CIESIN Reproduced, with permission, from: Nordhaus, W. D. 1991. Economic approaches to greenhouse warming. In Global warming: Economic policy approaches, ed. R. D. Dornbush and J. M. Poterba, 33-68. Cambridge, MA: MIT Press.

Economic Approaches to Greenhouse Warming

William D. Nordhaus

There is growing international concern about "greenhouse warming," which is a scientific theory that predicts that increases in CO2 and other atmospheric gases may produce significant climatic changes over the next century.(1) This concern is only one of a number of areas where population and economic growth have threatened to have significant impacts upon the global environment. Other concerns include increasing evidence of widespread damage from acid rain; the appearance of the Antarctic "ozone hole," interpreted by some as the harbinger of global ozone depletion that threatens to remove the shield that protects organisms from harmful ultraviolet radiation; deforestation, especially in the tropical rain forests, which may upset global and local ecological balance; and a depletion of genetic resources that arises from urbanization and other impacts upon major ecosystems.

Global environmental problems raise a host of major policy questions. They are all scientifically complex and controversial, and no scientific consensus is likely to emerge until irreversible decisions have been made. The costs and benefits of these changes transcend national boundaries, and nations, which cannot appropriate the global costs and benefits of such changes, are unlikely to be able or willing to make efficient decisions on how to combat these global externalities. In addition, these concerns sometimes have impacts over hundreds of years and thereby strain political decision making, which often functions effectively only when the crisis is at hand.

This chapter considers some of the economic issues involved in deciding how to react to the threat of global warming. I first review the theory and evidence on the greenhouse effect. I then present evidence on the impacts of greenhouse warming, the costs of stabilizing climate, and the kinds of adaptations that might be available. In the final section, I review the policy initiatives that nations might follow in the near term.

Scientific Issues

The Greenhouse Effect and Greenhouse Gases

For almost two centuries, scientists have suspected that changing the chemical composition of the atmosphere would alter our planet's climate. In the first careful numerical calculations, S. A. Arrhenius estimated in 1896 that a doubling of the atmospheric concentrations of carbon dioxide (CO2) would increase global mean temperature by 4deg. to 6deg. c.2

What causes the greenhouse effect? The atmosphere consists of several "radiatively active" gases that absorb radiation at different points of the spectrum. The "greenhouse gases" are transparent to incoming solar radiation but absorb significant amounts of outgoing radiation. The net absorption of radiation produces a happy result, raising the earth's temperature about 33deg. C (59deg. F). The greenhouse effect helps explain the hot temperatures on Venus along with the frigid conditions of Mars.

The concern about the greenhouse effect arises because human activities are currently raising atmospheric concentrations of greenhouse gases and threatening a significant and undesirable climate change. The major greenhouse gases (GHGs) are carbon dioxide, methane, nitrous oxides, and chlorofluorocarbons (CFCs). Scientific monitoring has firmly established the buildup of the major GHGs. Table 2.1 shows the important GHGs, recent and projected concentrations, and the past and estimated future growth rates of major GHGs.

Greenhouse gases differ greatly in their quantitative impact on climate because they have different radiative properties and different lifetimes. Table 2.2 shows the important greenhouse gases, their "instantaneous" and "total" contribution to global warming,(3) and the industries in which the emissions originate. Carbon dioxide is the major contributor to global warming, with most CO2 emissions coming from the combustion of fossil fuels. Of the fossil fuels, natural gas has 60 percent as much CO2 emissions per unit energy as coal, and petroleum has 80 percent as much CO2 per unit energy as coal. The second most important source of GHG emissions is the CFCs, which are small in volume but have a warming potential almost 20,000 times as powerful as CO2 per unit of volume.

In analyzing policies, it is important to find a common unit of measurement for different gases. In this paper, I will translate all impacts and costs into a common unit, the "CO2 equivalent" of GHG emissions, measured in terms of the carbon content of CO2. By working in this metric, we can attempt to ensure cost effectiveness of policies in different sectors.

Climate Models and Forecasted Climate Change

There is no dispute about the buildup of greenhouse gases in the atmosphere. To project climate changes many years into the future, however, requires use of climate models that trace out the effect of a changing radiative balance upon major climatic variables. Because we are heading into uncharted waters, the models cannot rely upon historical experience but must extrapolate beyond current observations.

Climate models are mathematical representations of important variables such as temperature, humidity, winds, soil moisture, and sea ice. Large "general-circulation models," or GCMs, simulate changes in weather, in steps of a few minutes, over a century or more. The largest models use 500-kilometer-square grids through several layers of the atmosphere. Such models are unfortunately extremely expensive to run, and a single CO2 scenario might take the largest supercomputer nearly a year to calculate.

Three important features of the results emerge from existing studies. First, the central estimates of the equilibrium impact of a CO2-equivalent doubling have changed little since the earliest calculations. The last thorough review by the National Academy of Sciences concluded, "When it is assumed that the CO2 content of the atmosphere is doubled and statistical thermal equilibrium is achieved, all models predict a global surface warming. (IGCMs) indicate global warming [from CO2 doubling] to be in the range between about 1.5 and 4.5deg. C" (National Research Council 1983, 28).

Second, though short-run weather forecasting has improved dramatically in recent years, model estimates of the impact of CO2 doubling are not converging. A recent survey of eighteen GCM simulations from seven different modeling groups found a median equilibrium increase from CO2 doubling of 4deg. C, with a range of estimates from 1.9deg. to 5.2deg. C. While all current models find a positive impact of elevated GHG concentrations on global temperature, the range is uncomfortably wide and is not narrowing.

Third, while the climate modelers have concentrated upon the equilibrium climate change, the more important question for economic policy concerns the rate of realized or actual warming. Because of the heat capacity of the oceans, the actual warming is likely to lag behind the equilibrium by several decades. To illustrate, note that estimates are that GHG concentrations have already increased by an amount that will in equilibrium produce about one-half of a CO2 doubling (say, 0.95deg. to 2.6deg. C, according to the just-cited range). Yet the actual warming over the last century is clearly far less than the estimated equilibrium amount (being, say, 0.3deg. to 0.6deg. C). Considerable scientific work is required to narrow down the time lags of the climate behind the GHG emissions.

General-circulation models produce a vast output of interesting numerical results on predicted future climates. Table 2.3 reports an attempt to characterize the results and uncertainties about future climate change. Most experts believe that mean temperature will rise and that the warmer climate will increase precipitation and runoff. Some models foresee hotter and drier climates in midcontinental regions, such as the U.S. Midwest. Forecasting climate changes at particular locations (such as in California or Cortina) has proven intractable, and many climate modelers do not expect to be able to forecast regional climates accurately in the foreseeable future.

Consistency with the Historical Temperature Record

One of the major problems with GCMs is the challenge of validating predictions. There are several possible ways to test the models, but the proof of the pudding will ultimately occur when and if global temperatures actually begin to rise.

Historical records indicate that global mean temperature has increased by 0.3deg. to 0.6deg. C since the 1880s. Whether the observed temperature record is consistent with the predictions of climate models is a hotly debated question. Some authors have used statistical techniques to test for the presence of a "greenhouse signal" in the upward trend of temperature over the last century. The hypothesis that the climate is a trendless process can be rejected at a high level of confidence. Still, a great deal of evidence suggests that climatic variables fluctuate over periods of a century or more.

Unfortunately, we do not know enough about the background

trends and cycles to know whether the warming in recent years is normal climatic fluctuation or something new and different.(4) To date, statistical analysis of the historical record has lagged far behind construction of new and more refined GCMs. But the historical record is an important, independent source of evidence about the pace of global warming, and the signal-to-noise ratio is likely to increase over the coming years, allowing a firmer test of the temperature-CO2 sensitivity.

Impacts of Climate Change

Although a great deal of effort has gone into constructing scenarios about climate change, studies of the impact of climate change on society are in their infancy. Even though we may not be able to predict the future climate change, it may, ironically, be possible to say a great deal about the economic impact of future climate changes. This section first describes the effects of greenhouse warming upon the economy, then presents some estimates of what measures to slow greenhouse warming would cost, and finally addresses the issue of potential adaptations to greenhouse warming.

Impacts of Greenhouse Warming

It will be useful to begin with some general remarks about the relationship of climate to human societies. To begin with, it must be recognized that human societies thrive in a wide variety of climatic zones. People today live in virtually every climatic zone from the tropics to the arctic, with the zone of tolerance ranging from - 60deg. to 55deg. C. Climate variables like temperature or humidity have little effect upon the net value of economic activity in advanced countries. Indeed, owing in part to technological changes like air conditioning, migration patterns in the United States have favored warmer regions. Today, for the bulk of economic activity, variables like wages, unionization, labor-force skills, and political factors swamp climatic considerations. When a manufacturing firm decides between investing in Hong Kong and Warsaw, climate will probably not even be on the list of factors considered.

At the same time, although most analyses focus primarily upon globally averaged surface temperature, this variable is not the most important for impacts. Variables like precipitation or water levels and extremes of droughts or freezes are likely to be more important. Mean temperature is chosen because it is a useful index of climate change that tends to be associated with most other important changes.

A related point involves the size of projected climate changes in comparison to the day-to-day changes we normally experience. The variations in weather that we experience in our daily lives will swamp the likely changes over the next century. The change in temperature while this paper is being read is likely to be greater than the expected change from 1990 to 2090. Few people are likely to notice the CO2 signal amidst the noisy pandemonium of their daily lives.

Economic Effects of Climate Change: United States

I next consider detailed studies of the impact of climate change on economic activity, beginning with studies of advanced countries and then turning to developing countries.

Climate change is likely to have different effects on different sectors. In general, sectors of the economy that have a significant interaction with unmanaged ecosystems--that is, those that are heavily dependent upon naturally occurring rainfall, runoff, or temperatures--may be significantly affected by climate change. Agriculture, forestry, and coastal activities fall in this category. Most of the U.S. economy has little direct interaction with climate, and the impacts of climate change are likely to be very small in these sectors. For example, cardiovascular surgery and microprocessor fabrication are undertaken in carefully controlled environments and are unlikely to be directly affected by climate change.

Table 2.4 presents a breakdown of U.S. national income, organized by the sensitivity of the sector to greenhouse warming.(6) Approximately 3 percent of U.S. national output originates in climate-sensitive sectors and another 10 percent in sectors modestly sensitive to climatic change. About 87 percent of national output comes from sectors that are negligibly affected by climate change. These measures of output may understate the impact of climate change on well-being because they omit important nonmarket activities or noneconomic value that may be more sensitive to climatic change than measured output, an important qualification that I will return to shortly.

What are the likely effects of climate change on individual sectors? The following is a synopsis of recent studies, and the quantified impacts are summarized in table 2.5.


Agriculture is the most climate-sensitive of the major sectors. Studies suggest that greenhouse warming will reduce yields in many crops. On the other hand, the associated fertilization effect of higher levels of CO2 will tend to raise yields, particularly in C3 species (which included most major crops except corn). After a careful review, a recent National Academy of Sciences report stated, "Thus, we do not regard the hypothesized CO2-induced climate changes as a major direct threat to American agriculture over the next few decades" (National Research Council 1983, 45). The Environmental Protection Agency found that the value of U.S. agricultural output is likely to rise or fall by as much as $10 billion annually depending on the magnitude of the climate change (EPA 1989a).

A recent study of Kane, Reilly, and Tobey (1990) goes beyond yield estimates to estimate the general-equilibrium impacts of climate change in a world agricultural model. Their estimates range from an optimistic finding that world real income would increase over a half century or more by 0.1 percent to a pessimistic scenario in which world output decreases by 0.3 percent. In the pessimistic scenario, the major loser is China, with a 5 percent decrease in GDP. To put these figures in perspective, measured per capita income in China grew at a rate of 5.2 percent over the period 1965-1987. Say the rate of growth was only half as high over the next half century. Then the pessimistic scenario of Tobey et al. suggests that climate change would lower the average growth rate from 2.6 to 2.5 percent per annum.

Sea-Level Rise

Most studies indicate a gradual rise in average sea level over the next century. The Intergovernmental Panel on Climate Change (IPCC) estimates that sea-level rise over the next eighty years will be 44 cm, although this figure has a large range of uncertainty. By comparison, over the past eighty years the sea level has risen 8 to 12 cm.

For a sea-level rise of 50 cm, EPA projects the costs to be land loss of around 6,000 square miles, protection costs (by levees and dikes) of high-value property, and miscellaneous protection of open coasts. The total capital outlay is on the order of $100 billion (EPA 1989a), which is approximately 0.1 percent of cumulative gross private domestic investment over the period 1985-2050.


Greenhouse warming will increase the demand for space cooling and decrease the demand for space heating. The net impact of CO2 doubling is estimated to be less than $1 billion at 1981 levels of national income.

Other Marketed Goods and Services

Many other sectors are likely to be affected, although numerical estimates of the effects are not available. The forest products industry may benefit from CO2 fertilization (Binkley 1988). Water systems (such as runoff in rivers or the length of ice-free periods) may be significantly affected, but the costs are likely to be determined more by the rate of climate change than the new equilibrium climate. Construction in temperate climates will be favorably affected because of a longer period of warm weather. The impact upon recreation and water transportation is mixed depending upon the initial climate. Cold regions may gain, and hot regions may lose; investments in waterskiing will appreciate, while those in snow skiing will depreciate. (7) But for the bulk of the economy--manufacturing, mining, utilities, finance, trade, and most service industries--it is difficult to find major direct impacts of the projected climate changes over the next fifty to seventy-five years.

Nonmarketed Goods and Services

Many valuable goods and services escape the net of the national income accounts and might affect the calculations. Among the areas of importance are human health, biological diversity, amenity values of everyday life and leisure, and environmental quality. Some people will place a high moral, aesthetic, or environmental value on preventing climate change, but I know of no serious estimates of what people are willing to pay to stop greenhouse warming. One study projects important gains for the United States from modest increases in average temperature (see National Research Council 1978). I am aware of no studies that point to major nonmarket costs, but further analysis will be required to decide whether these omitted sectors will affect the overall assessment of the cost of greenhouse warming.

In sum, the economic impact upon the U.S. economy of the climatic changes induced by a doubling of CO2 concentrations is likely to be small. The point estimate today is that the impact, in terms of variables that have been quantified, is likely to be around one-fourth of 1 percent of national income. However, current studies omit many potentially important effects, so this estimate has a large margin of error.

Economic Effects of Climate Change: Outside the United States(8)

To date, studies for other countries are fragmentary, and no general conclusions are possible at this time. Existing evidence suggests that other advanced industrial countries are likely to experience modest impacts similar to those of the United States. On average, high-income countries have less than 5 percent of their GDP originating in agriculture. Detailed studies for the Netherlands, as well as a less comprehensive study for six large regions (the United States, Europe, Brazil, China, Australia, and the USSR) found that the overall impact of a CO2-equivalent doubling will be small and probably difficult to detect over a half century or more (see Coolfont Workshop 1989).

On the other hand, small countries that are heavily dependent on coastal activities or suffer major climate change may be severely affected. Studies suggest that significant parts of Bangladesh and the Maldives may be inundated. Particular concerns arise where activities cannot easily migrate in response to climate change. Such situations include natural reserves (like Bharatpur) or populations limited to small areas (like South Sea Islanders).

Developing countries are probably more vulnerable to greenhouse warming than are advanced countries, particularly those poor countries living on the ragged edge of subsistence with few resources to divert to dealing with climate change. However, most poor countries are heavily dependent upon agriculture, so the benefits of CO2 fertilization might offset the damages from climate change. Countries classified by the World Bank as "low-income economies" had 31 percent of GDP produced in the agricultural sector in 1987; these countries hold 2.8 billion people. Figure 2.1 shows the share of output in agriculture for nine major countries in 1965 and 1987, with a mechanical projection of past trends to 2050. This illustration suggests that climate vulnerability of most countries has decreased over time and will further decrease as the climate changes.

These reflections lead to a surprising conclusion: Our best guess is that CO2-induced climate change will produce a combination of gains and losses with no strong presumption of substantial net economic damages. However, these changes are likely to take place over a period of a half century or more and may get lost in the background noise of social, economic, and political change. This conclusion should not be interpreted as a brief in favor of climate change. Rather, it suggests that those who paint a bleak picture of desert Earth devoid of fruitful economic activity may be exaggerating the injuries and neglecting the benefits of climate change.

Policies to Cope with the Threat of Global Warming

In response to the threat of global warming, a wide variety of responses is available. A first option, taking steps to slow or prevent greenhouse warming, has received the greatest public attention. Most policy discussion has focused on reducing energy consumption or switching to nonfossil fuels, while some have suggested reforestation to remove CO2 from the atmosphere. One important goal of policy should be cost-effectiveness--structuring policies to get the maximal reduction in harmful climatic change for a given level of expenditure.

A second option is to offset greenhouse warming through climatic engineering. Measures in this category include changing the albedo (reflectivity) of the earth, increasing the rate of removal of greenhouse gases, or changing water flows to cool the earth.

A final option is to adapt to the warmer climate. Adaptation could take place gradually on a decentralized basis through the automatic response of people and institutions or through markets as the climate warms and the oceans rise. In addition, governments could prevent harmful climatic impacts by land-use regulations or investments in research on living in a warmer climate.

Preventive Policies

The major policy question surrounding the greenhouse effect is whether steps should be taken in the near term to prevent global warming. Whether preventive action should be taken depends on the costs of preventing GHG emissions relative to the damages that the GHGs would cause if they continue unchecked.

Knowledge of the costs of slowing climate change is rudimentary. This section will review the costs of slowing greenhouse warming through reduction of emissions and atmospheric concentrations of greenhouse gases. These examples--reducing CFC emissions, reducing CO2 emissions, and reforestation--are not the only options, but they have been studied most intensively.

In calculating the cost of preventive measures, we measure costs in terms of tons of CO2 equivalent. Measures that cost up to $5 per ton of CO2 equivalent are inexpensive; at this cost, global warming could be stopped dead in its tracks at a total cost of less than $40 billion per year (about 0.2 percent of global income). Costs near $10 to $50 per ton CO2 equivalent are expensive but manageable (costing 0.5 to 2.5 percent of global income). Measures in excess of $100 per ton of CO2 are extremely expensive.

Reducing CFC Emissions

A first strategy involves reducing emissions of chlorofluorocarbons (CFCs) into the atmosphere. This step is particularly important because CFCs are extremely powerful greenhouse gases. It is currently believed that new chemical substitutes for the two most important CFCs can be found that will significantly reduce greenhouse warming. A rough estimate is that these substitutes can reduce warming at a cost less than $5 per ton of CO2 equivalent. This policy is extremely cost-effective, bringing a significant reduction of warming at modest cost.

Reducing CO2 Emissions

Any major reduction in GHGs will require a significant reduction in CO2 emissions, of which more than 95 percent come from the energy sector and deforestation. Carbon dioxide emissions can be reduced through increases in energy efficiency, decreases in final energy services, substitution of less GHG-intensive fossil fuels for more GHG-intensive fossil fuels, substitution of nonfossil fuels for fossil fuels, and technological change that allows new techniques of production along with new products and services.

Because energy interacts with the economy in so many ways, good estimates of the costs of reducing CO2 emissions require complex models of the energy system. Such models must incorporate the behavior of both producers and consumers along with consideration of each of the possible methods for reducing emissions. I have reviewed a number of studies, (9) and I will discuss the current estimates of the long-run costs of reducing CO2 emissions. (10) Figure 2.2 shows a compilation of the estimates of the marginal cost of reducing CO2 emissions as a function of the percentage reduction from a baseline projection .

(Note that in this figure as elsewhere in this paper, the estimates of the cost of reduction are relative to a trajectory in which there is no greenhouse policy. That is, when we discuss "the cost of a y percent reduction in greenhouse gas x," this calculates the cost of a y percent reduction from a baseline path in which no taxes, regulations, or other greenhouse policies are exercised. The baseline path is not a constant emissions or concentrations path, but a path in which emissions would be growing at the rate determined by fundamental factors such as GNP growth, energy prices, and technological factors.)

The studies of the long-run costs of CO2 reduction that I have reviewed lead to two major conclusions. First, the cost of reducing CO2 emissions is low for small curbs. Reductions of up to 10 percent of CO2 emissions from the energy sector can be attained at an average cost of around $10 per ton of CO2 reduced. With current global annual emissions of around 6 billion tons of CO2, a 10 percent reduction would cost around $6 billion annually.(11)

The second conclusion is that the cost of reducing CO2 emissions grows rapidly and becomes extreme for substantial reductions. I estimate that, in the long run but with today's energy technologies, the marginal cost of a 50 percent reduction in CO2 emissions is approximately $130 per ton of CO2. In other words, inducing producers and consumers to reduce their CO2 emissions by one-half would require a carbon tax (or the regulatory equivalent thereof) of $130 per ton of CO2, which would generate annual taxes of around $400 billion. The total resource cost of a 50 percent reduction in CO2 emissions is about $180 billion annually, slightly less than 1 percent of world output at current price and output levels.(12)

The incremental costs of reducing CO2 emissions rise rapidly because no substitutes currently exist for many uses of fossil fuels. For example, a major reduction in CO2 emissions from transportation would require either that people travel less or that fewer goods be transported, both of which would be quite costly.


Several studies have proposed using trees as a method of removing carbon from the atmosphere. Among the major proposals are slowing the deforestation of tropical forests; reforesting open land, thereby increasing the amount of carbon locked into the biosphere; a "tree bounty," which subsidizes the sequestration of wood in durable products; and a "tree pickling" program, in which trees are stored indefinitely.

No detailed study of the economics of tropical forests has been undertaken here. However, deforestation may be adding 500 million to 3 billion tons of carbon per year to CO2 emissions (amounting to between 5 and 30 percent of total GHG emissions). Much deforestation is uneconomic in tropical regions even without invoking greenhouse considerations. If so, the cessation of uneconomical deforestation can significantly and inexpensively slow greenhouse warming.

I estimate that the three other reforestation options can remove carbon from the atmosphere at modest costs; however, they can contribute only marginally to reducing atmospheric concentrations.

Summary on Costs of Prevention

As shown in Figure 2.3, it appears that a significant fraction of GHG emissions--perhaps one-sixth--can be eliminated at relatively low cost. The most cost-effective policies to slow greenhouse warming include curbs on CFC production and preventing uneconomic deforestation. Putting all the low-cost options together, I estimate that around one-sixth of C02-equivalent emissions can be reduced at an average cost of $4 per ton of CO2 equivalent, for a total cost of about $6 billion annually.

After the low-cost options have been exhausted, further reductions in GHG emissions will require curbing CO2 emissions--say, through taxes or regulations on the carbon content of fuels. But curbing GHG emissions rapidly hits diminishing returns: a 50 percent reduction in GHG emissions in the long run will cost about $200 billion annually, which is about 1 percent of global output. Attempts to restrict GHG emissions severely in a short period would be even more costly.

Figure 2.4 shows the trade-off between economic growth and slowing emissions. The upper path shows the trade-off if policies are efficient (say through a uniform global carbon tax) and introduced gradually over a period of time. The middle path shows the trade-off with an efficient policy and a rapid phase-in (phasing out existing capital over a twenty-year period). The bottom curve shows the trade-off with a rapid phase-in and inefficient policies (such as sector-by-sector regulations) that double the cost of controls. These curves show the importance of a gradual phase-in and efficient design of policies.

Climatic Engineering

In approaching a task, my grandfather always advised me, "Use brains, not brawn." The preventive strategies of planting hundreds of billions of trees or reducing fossil-fuel use through trillions of dollars of investments represent the brawn philosophy. A promising new approach to the threat of greenhouse warming is to use our brains to find a way to offset greenhouse warming through climatic engineering; this is the global equivalent of turning on an air conditioner. Potential approaches are changing the albedo (reflectivity) of the earth, increasing the rate of removal of greenhouse gases, or changing the circulation of water to cool the earth.

Careful analysis of these proposals is just beginning, but a number have already been identified that appear much more cost-effective than plugging the oil wells and shutting down the coal mines. One approach would be to create a sunscreen by sending tiny particulates into the stratosphere to cool the earth. These particles could be shot up with 16-inch naval rifles, lifted by hydrogen balloons, or deposited by tuning the engines of aircraft to burn somewhat richer than normal. One estimate finds that 100,000 kilograms of carbon can be offset with 1 kilogram of particles. That's real leverage!

A further intriguing possibility raised by Martin and his associates is placing trace quantities of iron in the North Pacific and Antarctic ocean regions. Studies suggest that this procedure would increase the limiting nutrients, foster much greater photosynthesis, and increase the rate of carbon precipitation to the ocean floor. Preliminary estimates indicate that this option might well annually remove from the atmosphere a quantity equal to current CO2 emissions.(13)

Preliminary estimates suggest that the cost of the geoengineering options is in the order of $0.10 to $10 per ton carbon equivalent, which is far less than many perennial favorites like reforestation, fuel switching, or energy conservation. These proposals surely sound like panaceas. At the same time, they pose unknown risks by perturbing a complex system that is already being perturbed. They deserve careful study.

Adaptation Policies

Faced with the prospect of changing climate, societies may decide to adapt. The most important adaptations are those taken by private agents--consumers and businesses, for example. Decentralized adaptations--population migration, relocation of capital, land reclamation, and technological change--will occur more or less automatically in response to changing relative incomes, prices, and environmental conditions.

Governments also play an important role by ensuring that the legal and economic structure is conducive to adaptation, particularly by making sure that the environmental or climatic changes get reliably translated into the price and income signals that will induce private adaptation. Fulfilling this role may prove difficult because so many of the impacts of climate change are not properly priced. For example, greenhouse warming may alter runoff patterns of major rivers (see Ravelle and Waggoner 1983; Waggoner 1990). Because water is allocated in such an archaic way, there is no guarantee that it will be efficiently allocated when water availability changes. Governments can improve adaptation by introducing general allocational devices (such as water auctions) that will channel water resources to their highest-value uses. Use of land near seacoasts and in floodplains poses similar issues.

Speeds of Adjustment in Prevention and Adaptation

Adaptation and prevention are often treated as symmetrical policies. They differ in one crucial respect, however: While preventive policies must generally precede global warming, adaptation policies can occur simultaneously. This distinction is crucial here, for cause precedes effect by a half century or more. To stabilize climate, immediate action is necessary; adaptations can wait for many years. This point of contrast underscores one of the major obstacles to responding intelligently to threatening climate change--the long time scale involved in climate change.(l4)

A common mistake in thinking about this issue is to impose a slowly changing climate upon today's world and to ignore the inevitable evolution that will take place over the coming decades. If it takes eighty years or more for CO2 doubling, as suggested earlier, adaptations will be spread over a similar period. Yet social and economic structures change enormously over such a time. Recall how much the world has changed since 1910. That was the age of empires, when the Ottoman, Austrian, and czarist regimes ruled much of Eurasia. The map of Europe has been redrawn three times since 1910 and is being restructured again today. The power density of the United States was about 1.5 horsepower per capita as opposed to 130 horsepower per capita today; one-sixth of horsepower was horses, and the 21 million horses were the major polluters. Air conditioning, nuclear power, and electronics were unheard of.

This catalog makes clear how foolish it would be to prescribe adaptive steps now to smooth the transition to climate changes over the next century. The time scale of most adaptations is much shorter than the time scale of climate change. Carbon dioxide doubling will take place over the next century. By contrast, financial markets adjust in minutes, product prices in weeks, labor markets in a few years, and the economic "long run" is usually reckoned at no more than two decades. To adapt now would be akin to building the Maginot Line in 1935 to cope with military threats of the 1990s, which would be little use for the petroleum wars of today.

These considerations suggest that it would be unwise to undertake costly adaptive policies unless they satisfy one of three criteria: (1) they have such long lead times that they must be undertaken now to be effective; (2) they have a clear presumption of being economical even in the absence of climate change; or (3) the penalty for delay is extremely high. By these criteria, it is difficult to enumerate any adaptive measures other than the general maxim to promote a healthy economy, to strive to internalize most external effects to ensure an appropriate response to changing climatic signals, to broaden the scope of markets so that individuals, firms, and nations receive the appropriate signals of scarcity, and to ensure a high saving rate to provide the investments needed for changing infrastructure.

Approaches to Policies on Global Warming

Research to date suggests that the costs of climate change are likely to be small for most advanced economies over the next fifty to seventy-five years. At the same time, the prospect that the climate may change in a catastrophic fashion might justify steps to slow climate change. How should nations respond today to the threat of global warming over the next century?

In related work, I have attempted to estimate the costs and benefits of policies to slow global warming (see especially Nordhaus 1990a). In studies of policy, it is useful to define government actions as carbon taxes that penalize emissions of greenhouse gases in proportion to their global warming potential. These "taxes" are a metaphor for explicit government steps to reduce GHG emissions through energy or gasoline taxes, CFC bans or regulatory limits, prohibitions on tree cutting, taxes on carbon emissions, or energy-efficiency standards.

Using the estimates of damages outlined previously and assuming a low discount rate on future damages from climate change, I calculate that an efficient policy would impose a penalty on GHG emissions of around $5 per ton CO2 equivalent (carbon weight).(15) This level of penalty would produce a total reduction of about 12 percent of GHG emissions, including a large reduction in CFCs and a small reduction in CO2 emissions. As table 2.6 shows, such a tax amounts to $3.50 per ton tax on coal, 58 cents per barrel on oil, and 1.4 cents per gallon on gasoline. Annually, U.S. revenues from a $5 per ton carbon tax would amount to about $10 billion. Table 2.6 also shows the impact of a severe restraint--$100 per ton CO2--which would be close to the tax required to reduce CO2 emissions by one-half. The "high-tax" strategy would have a significant impact upon the U.S. and other economies.

It will be useful to compare these costs with historical events or regulatory programs. A low-cost program for slowing global warming (say one associated with the low-tax proposal in table 2.6) would impose a burden equivalent to a major U.S. regulation, such as those on drinking water, noise, or surface mining (see Litan and Nordhaus 1983, ch. 2).

The more stringent program to cut GHG emissions by half (associated with the high-tax scenario in table 2.6) would impose annual costs (or, more technically, dead-weight losses) of around 1 percent of world output. This figure can be compared to the costs of all environmental, health, and safety regulations in the United States, which were estimated to cost 1 to 3 percent of GNP (ibid.). Another parallel is with the impact of the energy price increases of the 1970s. Jorgenson and Wilcoxen (1990) estimate that the energy-price increase lowered U.S. output growth by 0.2 percent per year, or a total of about 3 percent, since 1973.

Discounting, Nonlinearities, and Learning

Cost-benefit analyses are a useful starting point for considering government policies, but they raise several issues that must be addressed before making policy recommendations. To begin with, many values cannot be incorporated in a quantitative cost-benefit analysis. For example, climate change may threaten a society's cultural heritage in ways that are not possible to evaluate in an economic framework but which are nonetheless unacceptable. While being unable to put a price tag on Venice, we might decide that it is unacceptable to take actions that threaten Venice's existence. There is not much economic science can say about this issue except to identify such trade-offs.

In addition, greenhouse warming poses particularly difficult issues because of the importance of the discount rate and the presence of nonlinearities and learning over time. On these issues, economics has a great deal to say.

The Discount Rate and Future Climate Damages

How should the costs of future climate change be discounted in making current decisions? This issue is particularly thorny because of the long lags in the carbon cycle. Carbon dioxide emissions have an extremely long atmospheric residence time, in the range of 200 to 500 years, so actions today can affect economic welfare in the distant future. How should we balance CO2-reduction costs in 1990 against benefits in lower costs of climate change in 2040 or 2090?

In part, the issue of discounting is an ethical question, reflecting the relative valuation of well-being of current and future generations. But the revealed social discount rate is embedded in numerous public and private decisions, such as government fiscal and monetary policy and the rate of public investment, so the discount rate on climate change should not be chosen arbitrarily and without regard to other decisions. A real discount rate on goods and services close to the return on capital in most countries--say 8 percent per year or more--would imply that we should invest little today to slow the projected climate changes and concentrate instead on more immediate problems.

On the other hand, a low discount rate--say 4 percent per year or less--would give considerable weight today to climate changes in the late twenty-first century. But such a low discount rate would also imply that all investment opportunities with yields above 4 percent are exhausted--an assumption that is inconsistent with knowledge about rates of return on business and human capital in most advanced countries.

We might also ask whether a major commitment to slowing climate change is a worthwhile investment for developing countries who are likely to be the regions most vulnerable to climate change. Surveys of developing countries suggest that social rates of return well in excess of 10 percent per year are abundantly available in poor countries. For example, the social rate of return to investments in education in poor countries is estimated to be around 26 percent for primary education, 16 percent for secondary education, and 13 percent for higher education (see Psacharopoulos 1985). To devote many billions of dollars of resources to slowing climate change at the expense of equivalent investments in education, energy conservation, or tangible capital in developing countries would probably hurt poor countries and give little return in high-income countries.

A low discount rate on climate change along with a high return on capital is simply inconsistent. Faced with the dilemma of a low social discount rate and a high return on capital, the efficient policy would be to invest heavily in high-return capital now and then use the fruits of those investments to slow climate change in the future.


Clearly, global warming is rife with uncertainty--about future emissions paths, about the GHG-climate linkage, about the timing of climate change, about the impacts of climate upon flora and fauna, about the costs of slowing climate change, and even about the speed with which we can reduce the uncertainties. How should we proceed in the face of uncertainty?

One approach would be to take a "certainty equivalent" or "best guess" analysis, which would ignore uncertainty and the costs of decision making and charge ahead. The cost-benefit analysis performed above embodies this approach. It is appropriate as long as the risks are symmetrical and as long as the uncertainties are unlikely to be resolved in the foreseeable future. Unfortunately, neither of these conditions is likely to be satisfied for the greenhouse effect.

Virtually all observers agree that the uncertainties of climate change are asymmetrical; we are likely to be increasingly averse to climate change as the change becomes larger. To go from a 2deg. to a 4deg. warming is much more alarming than to move from a 0deg. to a 2deg. warming. The greater the warming, the further we move from our current climate and the greater the potential for unforeseen events. Moreover, it is the extreme events--droughts and hurricanes, heat waves and freezes, river flooding and lake freezing--that produce major economic losses. As probability distributions shift, the frequency of extreme events increases (or decreases) proportionately more than the change in the mean. Whether the increases in unpleasant extremes (like droughts in the corn belt) will be greater or less than the increases in pleasant extremes (like frost-free winters in the citrus belt) is, like most questions about climate change, unanswered.

In addition, climatologists generally think that the chance of unpleasant surprises rises as the magnitude and pace of climate change increases. We must go back 5 million to 15 million years to find a climate equivalent to what we are likely to produce over the next 100 years; the concentrations of GHGs in the next century will exceed levels previously observed.

Moreover, climate systems are complex systems, and some models have shown two or more locally stable equilibria (see Manabe and Stouffer 1988). There is historical evidence that climates have changed sharply (in or out of ice ages) in as little as a century, as occurred in the Younger Dryas period.

Among the kinds of responses that have been suggested and cannot be ruled out are major shifts of glaciers, leading to a rise in sea level of 20 feet or more in a few centuries; drastic changes in ocean currents, such as displacement of the North Atlantic deep currents that would lead to a major shift in climates of Atlantic coastal communities; large-scale desertification of the grain belts of the world; and the possibility that climate changes will upset the delicate balance of bugs, viruses, and humans as the tropical climates that are so hospitable to spawning and spreading new diseases move poleward. No one has demonstrated that these impacts will occur. Rather, it seems likely that unexpected and unwelcome phenomena, like the antarctic ozone hole, will occur more frequently under conditions of more rapid climatic change.

The threat of an unforeseen calamity argues for more aggressive action than a plain-vanilla cost-benefit analysis would suggest. However, the possibility of resolving uncertainties about climate change argues for postponing action until our knowledge is more secure. Most scientists believe that research can improve our understanding about the timing, extent, and impacts of climate change. Improved understanding could sharpen our calculations about appropriate policies. The best investment today may be in learning about climate change rather than in preventing it.

Putting this propostion concretely, we could easily make serious mistakes in attempting to prevent climate change. Imagine that the United States had mandated a massive nuclear-power program twenty years ago, only to find that the technology was expensive and unacceptable. Learning to cope with the threat of climate change includes not only improving our estimates of the consequences of climate change but performing R&D on inexpensive and reliable ways of slowing climate change.

A Framework for Policies to Slow Greenhouse Warming

I conclude by suggesting the direction that policy should follow in slowing greenhouse warming. In designing policies to slow global warming, we must first take into account that this is a global issue. Efficient policies will involve steps by all countries to restrict GHG emissions. In order to induce international cooperation, the United States and other rich nations may need to subsidize actions by poor nations (say to slow tropical deforestation or to phase out CFC use). While unilateral action may be better than nothing, concerted action is better still.

Given the identified costs of global warming, it would be sensible to take three modest steps to slow global warming while avoiding precipitous and ill-designed actions that may later be regretted.

1. Improve Knowledge. A first set of measures should aim to improve our understanding of greenhouse warming. Such steps would include augmented monitoring of the global environment; analyses of past climatic records, as well as intensive analysis of the environmental and economic impacts of climate change, past and future; and analyses of potential steps to slow climate change. Understanding of climate change has improved enormously over the last two decades, and further research will help to sharpen our pencils for the tough decisions to be made in the future.

2. Develop New Technologies. Countries should support research and development on new technologies that will slow climate change-- particularly on energy technologies that have low GHG emissions per unit of output. Too little is invested in these technologies because of a "double externality": Private returns are less than social returns both because the fruits of R&D are available to those who spent nothing on research and because the benefits of GHG reductions are currently worth nothing in the marketplace.

Energy technologies that replace fossil-fuel use require greater government support than they currently receive. Inherently safe nuclear power, solar energy, and especially energy conservation are particularly promising targets for government R&D support.

In addition, a number of technical fixes should be investigated to determine whether they might provide low-cost relief to future climate change. In particular, measures to sequester carbon and proposals for climate engineering should be carefully studied and field-tested to evaluate their merits. It is possible that these new approaches would be far more cost-effective than severe measures to curb fossil-fuel consumption.

3. "No-Regret" Policies. A third approach is to identify and accelerate the myriad otherwise-sensible measures that would tend to slow global warming. Many steps could contribute to slowing global warming at little or no economic cost. These steps include efforts to strengthen international agreements to restrict CFCs, moves to slow or curb uneconomic deforestation, and steps to slow the growth of uneconomic use of fossil fuels, say through higher taxes on gasoline, on hydrocarbons, or on all fossil fuels. If nations were to take such actions and climate change were to disappear, there would be few regrets about such policies.

Should we go further than these three steps? I believe not. The agenda of unsolved problems is long; resources and political will are scarce; and I believe it is premature to take costly steps to slow climate change at this time. But others might disagree and find the risks of climate change more frightening, or we might tomorrow uncover new information that would increase the likely future costs of climate change. In these cases, if we desire to press further in reducing longterm risks, I would turn to carbon taxes as a way of further reducing GHG emissions.

4. Carbon Taxes. A final measure to slow climate change would be a set of global environmental taxes levied on the C02-equivalent emissions of greenhouse gases, particularly on CO2 emissions from the combustion of fossil fuels. The analysis in this study suggests that a GHG tax in the order of $5 per ton of CO2 equivalent would be a reasonable response to the future costs of climate change.(l6)

The design of the carbon taxes goes beyond the scope of this paper, but a few remarks are in order. (l7) To be effective, carbon taxes should be imposed in all countries. Consider first the issue of fossil fuel combustion as the source of CO2. If the tax rate were harmonized among nations, then taxes on fossil fuel production would suffice as long as the production received no offsetting government subsidy. Since most carbon in fuels ends up in the atmosphere, no complicated chemical analysis is required.

For other sources of greenhouse gases, several complications arise. First, it would be necessary to convert the global warming potential of each gas into its CO2 equivalent. The translation will depend on the discount rate because different GHGs have different lifetimes. The translation is relatively straightforward for the CFCs, but the sources and chemical transformations for methane are extremely complicated. Second, some sources are not immediately emitted into the atmosphere, but gradually decay, as is the case for CFCs, which requires a complicated economic analysis of the shadow price of emissions at different periods. Third, there might be a "bounty" on sequestering activities, such as reforestation, which raises complicated problems of monitoring and keeping carbon inventories; tax specialists are generally wary of tax credits, and the possibility for abuse is significant.

Another set of issues concerns the international application of carbon taxes. The most significant question concerns whether they are production or consumption taxes, and who should receive the revenues. I believe that it is not politically practical to have much of the revenues accruing to those outside the nations that levy them. Moreover, production-based taxes are much simpler to administer, but the distribution of consumption-based taxes will be more appealing to industrial countries.

The design of consumption-based taxes is complicated by the need to calculate the carbon content of international trade flows on the basis of their C02-equivalent content. These adjustments might be significant for coal-based and petrochemical feedstocks, but to pursue carbon content far into the input-output table would mainly create employment for tax specialists and matrix inverters. With differential international standards or taxes, another set of issues concerns whether firms could employ "emissions offsets," whereby they would get credit in the high-tax regions for emissions reductions or carbon sequestration in other regions. Already, some utilities are assuaging environmental consciences bothered by the combustion of fossil fuels by planting trees in tropical regions.

Clearly, carbon taxation is a complicated issue of tax design; it would be useful to have a team of specialists design such a tax to see whether it is administrable. However, a carbon tax would be far preferable to regulatory interventions because taxes provide incentives to minimize the costs of attaining a given level of GHG reduction while regulations often do not. None of the dilemmas facing carbon taxation would be avoided by adopting emissions limitations or quotas; rather, the monetary flows and redistributions simply become murkier because the impact on prices and incomes of quantitative limitations is less visible.

While these arguments for a carbon tax are persuasive, I do not recommend it today. Negotiating a global carbon tax would be a daunting task even for a president who likes taxes and is not occupied in the Saudi desert with making the world safe for gas guzzlers. Reducing the risks of climate change is a worthwhile objective, but humanity faces many other risks and has many other worthy potential investments, such as oil conservation, factories, equipment, training, education, health, hospitals, transportation, communications, research, development, housing, environmental protection, population control, and curing drug dependency.

In conclusion, the United States and other major countries would be well served by continuing to take steps in the three areas outlined above--improving knowledge, investing in R&D in new technologies, and adopting "no-regret" policies that tilt away from greenhouse gases. Pursuing this approach, we will be prepared for whatever developments unfold in the future--for a tightening of the screws if the threat of global warming accelerates or for a relaxation of policy if science or technology were to alleviate our concerns. The struggle with the threat of greenhouse warming is likely to be a long one, and flexibility is the key to sensible policies.


1. This chapter draws on a number of earlier articles by the author, and helpful comments were provided along the way by Henry Aaron, Jesse Ausubel, Thomas Schelling, Charles Schultze, Paul Waggoner, and Gary Yohe.

2. A short history of scientific concerns about the greenhouse effect is provided in Ausubel 1983.

3. The traditional estimate of the relative importance of different greenhouse gases uses the instantaneous contribution of a gas to global warming (in deg.C). The traditional estimate has the defect that GHGs differ in their lifetimes and chemical transformations. In order to calculate the total contribution of each GHG, which is sometimes called "total warming potential," the table shows the sum of the instantaneous contributions over the indefinite future (in deg.C-years).

4. One eminent climatologist stated that he had "99 percent" confidence that the warming of the 1980s was associated with the greenhouse effect (see Hansen 1988). By contrast, four other respected scientists wrote that "no conclusion about the magnitude of the greenhouse effect in the next century can be drawn from the 0.5deg. C warming that has occurred in the last 100 years" (Marshall Institute 1989, 8).

5. The most careful studies of the impact of greenhouse warming have been conducted for the United States, and this review will therefore concentrate here. The most comprehensive is a recent study by the U.S. EPA (1989a). Although the studies reviewed here use different assumptions, we should envisage the estimated impacts as occurring late in the second half of the next century.

6. "National income" is total national output measured at factor costs. It equals GNP less indirect business taxes and depreciation.

7. A recent study of the impact of climate change in the Canadian Great Lakes region illustrates the substitution of different activities. According to the projected impact, global warming would shorten the length of the ski season by one-third, causing a $50 million loss in revenues. However, the camping season is expected to lengthen by forty days, offsetting the revenue losses .

8. I reiterate that the studies reviewed here represent "best-guess" scenarios of climate change. They omit uncertainties and possible nonlinearities, a topic that I shall examine in the last section.

9. These studies include estimates from specific technologies (such as CO2 scrubbing and substitution of methane for oil and coal); econometric or elasticity studies (often using highly aggregated models); and mathematical programming or optimization approaches (which often use activity-analysis specifications of energy technologies). For all three of these approaches, one can estimate the cost of reducing CO2 emissions as a function of the penalty or tax imposed upon those who emit CO2. See Nordhaus 1990c.

10. Note that these estimates are of the long-run cost--that is, the cost after the capital stock has fully adjusted. Attempts to reduce CO2 emissions in the short run would be much more expensive. Also, these costs do not include any adjustments for unmeasured or external environmental, health, or economic effects.

11. Indeed, some studies suggest that Cdeg.2 emissions can be reduced at "negative" costs. That is, there are opportunities to reduce GHGs whose costs are less than the benefits when these opportunities are evaluated at the appropriate social shadow prices. One set of examples concerns activities that emit greenhouse gases and have other environmental externalities. For example, use of CFCs is contributing to ozone depletion. Steps to protect the ozone layer by phasing out CFCs would have the additional benefit of slowing the buildup of greenhouse gases. Other areas are murkier from an economic point of view but might also qualify as negative-cost actions. For example, subsidies to energy prices in many countries lead to much higher levels of energy consumption than would be the case were prices to reflect realistic market prices or world prices. Market imperfections or informational deficiencies in capital markets, electric utilities, or capital-goods purchases are also alleged examples. The extent of negative-cost opportunities is controversial, but from a policy perspective the real issue may not concern the existence of negative-cost opportunities but the difficulty of finding tools to exploit them.

12. The chapter by John Whalley in this volume provides an estimate that is higher than this consensus estimate.

13. These estimates are derived from a recent unpublished study by Robert Frosch, to whom I am grateful for clarifying several issues.

14. Many of the issues in this section are developed at length in a superb essay by Schelling (1983).

15. The analysis draws upon Nordhaus 1990a. More precisely, it assumes that the discount rate on goods and services exceeds the growth rate of the economy by 1 percent per year. If the damage from a doubling of CO2 is 0.25 percent of total output, then the efficient CO2 tax is $3.2 per ton CO2 equivalent; if the damage is 1 percent of output, the efficient tax is $12.7 per ton. I have chosen $5 as an illustrative intermediate figure.

16. Some would argue that carbon taxes fall in category 3 as sensible economic policy. Consumption of fossil fuels has many negative spillovers beside the greenhouse effect, such as local pollution, traffic congestion, wear and tear on roads, accidents, and so forth. In addition to slowing global warming, carbon taxes would restrain consumption of fossil fuels, encourage R&D on nonfossil fuels, favor fuel switching to low-GHG fuels like methane, lower oil imports, reduce the trade and budget deficits, and raise the national saving rate. Indeed, in the tax kingdom, carbon taxes are the rara avis that increases rather than reduces economic efficiency.

17. The chapters in this volume by James Poterba and John Whalley go into the design of carbon taxes in greater detail.


Ausubel, Jesse H. 1983. Historical note. In National Research Council 1983, 488-491.

Binkley, Clark S. 1988. A case study of the effects of CO2-induced climatic warming on forest growth and the forest sector: B. Economic effects on the world's forest sector. In M. L. Parry, T. R. Carter, and N. T. Konijn, eds. The Impacts of Climatic Variations on Agriculture, 197-218. Dordrecht, Netherlands: Kluwer Academic Publishers.

Coolfont Workshop. 1989. Climate Impact Response Functions: Report of a Workshop Held at Coolfont, West Virginia, September 11-14, 1989. Washington, DC: National Climate Program Office.

EPA. 1989a. U.S. Environmental Protection Agency. The Potential Effects of Global Climate Change on the United States: Report to Congress, EPA-230-05-89050, December 1989.

EPA. 1989b. U.S. Environmental Protection Agency. Policy Options for Stabilizing Global Climate, Draft Report to Congress, February 1989.

Hansen, James. 1988. Testimony before the Senate Energy Committee, June 23, 1988.

Jorgenson, Dale W., and Wilcoxen, Peter J. 1990. The cost of controlling U.S. carbon dioxide emissions. Processed, Harvard University.

Kane, Sally, Reilly, John, and Tobey, James. 1990. An empirical study of the economic effects of climate change on U.S. agriculture. Processed, July 1990.

Litan, Robert E., and Nordhaus, William D. 1983. Reforming Federal Regulation. New Haven, CT: Yale University Press.

Manabe, S., and Stouffer, R. J. 1988. Two stable equilibria of a coupled ocean-atmosphere model. Journal of Climate 1:841-866.

Marshall Institute. 1989. Frederick Seitz, Karl Bendelsen, Robert Jastrow, and William A. Nierenberg, Scientific Perspectives on the Greenhouse Problem: Executive Summary. Washington, DC: George C. Marshall Institute.

Martin, J. H., Gordon, R. M., and Fitzwater, S. E. 1990. Iron in Antarctic waters. Nature 345:156-159.

National Research Council. 1978. International Perspectives on the Study of Climate and Society. Washington, DC: National Academy Press.

National Research Council. 1983. Changing Climate. Washington, DC: National Academy Press.

Nordhaus, William D. 1990a. To curb or not to curb: The economics of the greenhouse effect. Paper presented to the annual meetings of the American Association for the Advancement of Science, New Orleans, February.

Nordhaus, William D. 1990b. Contribution of different greenhouse gases to global warming: A new technique for measuring impact. Processed, February Nordhaus, William D. 1990c. A survey of the costs of reduction of greenhouse gases. Energy Journal, Fall.

Psacharopoulos, George. 1985. "Returns to Education: A Further International Update and Implications," Journal of Human Resources 20 (Fall):583-604.

Ravelle, Roger R., and Waggoner, Paul E. 1983. Effects of a carbon dioxide-induced climatic change on water supplies in the western United States. In National Research Council 1983, 419-432.

Schelling, Thomas C. 1983. Climatic change: Implications for welfare and policy. In National Research Council 1983, 449-482.

Wagonner, Paul E., ed. 1990. Climate Change and U.S. Water Resources. New York: Wiley.

Wuebbles, Donald J., and Jae Edmonds. 1988. A Primer on Greenhouse Gases, prepared for the Department of Energy, DOE/NBB-0083, March.