Elsevier

Global Environmental Change

Volume 24, January 2014, Pages 123-131
Global Environmental Change

Risk mitigation and the social cost of carbon

https://doi.org/10.1016/j.gloenvcha.2013.11.012Get rights and content

Highlights

  • This paper explores the social cost of carbon in a stochastic climate–economy model characterized by fat-tailed risks to future human welfare.

  • Accounting for uncertainty favors aggressive steps to stabilize climate given the degree of risk aversion people reveal on financial markets.

  • The social cost of carbon depends sensitively on the stringency of greenhouse gas controls, attaining the notably high value of $25,700 per metric ton of carbon dioxide in a scenario where greenhouse gas emissions are unregulated.

Abstract

The social cost of carbon – i.e., the marginal present-value cost imposed by greenhouse gas emissions – is determined by a complex interaction between factual assumptions, modeling methods, and value judgments. Among the most crucial factors is society's willingness to tolerate potentially catastrophic environmental risks. To explore this issue, the present analysis employs a stochastic climate–economy model that accounts for uncertainties in baseline economic growth, baseline emissions, greenhouse gas mitigation costs, carbon cycling, climate sensitivity, and climate change damages. In this model, preferences are specified to reflect the high degree of risk aversion revealed by private investment decisions, signaled by the large observed gap between the average rates of return paid by safe and risky financial instruments. In contrast, most climate–economy models assume much lower risk aversion. Given high risk aversion, the analysis finds that investment in climate stabilization yields especially large net benefits by forestalling low-probability threats to long-run human well-being. Accordingly, the social cost of carbon attains the markedly high value of $25,700 per metric ton of carbon dioxide in a baseline scenario in which emissions are unregulated. This value falls to just $4 per ton as the stringency of control measures is successively increased. These results cast doubt on the idea that the social cost of carbon takes on a uniquely defined, objective value that is independent of policy decisions. This does not, however, rule out the use of carbon prices to achieve the benefits of climate stabilization using least-cost mitigation measures.

Introduction

The social cost of carbon – defined as the marginal present-value cost imposed by greenhouse gas emissions – has emerged as a central concept in the economics of climate change (Tol, 2011). In 2002, for example, the United Kingdom adopted an official social cost of £19 per metric ton of carbon dioxide (or $29 per ton) for use in policy evaluation (see Department for Environment Food and Rural Affairs, 2002, Pearce, 2003). In the United States, carbon pricing is now required under the procedures of the Office of Management and Budget, which mandate the use of cost-benefit analysis to review all significant new and revised federal regulations, even when statutory requirements explicitly rule out a balancing of costs and benefits in the promulgation of environmental standards (Clinton, 1993, Hahn and Sunstein, 2002). Applications of this approach have established that accounting for net carbon emissions can have non-trivial impacts on the desirability of policy options (Kopp and Mignone, 2012). This is true, for example, of the U.S. Corporate Average Fuel Economy standards, where the level of fuel economy that is judged to be economically efficient is sensitive to the monetary value assigned to reductions in greenhouse gas emissions (Masur and Posner, 2011).

In 2009, the Obama Administration convened an Interagency Working Group with representation from the Environmental Protection Agency and five cabinet level departments (Agriculture, Commerce, Energy, Transportation, and Treasury) to survey the literature and assign a range of quantitative values to the social cost of carbon for use in official policy analysis. In the pursuit of this task, the Working Group employed three major models of the interplay between climate change and the global economy: Nordhaus’ (2008) “Dynamic Integrated Climate Economy” (DICE) model; Hope's (2008) “Policy Analysis of the Greenhouse Effect” (PAGE) model; and Anthoff and Tol's (2010) “Climate Framework for Uncertainty, Negotiation, and Distribution” (FUND) model (see also Tol, 1997). Although these models differ in various details, they adopt broadly similar assumptions regarding the costs of greenhouse gas emissions reductions, the economic impacts of climate change, and future trends in technology, population, and economic growth. (FUND, however, is relatively optimistic about climate impacts and adaptation, especially for small changes in mean global temperature.)

One point of difficulty for the Interagency Working Group was choosing the rate at which to discount future costs and benefits. Nordhaus and others have asserted that investments in greenhouse gas mitigation are warranted if and only if they provide returns at least as high as those available on financial markets, or approximately 6% per year (Nordhaus, 2008). In contrast, authors including Cline (1992) and Stern (2007) have argued that annual discount rates on the order of 1–2% are justified if one accepts the moral premise that equal weight should be attached to the welfare of present and future generations. Based on its review of the literature, the Working Group decided to consider discount rates of 2.5%, 3%, and 5% for all three models. With a discount rate of 5% per year, the Working Group concluded that the social cost of carbon attains a value in the year 2010 of $4.7 per metric ton of carbon dioxide, or just 4 cents per gallon of gasoline equivalent. With a 2.5% discount rate, the Working Group estimated a social cost of carbon of $35.1 per ton for the year 2010. Thus the use of low discount rates favors more aggressive steps to stabilize climate (Stern, 2007).

The reception of the Interagency Working Group report on the social cost of carbon has been mixed. On the one hand, it is clearly very important to assign an accounting price to changes in greenhouse gas emissions for use in regulatory impact analysis (Rose, 2010). An appropriately chosen carbon price can guide decision-makers to the adoption of behaviors and technologies that achieve society's environmental goals at the least economic cost. Pragmatically, the Working Group report provides a framework that federal agencies can utilize to pursue this objective.

On the other hand, critics have argued that the three models considered by the Interagency Working Group are based in part on optimistic assumptions concerning the projected economic impacts of climate change coupled with an incomplete analysis of risk (Ackerman et al., 2009, Pindyck, 2013, Stern, 2013). The DICE model, for example, assumes that a 3 °C increase in mean global temperature would lead to a 2.5% reduction in economic output. PAGE and FUND assume even lower damages. This contrasts with Hansen et al.’s (2008) warning that increases in mean global temperature exceeding 1–2 °C could potentially trigger positive feedback processes related to ice sheet collapses and the destabilization of global ecosystems that would impose truly catastrophic costs (see also Lenton et al., 2008).

To address uncertainty, the Interagency Working Group adopted Roe and Baker's (2007) fat-tailed distribution on climate sensitivity – i.e., the change in mean global temperature caused by a doubling of greenhouse gas concentrations – which implies a 20% chance of exceeding 5.0 °C. Monte Carlo simulations were then used to estimate a 95th percentile estimate for the social cost of carbon given a 3% discount rate. The resulting estimate of $64.9 per metric ton of carbon dioxide for the year 2010 is in one sense surprising. Ackerman and Stanton (2012), for example, found that assigning plausible values to uncertain parameters can result in a carbon price as high as $900 per ton of carbon dioxide. Anthoff et al. (2009) found that even higher values can arise in a sensitivity analysis involving low time preference and high risk aversion, though their central estimate was $16 per ton based on their interpretation of decision-makers’ revealed preferences in the absence of equity weighting. These points are linked to Neumayer's (2007) concern that the current generation of integrated assessment models does not fully account for the potentially “irreversible and non-substitutable damage” that climate change will inflict on the stability and functioning of ecosystems and the role of natural capital in supporting human activity.

In the present paper, we develop Kousky et al.’s (2011) argument that appropriately accounting for the role of risk mitigation might substantially alter the numerical value assigned to the social cost of carbon. Following Weitzman (2009), we work with a formal model of decision-making under uncertainty that allows for major risks of the type described by Hansen et al. and Roe and Baker. Using a theoretical model, Weitzman concluded that aggressive climate change policies might generate highly valuable (at face value potentially infinite) net benefits by reducing the statistically low probability that unmitigated climate change would lead to future economic collapse. In previous work (Gerst et al., 2013), we confirmed this finding in a plausibly specified numerical model in which preferences regarding time and risk were inferred from market data on consumption growth and the rates of return paid by safe and risky investments using methods from the macrofinance literature (Lucas, 1978, Mehra and Prescott, 1985, Barro, 2006). Iverson and Perrings (2011) provide a related analysis based on an application of the asymmetric minimax regret criterion as a framework for characterizing rational decisions under strong uncertainty. In a similar vein, McInerney et al. (2012) describe how an array of decision-theoretic approaches can be applied to evaluate climate change policies in a modified version of DICE.

Here, we employ the Gerst et al. model to produce a seemingly paradoxical result. On the one hand, we find that deep cuts in greenhouse gas emissions can produce very high net social benefits. On the other hand, once an aggressive control path is initiated, the marginal benefit of further emissions reductions is quite low. We see this result as consistent with the well-known “diamond–water” paradox, in which actions that are essential to sustaining human welfare have high total net benefits yet low marginal benefits given appropriate levels of provisioning (see Farber et al., 2002). In our model, this occurs when emissions cuts are sufficient to reduce the relatively low probability of catastrophic climate impacts to essentially zero.

These results contrast strongly with the Interagency Working Group's (2010) finding that the social cost of carbon is relatively independent of the stringency of emissions abatement. Such independence can occur in a deterministic model like DICE (Nordhaus, 2008), in which equilibrium temperature is logarithmic with respect to greenhouse gas concentrations and climate change damages are quadratic with respect to temperature. This implies a nearly linear relationship between temperature and damages and, hence, a marginal cost of greenhouse gas emissions that is independent of the state of the environment (Hope, 2006). This, however, appears to be an idiosyncratic condition rather than a general phenomenon, especially when the complex dynamics of risk mitigation are considered.

Our analysis suggests that risk-averse decision-makers attach especially high value to the early elimination of catastrophic risks and that the level of total net benefits provided becomes nearly invariant to specific policy scenarios under emissions control rates of 40% or more by the year 2050. This implies that the main risks to welfare are from failing to stabilize climate, not from cutting emissions by too much, too soon. Thus, balancing the marginal costs and benefits of emissions controls may be less important than attaining the overall benefits of climate stabilization. Pragmatically, this may favor an approach to carbon pricing aimed at the cost-effective achievement of policy-specified emissions targets, rather than a focus on the ‘correct’ social cost of carbon, especially given the uncertainties associated with integrated assessment models and the strong role of moral values in climate governance (see Howarth, 2011, Dietz, 2012).

Section snippets

The model

For the purposes of analysis, we work with Gerst et al.’s (2013) stochastic integrated assessment model of climate–economy interactions, which adopts the mitigation cost, damage cost, carbon cycle, and climate system modules of DICE (Nordhaus, 2008). To allow for tractable analysis of low probability risks, Gerst et al. base the economic module on the well-known Lucas–Mehra–Prescott model (Lucas, 1978, Mehra and Prescott, 1985), which plays a central role in understanding the coupled dynamics

Policy scenarios and welfare implications

The model summarized above defines an equilibrium path for the economy once policy-makers stipulate a set of greenhouse gas emissions abatement policies. As in Gerst et al. (2013), we focus on a family of alternative policy scenarios in which the greenhouse gas emissions control rate (μt) rises from a value of zero in 2010 to unity in the year 2270 following an S-shaped Weibull function (Fig. 2). The speed of the transition is then defined by the stringency of emissions reductions in the year

The social cost of carbon

The final step in our analysis is to calculate the social cost of carbon as a function of the stringency of emissions control policies and decision-makers’ risk aversion. As noted in the introduction, it is well-known that, all else equal, the use of low discount rates leads to a relatively high value for the social cost of carbon, which reflects the marginal social cost imposed by short-run greenhouse gas emissions (see Johnson and Hope, 2012). In addition, authors such as Hope (2006) have

Conclusions

This paper has assessed the social cost of carbon in a stochastic growth model adapted to account for the costs of greenhouse gas emissions reductions, the relationship between emissions and future mean global temperature, and the economic impacts of climate change. The model is novel because it provides an internally consistent approach to analyzing the net benefits of climate stabilization under strong uncertainty. It allows for fat-tailed uncertainty concerning climate sensitivity, and its

Acknowledgements

We thank the referees and the editors for their helpful and engaging comments on the initial version of this paper. We also thank the Yale School of Forestry and Environmental Studies for hosting a seminar on this work.

References (56)

  • K.J. Arrow et al.

    Uncertainty and the evaluation of public investment decisions

    American Economic Review

    (1970)
  • G. Atkinson et al.

    Siblings, not triplets: social preferences for risk, inequality and time in discounting climate change

    Economics E-Journal

    (2009)
  • M.B. Baker et al.

    The shape of things to come: Why is climate change so predictable?

    Journal of Climate

    (2009)
  • R.J. Barro

    Rare disasters and asset markets in the twentieth century

    Quarterly Journal of Economics

    (2006)
  • R.B. Barsky et al.

    Preference parameters and behavioral heterogeneity: an experimental approach in the health and retirement study

    Quarterly Journal of Economics

    (1997)
  • R.E. Bellman

    Dynamic Programming

    (1957)
  • J. Broome

    The ethics of climate change

    Scientific American

    (2008)
  • W.R. Cline

    The Economics of Global Warming

    (1992)
  • W.J. Clinton

    Executive Order 12866 of September 30, 1993 – Regulatory Planning and Review

    (1993)
  • Valuing the Social Cost of Carbon Emissions: DEFRA Guidance

    (2002)
  • S. Dietz

    The treatment of risk and uncertainty in the US social cost of carbon for regulatory impact analysis

    Economics E-Journal

    (2012)
  • P. Ding et al.

    Rare disasters and risk attitudes: international differences and implications for integrated assessment modeling

    Risk Analysis

    (2012)
  • Hahn, R.W., Sunstein, C.R., 2002. A new executive order for improving federal regulation? Deeper and wider cost-benefit...
  • R.E. Hall

    Intertemporal substitution in consumption

    Journal of Political Economy

    (1988)
  • J.E. Hansen et al.

    Target atmospheric CO2: where should humanity aim?

    Open Atmospheric Science Journal

    (2008)
  • C. Hope

    The social cost of carbon: what does it actually depend on?

    Climate Policy

    (2006)
  • C. Hope

    Optimal carbon emissions and the social cost of carbon under uncertainty

    Integrated Assessment Journal

    (2008)
  • R.B. Howarth

    Discounting and uncertainty in climate change policy analysis

    Land Economics

    (2003)
  • Cited by (25)

    • Do organic fertilizer impact on yield and efficiency of rice farms? Empirical evidence from Bangladesh

      2021, Heliyon
      Citation Excerpt :

      On the other hand, most Bangladeshi rice farmers do not follow the recommendation guide of fertilizer due to a lack of awareness about soil fertility and appropriate fertilizer use knowledge. As a result, long-term use of chemical fertilizer can negatively affect crop productivity and soil health (Howarth et al., 2014). Moreover, the rising production cost and shortage of chemical fertilizer, an alternative to chemical fertilizer is needed.

    • What explains carbon-pricing variation between countries?

      2020, Energy Policy
      Citation Excerpt :

      Global climate change is widely perceived to be the primary environmental challenge for the 21st century. The environmental and societal damage caused by carbon dioxide emissions is reflected in high, but widely varying, social cost of carbon calculations (Hepburn, 2017; Howarth et al., 2014; Pindyck, 2019). Carbon prices have been increasingly implemented in recent years to combat climate change risk but only covered around 15% of global emissions by the end of 2019 (Hepburn, 2017; World Bank, 2019a).

    View all citing articles on Scopus
    1

    Tel.: +1 704 516 5067; fax: +1 603 646 2277.

    2

    Tel.: +1 603 646 9944; fax: +1 603 646 2277.

    View full text