculture, the likelihood of climate change should be explicit in their deliberations.

These deliberations will have to confront immediately the high current uncertainty about the consequences for regional climates of global warming of 1.5-5.5 deg.C over the next 50 to 75 years. Because of this uncertainty, no mid-latitudes government can confidently predict whether its agriculture will gain or lose from climate change. In this circumstance, the prudent course would appear to be to build the institutional and technical capacity needed for flexible response to climate change consequences as they emerge. If the consequences are favorable, this capacity will permit resources to flow into agriculture to take advantage of the new opportunities. If the consequences are unfavorable, the capacity will give governments options between promoting the flow of resources out of agriculture and developing technologies and management practices better adapted to the changing climate regime.

What are the principal characteristics of a capacity for flexible response? One is an institutional structure which facilitates the flow of resources among sectors of the economy and among regions, both within countries and across country borders. Given the likelihood that climate change would shift agricultural comparative advantage among countries, international institutions, and agreements, affecting movement of people and goods among countries could become especially important. The importance of permitting trade in agricultural commodities to follow shifts in comparative advantage already has been discussed. The ability of people to move in response to such shifts also could be important in country adaptations to climate change. Whether countries think they will win or lose from climate change effects on agriculture, they may find it useful to begin factoring the likelihood of change into their thinking about immigration policy.

Institutions which ease the movement of resources into and out of agriculture will facilitate adjustments to climate change, but lack of institutional obstacles to movement is not enough where climate change puts agriculture in decline. People must have positive incentives to move to non-agricultural employment and the basic skills needed to make the transition. Policies to ensure sustained economic growth and to equip people with the general education and specialized knowledge needed to perform adequately in the modern world thus have an important role to play in the adjustment of agriculture to climate change.

Capacity to do research in agricultural science and technology also is key to a posture of flexible response in adapting to climate change. This is true whether countries are favored or disfavored by change, but especially if they are disfavored. Such countries may view a climate-induced loss of agricultural comparative advantage as an unacceptable threat to food self-sufficiency or geopolitical position. For such countries institutional flexibility facilitating the flow of resources out of agriculture will not serve because the policy objective is not to adjust to agricultural decline but to prevent it. Prevention will require development of new technologies and management practices which permit farmers in the adversely affected regions to continue to produce economically. Rosenberg's study (1982) of the spread of hard red winter wheat production in the United States across temperature differences greater than the temperature changes now in prospect suggests that such technologies and management practices can be developed. Moreover, the possibility of substituting better adapted crops or newly developed species cannot be discounted. But the capacity to do the necessary research must be created and sustained over the long-term. Some countries are big enough and rich enough to build this capacity for themselves, but many are not. The system of international food research institutions making up the Consultative Group for International Agricultural Research (CGIAR) could serve as a model of how to build the agricultural research capacity many countries will need in developing the technologies and management practices needed to adapt to climate change.

Countries wishing to guard against the risk of climate-induced loss of agricultural capacity may be tempted to adopt policies to hold land and water in agriculture which otherwise would shift to non-agricultural uses. The rationale for such policies would be that since land and water are vital to agricultural production, holding a stock of them in reserve beyond what ordinary economic processes would provide gives some insurance against the risk of an eventual climate-induced loss of capacity.

I expect such policies would be misguided. It is of course true that land and water are necessary for agricultural production, and this will remain so for the foreseeable future. However, the amount of each per unit of output is variable, depending upon the state of technology. The fact is that the relative contribution of land and water to agricultural production has steadily diminished in the last three or four decades as knowledge and other forms of capital have been substituted for them. This process is evident all around the world, in developed and developing market economies as well as in centrally planned economies. The process - technological change, broadly interpreted - is an inevitable part of the larger process of economic development, and it is implicit in the projections of world agricultural development discussed above. In that scenario, by the middle of the next century land and water will be even less important in world agricultural production than they are today. Efforts to prevent that from happening would exact high opportunity costs in the form of non-agricultural output foregone. It is likely that agricultural capacity would be much more economically protected against adverse effects of climate change by investing in capacity to do research in agricultural science and technology than by preventing land and water from shifting out of agriculture. The latter is the opposite of a policy of flexible response.

* For an extended discussion of the argument summarized here as it bears on prospects for world agricultural development through the end of the next century, see Crosson (1986a).

** Except where property rights in land are ill-defined. This is the 'tragedy of the commons' problem discussed by Hardin (1968), and many others. It undoubtedly is important in some places, particularly as a factor in tropical deforestation. In a global perspective on the supply of agricultural land, however, it probably is secondary.

* The ability of the major countries involved to agree in Montreal in 1987 on objectives and measures to control emissions of CFCs gives some encouragement that agreements in time could be reached to limit emissions of CO2 and other greenhouse gases. The analogy should not be pressed too hard, however. The costs of limiting CO2 and the non-CFC emissions are likely to be far higher and the incidence of the costs among countries far wider than costs of limiting CFCs. On both counts agreements to limit non-CFC emissions will likely be harder to achieve.


Ausubel, J.H.: 1983, 'Can We Assess the Impacts of Climatic Change', Climatic Change 5, 7-14.

Beran, M.: 1986, 'The Water Resource Impact of Future Climate Change and Variability,' in J. Titus, (ed.), Effects of Changes in Stratospheric Ozone and Global Climate, Vol. 1, Overview, U.N. Environment Programme and U.S. Environmental Protection Agency, Washington, D.C.

Bolin, B., Jager, J., and Doos, B.: 1986, 'The Greenhouse Effect, Climate Change and Ecosystem: A Synthesis of Present Knowledge,' in B. Bolin, B. Doos, J. Jager, and R. Warrick (eds.), The greenhouse Effect, Climatic Change and Ecosystems, John Wiley and Sons, New York.

Brown, L., Chandler, B., Flavin, C., Postel, S., Starke., L., and Wolf, E.: 1984, State of the World, World Watch Institute, Washington, D.C.

Cibrowski, P, and Abrahamson, D.: 1986, 'The Granery and the Greenhouse Problem,' in C. Runge (ed.), The Future of the North American Granery, Iowa State University Press, Ames.

Clark, W.: 1985, 'Scales of Climate Impact,' Climatic Change 7, 5-27.

Crosson, P.: 1986a, 'Agricultural Development - Looking to the Future', in W. Clark and R. Munn (eds.), Sustainable Development of the Biosphere, Cambridge University Press, Cambridge.

Crosson, P.: 1986b, 'Soil Erosion and Policy Issues', in T. Phipps, P. Crosson and K. Price (eds.), Agriculture and the Environment, Resources for the Future, Washington, D.C.

Crosson, P.: 1986c, 'Sustainable Food Production,' Food Policy 11, 43-56.

Eckholm, E.: 1976, Losing Ground, W.W. Norton, New York.

El-Swaify, S., Danger, E., and Armstrong, C.: 1982, Soil Erosion by Water in the Tropics, University of Hawaii, Honolulu.

English, B., Maetzold, J., Holding, B., and Heady, E.: 1984, Future Agricultural Technology and Resource Conservation, Iowa State University Press, Ames.

Food and Agriculture Organization: 1979, Toward the Year 2000. Rome.

Frederick, K., with Hanson, J.: 1982, Water for Western Agriculture, Resources for the Future, Washington, D.C.

Gleick, P.: 1986, 'Regional Water Resources and Global Climatic Change in J. Titus (ed.), Effects of Changes in Stratospheric Ozone and Global Climate, U.N. Environment Programme and U.S. Environmental Protection Agency, Washington, D.C.

Goeller, H. and Zucker, A.: 1984, 'Infinite Resources: the Ultimate Strategy', Science 223,456-462.

Hardin, G.: 1968, 'The Tragedy of the Commons Science', 162.1243-48.

Hayami, Y. and Ruttan, V.: 1985, Agricultural Development: an International Perspective, Johns Hopkins University Press, Baltimore.

Kopp, R. and Krupnick, A.: 1987, 'Agricultural Policy and the Benefits of Ozone Control Paper given at the Summer meeting of the American Agricultural Economics Association, Michigan State University, East Lansing. Forth coming in the American Journal of Agricultural Economics.

Langbein, W.: 1949, Annual Runoff in the United States, U.S. Geological Survey circular 5, U.S. Department of Interior, Washington, D.C.

Manabe, S. and Wetherald, R.: 1986, 'Reduction in Summer Soil Wetness Induced by an Increase in Atmospheric Carbon Dioxide in J. Titus (ed.), Effects of Changes in Stratospheric Ozone and Global Climate, U.N. Environment Programme and U.S. Environmental Protection Agency, Washington, D.C.

Mather, J. and Feddema, J.: 1986, Hydrologic Consequences of Increases in Trace Gases and CO2 in the Atmosphere in J. Titus (ed.), Effects of Changes in Stratospheric Ozone and Global Climate, U.N. Environment Programme and U.S. Environmental Protection Agency, Washington, D.C.

Nordhaus, W. and Yohe, G.: 1983, 'Future Paths of Energy and Carbon Dioxide Emissions in Changing Climate, National Academy Press, Washington, D.C.

Office of Technology Assessment: 1986, Technology, Public Policy and the Structure of American Agriculture, Government Printing Office, Washington, D.C.

Parry, M. and Carter, T.: 1987, Assessment of Climatic Variations on Agriculture, Vol . 1, Assessments in Cool Temperate and Cold Regions, Reidel, Dordrecht, The Netherlands.

Plueknett, D. and Smith, N.: 1982, 'Agricultural Research and Third World Food Production. Science 217,215-220.

Revelle, R. and Waggoner, P.: 1983, 'Effects of a Carbon Dioxide-Induced Climatic Change on Water Supplies in the Western United States in Changing Climate, National Academy Press, Washington, D.C.

Rijsberman, F. and Wolman, M.: 1984, Quantification of the Effect of Erosion on Soil Productivity in an International Context, Delh Hydraulic Laboratory, Delft, The Netherlands.

Rosenberg, N.: 1981, 'The Increasing CO2 Concentration in the Atmosphcre and Its Implcation on Agricultural Productivity. I. Effects on Photosynthesis, Transpiration and Water Use Efficiency Climatic Change 3, 265- 179.

Rosenberg, N.: 1982, 'The Increasing CO2 Concentration in the Atmosphere and Its Implication on Agricultural Productivity. II. Effects through CO2 Induced Climate Change Climatic Change 4, 239-254.

Rosenberg, N.: 1986, 'Climate, Technology, Climate Change and Policy: The Long Run in C. Runge (ed.), The Future of the North American Granary, Iwa State University Press, Ames,

Sehelling, T.: 1983, 'Climate Change: Implications for Welfare and Policy in Changing Climate: Report of the Carbon Dioxide Assessment Committee, National Academy of Sciences, Washington, D.C,

Stoeking, M.: 1984, Erosion and Soil Productivity in Review, Food and Agriculture Organization, Rome.

Stockton, C. and Boggess, W.: 1979, Geohydrological Implications of Climate Change on Water Resource Development, U.S. Army Coastal Engineering Research Center, Fort Belvoir, Virginia.

Teramura, A.: 1986, 'Overview of Our Current State of Knowledge of UV Effects on Plants in J. Titus (ed.), Effects of Changes in Stratospheric Ozone and Global Climate, Vol. 1, Overview, U.N. Environment Programme and U.S. Environmental Protection Agency, Washington, D.C.

Titus, J.: 1986, 'The Causes and Effects of Sea Level Rise in J. Titus (ed.), Effects of Changes in Stratospheric Ozone and Global Climate, Vol. 1, Overview, U.N. Environment Programme and U.S. Environmental Protection Agency, Washington, D.C.

Waggoner, P.: 1983, 'Agriculture and a Climate Changed by More Carbon Dioxide', in Changing Climate, National Academy Press, Washington, D.C.

Warrick, R., Gifford, R., and Parry, M.: 1986, 'Assessing the Response of Food Crops to the Direct Effects of Increased CO2 and Climatic Change in B. Bolin, B. Doos, J. Jager, and R. Warrick (eds.), The Greenhouse Effect, Climatic Change and Ecosystems, John Wiley and Sons, New York.

World Bank.: 1982, World Development Report 1982, Oxford University Press.

World Bank.: 1984, World Development Report 1984, Oxford University Press.

(Received 16 February, 1988; in revised form 27 December, 1988)