Science Progress | Where science, technology, and progressive policy meet

What We Don’t Know Can Hurt Us

Quantifying the Economic Risks of Climate Science Uncertainty

Map Of Climate Change Impacts SOURCE: Sandia National Laboratories A map developed by Sandia Labs shows the GDP risk due to climate uncertainty in 48 US states between 2010 and 2050, in billions of dollars at a 0% discount rate.

What we don’t know about climate science can really hurt us. So says a recent study conducted by researchers at Sandia National Laboratories.

One common argument used by those opposed to action on climate change is that there is simply too much uncertainty about the science to warrant any action. Even those policymakers supportive of efforts to address climate change feel hampered by an inability to make its risks palpably relevant to their constituents. Since any decision whose consequence plays out in the future contains some uncertainty — be it planning for retirement, evaluating a business venture, or fighting terrorism — there are many ways already known to contend with uncertainty. The challenge is to do so in a down-to-earth manner. This report uses everyday concepts already in use in the insurance industry to help policymakers better understand the risks associated with the uncertainty around climate science.

In response to discussions with DOE Secretary Chu, Congressmen Jeff Bingaman, Tom Udall, and Mark Heinrich during a visit to Sandia National Laboratories, our team worked to establish the near-term risk from climate change at the level of U.S. states for voters and businesses. The study finds that contrary to popular rhetoric, greater uncertainty about the impacts of climate change means greater economic risk, not less. Specifically, within an envelope covering 98% of the climate uncertainty as it pertains to rainfall alone, the U.S. economy is at risk of losing between $600 billion and $2.0 trillion and between 4 million and 13 million U.S. jobs over the next 40 years. Let’s examine how the study arrives at these estimates.

Insurance, uncertainty and risk

People expect that their house will be the same tonight as it was in the morning before going to work.  Nonetheless, they have insurance to protect them from a large array of risks. The risks are a combination of probability (uncertainty) and consequence. The insurance cost is dominated by events only relatively less common than the day-to-day expectations. These risks stem from the low-consequence events with intermediate probabilities, such as cracked windows or a leaking roof after a storm. The insurance premium is not strongly affected by the high-consequence events with low probability, such as the house being destroyed by a runaway bulldozer.  Nonetheless, all probabilities and consequences contribute to the calculation of the premium.  In all instances, the insurance company uses the best, albeit imperfect, understanding of those probabilities and their consequences.

There are many types of risks associated with climate change uncertainty, from increased disease to rising ocean levels.  The Sandia study looked at the uncertainty and consequence associated solely with climate-induced precipitation changes through the year 2050.  The equivalent premium to insure against adverse rainfall changes over that period for the United States would be over one trillion of today’s dollars. In more personal terms, the risk amounts to nearly 7 million lost full-time jobs. Moreover, it is certain that the quantified risk over this near term future is small compared to the exponentially growing risks from climate change in the years beyond 2050.

Risk derives and increases from “not knowing.”

In opposition to popular conjecture, the study demonstrates that the greater the uncertainty, the greater the risk.  This finding is consistent, for example, with the perspective that taking a commercial flight on Virgin Galactic’s SpaceShipTwo spacecraft is considered to have a relatively high risk because of a lack of information about its reliability.Risk derives and increases from “not knowing.”

The efforts of those skeptical of climate-change projections to demonstrate limitations in the accuracy of climate-change analyses may cause climate scientists to change the priorities of their research, but the real effect of emphasizing limitations is to accentuate the level of uncertainty in future climatic conditions. Rather than justifying a lack of response to climate change, the emphasis on the uncertainty enlarges the risk and reinforces the responsibility for pursuing successful long-term mitigation policy. If those skeptical of climate change want to halt government initiatives in climate policy, they must act to reduce the uncertainty and demonstrate that the future climatic conditions will remain below dangerous levels.

Economic impacts due to changes in precipitation

The Sandia study analyzed how consumers and individual industries adjust to the changing economic and physical conditions created by climate change. These responses attempt to lessen the economic impacts that would otherwise occur. The methodology underlying the analysis is based on historical response patterns of industries and consumers—how real people in business and as  individuals have behaved in the past to changing economic conditions, policies, and events. The study uses historical real-world behavioral experience because evidence indicates it is a more realistic approach than simulating the choices people make based on the commonly used economic assumptions of optimality and perfect knowledge of future conditions.

The study used precipitation precisely because it is one of the most uncertain outputs from the computer models used to estimate the climatic future. Further, the availability of water directly affects a large segment of economic activity and human wellbeing. The Sandia risk assessment considers economic losses exclusively in the absence of any climate change mitigation policies—in other words, we only consider what might happen if no action on climate is taken.

The analysis uses the concept of exceedance probabilities to describe the various levels of uncertainty. “Exceedance probability” is an intimidating-sounding term that has a very simple meaning.  It measures the likelihood (or chance) a particular consequence of climate change will exceed (be greater than) the value reported for that probability. For example, a 25% exceedance probability means there is an estimated 25% chance an impact will exceed the expected value (for example, in dollars of lost GDP).  The range of exceedance probabilities extends from 100% (the maximum realizable precipitation) to 0% (the minimum realizable precipitation). The body of the full report for this study (see Further Reading below) provides a detailed discussion of the analysis process and a thorough explanation of the results.

Figure 1 shows the estimated reduction in the U.S. GDP over the period 2010 to 2050 at various levels of exceedance probability. The values on the solid red line represent the total cost over the 40-year period.   Note how fast the losses accelerate at the lower probabilities. The dashed lines represent the uncertainty of the best-estimate exceedance-probability values. For any given point on the best-estimate line, it is highly likely that the impact will lie somewhere between the corresponding values on the enveloping dashed lines.

Figure 1. U.S. GDP impacts (2010–2050)

Figure 2 presents the summary-risk losses for the GDP for individual states. This information conveys the impacts of climate change with which state-level governments and business are likely to contend. The colors represent the relative nature of the impacts. In Figure 2, only six states, those colored green, experience apparent gains in the GDP as a result of climate change. The GDP losses exhibited by all the other states indicate what it would be worth to avoid climate change even within short-term planning horizons  — that is, if mitigation is possible. In Texas, for example, there is a risk of losing about $137 billion over the 40-year period. In lower exceedance probability conditions, only the three states of Washington, Oregon, and Idaho could realize “net” benefits.

Figure 2. GDP risk (2010–2050) in billions of dollars at a 0% discount rate.

Economic impacts due to water availability and migration effects

The study also considered the population migration across the states as unemployment and water-availability conditions evolved. The map of population migration would look similar to  the GDP map of Figure 2, expect that migration from the southern states to the northern states are more disproportionate than the GDP change indicates. Changes in employment generally follow changes in GDP, but due to variations in employment by industry, some states experience greater unemployment than the GDP impacts imply. The employment losses indicate the pressures to minimize the impacts of climate change that policy makers are likely to experience from voters.

A more detailed example may help in understanding the analysis displayed in Figure 2. Despite suffering relatively greater drought conditions on average relative to the rest of the nation, California shows improvements by 2050 because its economic impacts are estimated to become relatively less than those of other states. Populations from other affected states migrate to California and stimulate its economy. In the near term and at higher exceedance probabilities, California does incur largely negative impacts. Note that the impacts for many states change sign over time, that is, many states alternately experience gains (positive sign) and losses (negative sign).

The Pacific Northwest states show some improvement with climate change due to expected increased precipitation and population growth through migration. It is possible, however, that the damage to this region from climate change may be understated. Because the analysis is limited to the annual resolution of precipitation levels (other than capturing the monthly variation for agricultural assessments), it does not capture the impact of seasonal phenomena such as snow. In the Pacific Northwest, the dam system is not designed to accommodate significant changes in the timing of when and how fast snow melts. Consequently, the positive impacts shown could be an artifact of our simplifying assumptions.

Expected urban population growth and an expanding economy in the eastern United States will stress existing water supplies in the future even in the absence of climate change. Consequently, the Northeast and the Southeast experience negative impacts from climate change, even though reductions in long-term precipitation may be minimal. Areas such as Colorado go from having adequate water and benefits in high-exceedance-probability simulations to experiencing losses from reduced water availability in the low-exceedance-probability simulations. Other than in the Pacific Northwest, the uncertainty in climate change tends toward decreased water availability in the continental U.S.

Economic impacts felt by industry

The Study estimated the risks from climate-change uncertainty on 70 industry sectors. Companies generate the jobs and are also the actual entities that must produce the goods and services that allow the population to adapt to climate change.   At the national-revenue level, retail trade, mining, and food manufacturing each have risks in the range of $300 billion through the year 2050. Due to construction, especially of power plants to augment lost hydroelectric capacity, positive effects in terms of economic value are experienced by utilities, electric equipment, and other manufacturing. Construction itself experiences a decline because of the overall national decline in economic growth. Transportation experiences a net zero economic impact, despite an overall reduction in economic activity, because of the added need for interstate trade, especially for food. Many professional services, including medical, suffer a decline because unemployment constrains additional spending. Agriculture-dependent industries, such as the chemical industry, encounter disproportional declines. Like agriculture, climate change strongly affects the mining industry because of the mining industry’s relatively rigid dependence on water.

The study further addresses the dynamics of the impacts and the responses due to the volatility of climate change across the years. Figure 3 illustrates an example of the annual impacts on the national GDP as a function of varying exceedance probabilities for reduced water availability. As shown, greater losses are evident in succeeding years, and the lower exceedance probabilities are associated with greater impacts on the GDP. In this example of volatility (having approximately a 10% probability of occurrence), the trough of 50% exceedance-probability impact near 2030 exceeds the crest of 50% exceedance-probability impact near 2050. Volatility brings the impact of future “average” climate change into the present.  The recent Pakistan floods and Russian heat wave may be examples of the “here from the future” climate change.  The detailed, time-dependent approach used in the analysis shows the additional early consequences of the volatility in climate change.

Figure 3. Annual U.S. GDP impacts from climate change.


To reemphasize, the methods of this study reveal how compelling risk derives from uncertainty, not certainty. The greater the uncertainty, the greater the risk.  It is the uncertainty associated with climate change that substantiates the risks to the economy and society. Policymakers will most likely need to make decisions about climate policy before climate scientists have resolved all relevant uncertainties about the impacts of climate change.

From a policy perspective, the incentive to act comes by comparing the risk (cost) of inaction with the cost of action to successfully mitigate climate change. The study finds with 98 percent confidence that changes in rainfall patterns will cost the economy between $600 billion and $2 trillion over the next 40 years unless action is taken to prevent climate change.

Despite perpetual prospects for improvement, the current study does establish a process for superior and more-meaningful risk assessments of climate change than is currently presented. The impacts across the 70 industries and 48 states demonstrate interrelationships that produce consequences different from those consequences that would be indicated by the analysis of individual states or economic sectors in isolation.

The risk-informed approach used in this work relates physical climate science to the societal consequences and thus, the study offers a systematic foundation for policy debate. Uncertainty induces debate. In the presence of absolute certainty, there are no facts left to debate. This analysis used the current understanding of climate-change uncertainty to unambiguously quantify risk. The future evolution of climate change policy will necessarily rest on continuing improvements in the quantification of uncertainty for both climate change and its consequences.

George Backus works in the Discrete Mathematics & Complex Systems Department at Sandia National Laboratories and is the lead author of the Sandia climate-assessment study.  You can reach George at A version of this is cross-posted at Climate Progress.

Further Reading

Adler, R., “National Lab Calculates State by State Climate Change Risks, Jul 24, 2010,

Backus, G. et. al., (2010). Assessing the Near-Term Risk of Climate Uncertainty: Interdependencies among U.S. States. SAND 2010-2052. Albuquerque, NM: Sandia National Laboratories.

Backus, G. et. al. (2010), Executive Summary for Assessing the Near-Term Risk of Climate Uncertainty: Interdependencies Among the U.S. States, SAND 2010-2200. Albuquerque, NM: Sandia National Laboratories.

Comments on this article

By clicking and submitting a comment I acknowledge the Science Progress Privacy Policy and agree to the Science Progress Terms of Use. I understand that my comments are also being governed by Facebook's Terms of Use and Privacy Policy.