Joint CO2 and CH4 Accountability for Global Warming Kirk R

PNAS PLUS Joint CO2 and CH4 accountability for global warming Kirk R. Smitha,1,2, Manish A. Desaia,1, Jamesine V. Rogersa, and Richard A. Houghtonb

aEnvironmental Health Sciences, School of Public Health, University of California, Berkeley, CA 94720-7360; and bWoods Hole Research Center, Falmouth, MA 02540-1644

Contributed by Kirk R. Smith, May 18, 2013 (sent for review June 10, 2007)

We propose a transparent climate debt index incorporating both In this paper, we first briefly review the rationale behind the methane (CH4) and carbon dioxide (CO2) emissions. We develop use of climate debt metrics for determining “differentiated re- national historic emissions databases for both greenhouse gases sponsibility” and then offer an alternative to past approaches by to 2005, justifying 1950 as the starting point for global perspectives. presenting a global database of climate debt at the country level We include CO2 emissions from fossil sources [CO2(f)], as well as, in that integrates CO2(f) and CH4, the two GHGs with the most a separate analysis, land use change and forestry. We calculate the radiative forcing (RF) associated with human activity (10), into CO2(f) and CH4 remaining in the atmosphere in 2005 from 205 coun- a single combined climate debt metric. tries using the Intergovernmental Panel on Climate Change’sFourth We illustrate how a combined climate debt metric can bring Assessment Report impulse response functions. We use these calcu- GHGs with different atmospheric lifetimes together into a com- lations to estimate the fraction of remaining global emissions due to mon measure that is a function of the GHGs’ different depletion each country, which is applied to total radiative forcing in 2005 to functions and RFs, but without use of arbitrary time horizons or determine the combined climate debt from both greenhouse gases in discount rates. Difficulties in choosing among time horizons and units of milliwatts per square meter per country or microwatts per discount rates, for example, have plagued the calculation of global square meter per person, a metric we term international natural debt warming potentials, which are used to compare current emissions (IND). Australia becomes the most indebted large country per capita of different GHGs in a common metric of carbon dioxide equiv- because of high CH4 emissions, overtaking the United States, which is alents (11). To investigate the issues posed by excluding or in- highest for CO2(f). The differences between the INDs of developing cluding LUCF, we also calculate a combined climate debt metric and developed countries decline but remain large. We use IND to that consolidates both CO2(f) and LUCF as well as CH4 by region, SCIENCE assess the relative reduction in IND from choosing between CO2(f) fi as this is the nest geographic scale available for the LUCF dataset. SUSTAINABILITY and CH4`control measures and to contrast the imposed versus expe- We characterize the global landscape of historical account- rienced health impacts from climate change. Based on 2005 emissions, ability illuminated by the combined climate debt metric and the same hypothetical impact on world 2050 IND could be achieved by demonstrate two of its applications. First, we reveal how, in ad- decreasing CH4 emissions by 46% as stopping CO2 emissions entirely, dition to facilitating international comparisons, a combined cli- but with substantial differences among countries, implying differen- mate debt metric can help prioritize among mitigation options tial optimal strategies. Adding CH4 shifts the basic narrative about that target different GHGs. Second, we reanalyze global patterns differential international accountability for climate change. of health impacts from climate change to show how a combined climate debt metric alters relationships that have been examined differentiated responsibilities | sustainability metrics from the standpoint of CO2(f) alone in previous analyses. Although linked to the concept of “differential responsibility” he United Nations Framework Convention on Climate Change in UNFCCC, we use the term “accountability” here so as to T(UNFCCC), which was the basis for the Kyoto Protocol and the distance the concept from a moral judgment, which is a subject post-Kyoto negotiations initiated at the Copenhagen Conference for another venue. of Parties in December 2009, calls for allocating accountability for “ action on mitigation and adaptation based on common but dif- Significance ferentiated responsibilities” (1). Part of the reason that this concept has not been fully implemented is lack of an acceptable metric, We develop a transparent climate debt index, termed international often termed “climate debt,” that allows differentiated responsibility natural debt, which combines historical emissions of CO from to be transparently measured. The most frequently applied measure 2 fossil sources and land use/forestry as well as CH . It covers 205 of a country’s responsibility for global warming is its current annual 4 countries and is a function of emissions, lifetimes, and radiative emissions of greenhouse gases (GHGs)*. The second most common forcings. This index can be used to assess the implications of is cumulative emissions, simply the sum total of all past emissions. choosing between CO and CH control measures and facilitates Neither metric, however, fully reflects the causes of global warming, 2 4 more accurate international comparisons of a range of climate- because the amount of global warming occurring at any time is ac- change-related phenomena, as illustrated by imposed versus ex- tually due to the anthropogenic GHGs from past emissions still perienced health impacts. Including the two most important remaining in the atmosphere at a given time, a quantity that isusually greenhouse gases in one index shifts the basic international nar- intermediate between current and cumulative emissions. rative about differential accountability for climate change. Climate debt discussions have focused on carbon dioxide (CO ) emissions, the most important GHG. In particular, the 2 Author contributions: K.R.S. designed research; K.R.S., M.A.D., J.V.R., and R.A.H. performed emphasis has been on CO2 emissions consequent to fossil fuel research; R.A.H. contributed new analytic tools; K.R.S., M.A.D., and J.V.R. analyzed data; combustion and cement manufacture, which we term CO2(f), and K.R.S., M.A.D., and J.V.R. wrote the paper. referring to fossil carbon. Increasingly, net CO2 emissions from The authors declare no conflict of interest. land use change and forestry (LUCF) have also garnered atten- Freely available online through the PNAS open access option. tion. It is not just CO2, however, that is causing climate change, 1K.R.S. and M.A.D. contributed equally to this work. but a suite of human activities resulting in the emissions of GHGs 2 To whom correspondence should be addressed. E-mail: [email protected]. and aerosols, together termed climate-altering pollutants (CAPs). ’ This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. Indeed, CO2 s current impact on global warming is only about 1073/pnas.1308004110/-/DCSupplemental. one-half the total, depending on how calculated (8). There is *For an excellent and detailed discussion of the characteristics of different indices for relatively little in the published climate debt literature, however, determining equitable distributions of greenhouse gas emissions and their reductions, incorporating the impact of any GHG except CO2(f) (2, 9). see Grübler and Nakicenovic (2). Other discussions are found in refs. 3–7.

www.pnas.org/cgi/doi/10.1073/pnas.1308004110 PNAS | Published online July 11, 2013 | E2865–E2874 Downloaded by guest on September 27, 2021 Natural Debt generations acted out of ignorance in not controlling emissions, One reformulation of climate debt is “natural debt,” which better and thus their descendants should not be penalized (25). More- reflects physical reality than either current or cumulative emis- over, estimates of the emissions of GHGs become increasingly sions by focusing on the amount of a country’s or person’s past unreliable the farther back in time one goes, because they depend GHG emissions that remain in the atmosphere in any given year on records of fossil fuel use, cement production, leakage from (12, 13). A similar concept is “ecological debt” (14, 15). A na- pipelines, patterns of agriculture, etc. Even with good estimates tional debt is built by borrowing financial resources from the by geography, shifting political boundaries and past dominion of future. Similarly, the natural debt is built by borrowing assimi- one country over others could make it problematic to assign lative capacity of Earth’s natural systems from the future, as emissions to countries today, even for a GHG such as CO2(f) for GHGs are released faster than they can be removed by the which reasonable emissions data may be available (26). planet. Just as with their national debt, countries have built up The counterarguments address why it is unfair not to hold the their infrastructure and economic wealth faster than would present accountable for the past; that ignorant or not, we otherwise have occurred by borrowing against their natural debt. benefitted and ignorance should not be rewarded. The size of its natural debt indicates the degree to which a People alive today have directly benefited by the actions of country has avoided diverting resources from other economic their ancestors in borrowing environmental assimilative capacity. activities to GHG control during development. The current economic standard in the United States, for exam- The amount of a GHG remaining in the atmosphere today is ple, would most likely not be as high today if previous US gen- not equal to the total amount emitted throughout history be- erations had directed more resources to emitting fewer GHGs cause Earth’s assimilative capacity removes GHGs from the at- given the state of technology at the time. In the words of Bhaskar mosphere at a rate that varies with, inter alia, the physical and (27), “if current generations in the North accept assets from their chemical characteristics of each GHG. Thus, the most realistic parents, then it is incumbent upon them to also accept the cor- climate debt calculations allow for natural depletion over time, responding liabilities” (emphasis in original). counting only that portion that still survives as the current debt. Bhaskar (27) notes also that, because of past ignorance, it is An individual country’s contribution to the global natural debt not appropriate to place a moral opprobrium on past generations serves as a useful measure of its accountability for global warming and their descendants for these actions. It is fair, however, to that more accurately reflects physical reality than current yearly expect that current generations meet the obligations that come emissions, because the current global natural debt (excess re- with the benefits they receive. Simply put, it is a matter of remaining GHGs in the atmosphere) is what drives climate change. paying one’s debts with a fraction of the assets achieved in part This allocation approach also accords with the “polluter pays by taking on the debts. Similarly, the return from successful principle” from environmental ethics, policy, and law, which says investments made in the past for the wrong reasons, or even that those who release the pollution into the common environ- accidentally, still is subject to income taxes today. ment should be held accountable for the costs of the resulting In addition, if the present generation is to be expected to ac- negative impacts imposed on others and of remediation. cept accountability for the future, it must possess a feeling of The debut of a natural debt-like metric in international nego- control over the future. Without any control, there can be no tiations is attributed to the Brazilian Proposal, which recommends true accountability, because there is no reason to think the values use of “net anthropogenic emissions” from 1840 rather than and consequent sacrifices of today will be honored in the future. current emissions when calculating accountability for global Consequently, and perhaps paradoxically, to impart a perception warming (16). This approach was not taken up in the Kyoto of control over the future, the present generation must feel Protocol but remains an option discussed by the UNFCCC, by somewhat constrained by past. If this generation dismisses his- other countries [for example, China (17)], and in the scientific torical accountability, what is to keep the next generation from literature (18, 19). The approach was raised by various countries, doing so as well (28)? most recently by the BASIC coalition of Brazil, South Africa, One of the best ways to encourage this and future generations India, and China at the Doha COP-18 conference. to more seriously consider the long-term impacts of introducing Incorporating the impacts of additional GHGs into a climate new technology and other activities is to make it clear that they debt metric will help the metric to further reflect physical reality will be held accountable for problems that arise as a result of and send signals regarding the most appropriate control priorities. their decisions, no matter how much ignorance is claimed. With In addition, as additional GHGs are considered when allocating this shift in perspective, they will then be more likely to apply the historic accountability, the spectrum of mitigation strategies to appropriate caution in their choices. To not do so is to provide achieve GHG targets will enlarge. There are, however, conceptual great incentive to remain ignorant. issues with integrating additional GHGs into the climate debt metric both in combining GHGs with different RFs and lifetimes International Natural Debt and in that available historical databases for non-CO2(f) GHG To address the potential uneasiness associated with historical sources have been much less elaborated than for CO2(f). accountability and the practical difficulties in determining current GHG concentrations from emissions dating back to Review of Historical Accountabilities ‡ preindustrial times, we use records back to 1950 .Thistime “ ” Although the polluter pays principle may be conceptually at- frame has the advantage of starting after World War II when tractive, some observers have expressed discomfort when applied national boundaries were altered across the globe to mostly ’ historically to a country s GHG emissions back to the beginnings resemble their modern form and is after the major colonial of the Industrial Revolution (20). Two classes of arguments focus empires began to dissolve. A starting point after these two on why it is unfair to hold the present accountable for the past: † events makes calculations less subject to arbitrary assumptions the past was ignorant and the present is ignorant . that link old national boundaries to current ones. Although the first warnings about greenhouse warming ap- peared in the century before last (24), it can be argued that past

‡A base year of 1950 was also used by Sagar (7), but other investigators have gone back to 1915 (29), 1850 (30), or, in probably the ﬁrst analysis done, 1800 (31). Subak (32) † In the context of replying to the critiques of Beckerman and Pasek (21) and others, Neumayer compares ﬁve different approaches to assessing emissions with data going back to (22) succinctly summarizes six separate arguments for using some form of cumulative emis- 1860. The complexities of systematically determining emissions so long ago and assign- sions as a responsibility index, including those discussed here. The philosopher Miller (23), in ing them to current populations, however, are daunting. Shorter periods have also been contrast, offers arguments against both current and historical emissions metrics. proposed, e.g., 1990 (14).

E2866 | www.pnas.org/cgi/doi/10.1073/pnas.1308004110 Smith et al. Downloaded by guest on September 27, 2021 PNAS PLUS Another justification for choosing 1950 as the starting point is and its components INDCO2(f) and INDCH4; current, total orig- that it roughly coincides with the beginning of the modern in- inal, and total remaining emissions for CO2(f) and CH4; and ternational era as indicated, for example, by the creation of the population and income data. For simplicity, we use the term United Nations (1947), World Bank (1946), World Health Or- “countries” to refer to all 181 countries and 24 dependencies ganization (1946), United Nations Children’s Fund (UNICEF, included in the IND database. “Original emissions” refers to the 1946), and other global institutions, such as the major private amount of GHG emitted without accounting for any subsequent charities [Oxfam (1942) and Cooperative for Assistance and depletion, and “remaining emissions” refers to the amount of Relief Everywhere (CARE, 1945)], as well as the beginning of original emissions that remained in the atmosphere in 2005 given official bilateral development aid programs. By the early 1950s, depletion over time (see Methods for further details). most of these organizations had taken on their modern global Total INDCO2(f)+CH4, predictably, spans a vast range incor- character. Arguably, therefore, this baseline year represents the porating the extremes of population and economic size exhibited point in world history when a large fraction of the world first ac- by the world’s countries, with the United States’ 364 mW/m2 at − cepted some shared accountability for global action, and the one end and Niue’s 3.51 × 10 4 mW/m2 at the opposite. More emergence of many developing countries from colonization, as notably, per capita INDCO2(f)+CH4 varies dramatically, from the well as modern transport and communication. International com- Falkland Islands’ 5,045 μW/m2 perpersontoRwanda’s29μW/m2 mitments focusing on climate issues are now expressed through per person, a factor of nearly 175. these same 1950-era institutions via the Intergovernmental Panel The chart in Fig. 1A illustrates total world CO2(f) climate on Climate Change (IPCC) and UNFCCC. Because it focuses on debt, emphasizing the contribution from the 10 countries with accountability developed during the modern international era, the the largest INDCO2(f). LUCF is not attributed to individual full term for our suggested metric is “international natural debt” countries but is instead included as its own separate chart (D). (IND) (which we may further specify with subscripts to indicate the Given INDCO2(f)’s genesis from fossil fuel combustion, the largest precise GHGs involved). debtor countries reflect economic size. The INDCO2(f) from the Most previous CO2-based climate debt metrics have focused United States alone comprises 25% of total world INDCO2(f) and only on emissions from the energy and cement sectors, for which the next nine countries account for an additional 43%. Thus, the reliable databases have been widely available. We term these top 10 INDCO2(f) debtor countries collectively account for nearly CO2 emissions as CO2(f), for fossil carbon. In reality, human- three-quarters of total world INDCO2(f). induced net LUCF have also contributed significantly to CO The chart in Fig. 1B similarly illustrates total world CH climate

2 4 SCIENCE emissions over time. At the country level, it is difficult to attri- debt. The 10 countries with the largest INDCH4 broadly reflect bute LUCF empirically, because of incomplete and uncertain population size. Within Fig. 1, combining charts A and B yields SUSTAINABILITY historical records of LUCF sinks and sources, and conceptually, chart C, total world CO2(f) and CH4 debt or INDCO2(f)+CH4. due to the ambiguities of deciding which changes were natural Comparing the shift in rankings from charts A to C reveals that versus human-induced and what credit to assign for avoiding the contribution of developed countries often decreases, for ex- degradation of carbon stocks (33). Nevertheless, attempts have ample Japan’s from 4.8% to 2.9%, whereas the contribution of been made (34). Also, unlike CO2(f), there is no appropriate developing countries often increases, for example India’s from baseline starting point, i.e., the start of the Industrial Revolution 3.2% to 5.3%. (35, 36). LUCF’s estimated contribution is therefore reported Just as other metrics related to human welfare, such as income separately from CO2(f) in most of the tables and figures that and health, are best judged on a per capita basis, per capita IND follow. Given these and many other challenges inherent to at- indicates the average use of the assimilative capacity of the tributing climate debt from LUCF, a national-level assessment is planet by individuals within a country. Fig. 2 compares per capita beyond the scope of this paper. We do, however, consider the INDCO2(f)+CH4 for the 10 countries (minimum population, 10 implications of LUCF’s inclusion through a region-level version million) with the largest values and the 10 most populous de- of IND both with and without LUCF. veloping countries. The divergence in per capita INDCO2(f)+CH4 The most basic measure of warming by a GHG is RF, which is between these two sets of countries is striking. However, even measured in extra energy per surface area of Earth—watts per Brazil and Mexico are close to the world average per capita square meter above a natural baseline (8). We rely on the RF INDCO2(f)+CH4. In general, CH4 constitutes a higher fraction of values reported in the IPCC Fourth Assessment Report (AR4) climate debt in developing countries. For illustration, the mean for CH4 and CO2(f) in 2005 as the means to combine and percentage for the CH4 proportion among the 10 most populous compare the relative contributions by different countries. Thus, developing countries is 74%, compared with 28% for the 10 the IND enables (i) measurement of accountability in terms of countries with the largest values of per capita INDCO2(f)+CH4. the physical consequence of concern, which is increased warming Fig. 3 demonstrates the distinction between distributions of as measured by RF; (ii) comparison of INDs across two GHGs per capita INDCO2(f) and INDCH4 among countries by income, as with otherwise variable characteristics; and (iii) avoidance of the measured by per capita gross domestic product, adjusted for vexing issues of discount rates and time horizons. purchasing power parity (GDP-PPP). The analysis includes the In summary, we advocate a metric that, while capturing the 153 countries in the IND database with populations greater than causal drivers of climate change through RF, remains straight- one million. CO2(f) emissions are relatively closely associated forward and transparent, and allows readily for future updates with economic development (r2 = 0.55) and the ratio of highest fi and modi cations, including new emissions data, different RF to lowest per capita INDCO2(f) is more than 1,250 (Kuwait–Chad). values, and additional CAPs. Thus, IND draws on the concepts In comparison, CH4 emissions are far more evenly distributed of GHG depletion and RF to “measure” climate debt as opposed across the income spectrum, and as a result, per capita income is 2 to metrics relying on more opaque global climate models and a poorly correlated with per capita INDCH4 (r =0.085)andthe other methods that may in some sense be more accurate, but are ratio of highest to lowest per capita INDCH4 is only about 50 not transparent or able to be manipulated by nonspecialists (Trinidad and Tobago–Taiwan). quickly or without sophisticated tools. IND also relies on publicly Table 1 presents a preliminary exploration, at a region-level available and frequently updated emissions datasets, contribut- (defined in Supporting Information), of LUCF’s contribution to ing to the ease of future elaboration. IND, contrasting INDCO2(f)+CH4, which does not include LUCF, and IND , which does. In both the IND Results CO2(f)+LUCF+CH4 CO2(f)+CH4 and INDCO2(f)+LUCF+CH4 sections of the table, the United Supporting Information summarizes IND results for all 205 States, Europe, and China regions possess the largest total INDs, countries considered in our analysis, and includes INDCO2(f)+CH4 and the Oceania, United States, and Canada regions, the largest

Smith et al. PNAS | Published online July 11, 2013 | E2867 Downloaded by guest on September 27, 2021 Fig. 1. Top 10 debtor countries for INDCO2(f),INDCH4, and INDCO2(f)+CH4. Stacked bar charts represent total world (A)INDCO2(f),(B)INDCH4,(C)INDCO2(f)+CH4,and (D) LUCF climate debts. Chart sizes are proportional to RF. Within each chart, the 10 countries with the largest INDs are indicated top-most by individual segments in rank order. The thin arrows between charts track the change in countries’ positions from the chart in A to the chart in C or from the chart in B to the chart in C. The remainder of the world is represented by the segment “Other Countries.” The percentages immediately following a segment’s name indicate its

contribution to total global IND for that chart. LUCF climate debt is represented separately because it is not a component of INDCO2(f) or INDCO2(f)+CH4. Instead, the broad arrow shows how the chart in C would enlarge if LUCF were attributed to countries (or regions) and included in an INDCO2(f)+LUCF+CH4 metric as is the case in Table 1 [see “INDCO2(f)+LUCF+CH4 (IND with LUCF)” in table]. Data are from Dataset S1.

per capita INDs. As would be expected, assigning accountabili- uncertainty in emissions, we conducted Monte Carlo simulation ty for LUCF leads INDCO2(f)+LUCF+CH4 to be greater than analysesbasedonassumed90%conﬁdence intervals (CIs) for INDCO2(f)+CH4 for all regions (except the Caribbean). The CO2(f) and CH4 emissions of ±25% and ±100%, respectively, increases are proportionately much larger, however, in those conservative assumptions in line with expert opinion (9, 38–40). ﬁ tropical regions that experienced signi cant deforestation over Not surprisingly, the width of 90% CIs for INDCH4 exceeded 1950–2005 (Tropical South America, Central Africa, West Africa, central estimates of INDCO2(f) for the many countries with East Africa, and Southeast Asia), so much so that these regions substantial emissions from both sources. Similarly, the width leapfrog others in both total and per capita IND. However, intact of 90% CIs for INDs of large countries were frequently larger tropical forests in these same regions, as well as undisturbed than the central estimates of INDs for small countries. As would ecosystems in other regions, undoubtedly served as carbon sinks be expected, both INDs were disproportionately sensitive to both before and after 1950. Theoretical and methodological emissions from recent years because, for CO (f), emissions have risen obstacles continue to limit incorporation of this and other 2 dimensions of LUCF into a robust climate debt metric. Such very rapidly, and, for CH4, emissions decay comparatively quickly. In contrast to emissions, uncertainty in RFs essentially serves challenges serve to further motivate research into the carbon ’ cycle across spatial and temporal scales, as well as to highlight to renormalize all countries INDCO2(f) or INDCH4 to a different value. Clearly, if one GHG’s RF were to shift proportionately the utility of the more transparent INDCO2(f)+CH4. As with any composite metric, IND is subject to uncertainty more or less than the other, this would alter the relative contri- in its constituent parameters—emissions, lifetimes, and RFs— bution of the two INDs to INDCO2(f)+CH4 and thereby redistribute each of which contributes to a different step of calculating IND. countries. The effect of uncertainty in lifetimes lies, in a sense, Emissions underpin IND, and uncertainty in this parameter between emissions and RFs. Within INDCO2(f) or INDCH4,a category has the potential to reverberate across all countries change in lifetime would affect all countries, but late emitters (37). To explore the sensitivity of INDCO2(f) and INDCH4 to would be affected less than early emitters. Uncertainties in life-

E2868 | www.pnas.org/cgi/doi/10.1073/pnas.1308004110 Smith et al. Downloaded by guest on September 27, 2021 PNAS PLUS SCIENCE SUSTAINABILITY

Fig. 2. Per capita INDCO2(f)+CH4 for top debtor countries, world, and largest developing countries. The upper bars list the 10 countries, minimum population of 10 million, with the largest per capita INDCO2(f)+CH4. Together, these 10 countries comprise 11% of global population. The middle bar provides the world average for per capita INDCO2(f)+CH4. The lower bars list the 10 most populous developing countries, collectively comprising 55% of global population, with each country’s per capita INDCO2(f)+CH4. Each bar is divided into INDCO2(f) and INDCH4, the sum of which equals INDCO2(f)+CH4. To the right of each bar is the percentage of INDCO2(f)+CH4 from CH4. Data are from Dataset S1.

times and RFs, reported as 90% CIs in AR4 (8), are recapitulated rate, world INDCO2(f)+CH4 in 2050 would decrease by an amount in Supporting Information. equivalent to completely ceasing global CO2 emissions over the identical period. Although such a swift reduction, let alone an Discussion instant cessation, of emissions is not realistic for either GHG, the Use of IND to allocate accountability for mitigation and adapta- conclusion drawn from equating these two hypothetical scenarios tion could have several impacts on policy. The relative importance reflects how slowly the atmosphere responds to changes in of different CO2(f) and CH4 emission sectors for mitigation re- CO2(f) emissions compared with CH4, owing to their very dif- search and development would change. Effort could be put into ferent impulse response functions. It is worth emphasizing that finding ways for carbon trading and offset schemes to increasingly analysis of these scenarios does not account for dramatic rises in consider CH4. Less clear are the implications for negotiations re- future emissions, schemes to directly remove GHGs from the lated to “post-Kyoto” international climate change regime (41). On atmosphere, or impacts beyond 2050. Additionally, the analysis one hand, as a group, developing countries might be considered assumes that the impulse response functions and radiative effi- somewhat more accountable for current warming patterns than ciency of each GHG remains the same in 2050 as 2005. Despite currently perceived as a result of considering historical CH4 emis- these limitations and assumptions, the overarching conclusion sions in addition to historical CO2(f). Still, however, the differences corresponds with the results in Shindell et al. (44) of the global between developed and developing countries remain large. effects of CH4 versus CO2(f) control measures on temperature. On the other hand, a range of opportunities are revealed for An INDCO2(f)+CH4 in 2005 comprised of between 25-30% CH4 engaging developing countries to reduce warming quickly while at represents a zone of tipping points with regards to mitigation the same time protecting health and obtaining other cobenefits priorities, with the specific percentage a function of how a given ’ from the CH4 reductions, most of which would accrue in these country sCO2 and CH4 emissions have evolved over time. With same countries (42, 43). Unlike CO2(f) controls, reduction in CH4 respect to reducing INDCO2(f)+CH4 over the coming decades, for emissions would, in comparison, rapidly lower INDCO2(f)+CH4, countries with a CH4 fraction in 2005 below this zone, CO2(f) thus shifting the relative impacts of different countries. In schemes control might be considered a higher priority, whereas CH4 that reward progress in lowering climate debt, CH4 emitters would control might be more advantageous above this zone, and within then hold the advantage. the zone the two control options are broadly commensurate, although the relative costs, clearly, would also be important. This Opportunities for Control. For many countries possessing a signif- observation suggests an underappreciated rationale, as well as an icant degree of INDCO2(f)+CH4 in 2005 as CH4, the amount of opportunity, for emerging economies, many of which have INDs INDCO2(f)+CH4 that could be mitigated in the coming decades with a high percent CH4, to rapidly alleviate their climate debts may be greater from reducing CH4 rather than CO2(f) emissions. at potentially lower cost or disruption. Additionally, the same For example, consider the world as a whole, which has about reasoning offers further impetus to encourage focused and rapid 43% of its INDCO2(f)+CH4 in 2005 as CH4. If global CH4 emis- action on CO2(f) by rich countries, which often have INDs with sions from 2006 to 2050 continued at roughly 54% of the 2005 a low percent CH4. Of course, as control proceeds, the percen-

Smith et al. PNAS | Published online July 11, 2013 | E2869 Downloaded by guest on September 27, 2021 Table 1. IND by Region

A: INDCO2(f)+CH4 (IND Without LUCF) B: INDCO2(f)+LUCF+CH4 (IND With LUCF) Population GHG GHG Total IND Per Capita IND Fraction Total IND Per Capita IND Fraction Year 2005 Region* (# of μW/m2 % World μW/m2 % World 6 2 2 Countries) 10 people mW/m % World /person Average CO2 CH4 mW/m % World /person Average CO2 CH4

WORLD 6,498.893 1979.0 100.0% 304.5 100.0% 57% 43% 2416.0 100.0% 371.8 100.0% 65% 35% South Asia (7) 1,487.355 139.7 7.1% 93.9 30.8% 29% 71% 144.1 6.0% 96.9 26.1% 31% 69% China (4) 1,337.586 273.9 13.8% 204.7 67.2% 49% 51% 323.8 13.4% 242.1 65.1% 56% 44% S. E. Asia (12) 565.976 95.7 4.8% 169.0 55.5% 26% 74% 154.3 6.4% 272.6 73.3% 54% 46% Europe (33) 518.902 314.3 15.9% 605.7 198.9% 77% 23% 334.1 13.8% 643.8 173.2% 79% 21% M. E. & N. Af. (22) 436.970 110.2 5.6% 252.1 82.8% 51% 49% 122.5 5.1% 280.3 75.4% 56% 44% Trop. S. Am. (10) 313.169 94.5 4.8% 301.8 99.1% 25% 75% 205.0 8.5% 654.5 176.1% 65% 35% United States (1) 296.820 364.0 18.4% 1226.2 402.6% 77% 23% 394.8 16.3% 1329.9 357.7% 79% 21% F. S. U. (15) 284.920 252.9 12.8% 887.8 291.5% 62% 38% 274.6 11.4% 963.7 259.2% 65% 35% West Africa (17) 277.322 30.0 1.5% 108.2 35.5% 13% 87% 52.7 2.2% 190.1 51.1% 51% 49% East Africa (9) 229.289 23.2 1.2% 101.3 33.2% 4% 96% 37.6 1.6% 164.0 44.1% 41% 59% East Asia (4) 199.730 85.0 4.3% 425.4 139.7% 86% 14% 92.2 3.8% 461.5 124.2% 87% 13% S. Africa (14) 147.553 38.6 2.0% 261.6 85.9% 45% 55% 58.5 2.4% 396.7 106.7% 64% 36% Mesoamerica (7) 145.103 34.3 1.7% 236.3 77.6% 48% 52% 49.4 2.0% 340.2 91.5% 64% 36% Central Africa (9) 101.109 17.1 0.9% 169.3 55.6% 4% 96% 39.0 1.6% 385.9 103.8% 58% 42% Temp. S. Am. (4) 58.308 26.2 1.3% 449.9 147.7% 33% 67% 34.2 1.4% 586.7 157.8% 48% 52% Caribbean (16) 39.081 9.3 0.5% 238.5 78.3% 52% 48% 7.3 0.3% 187.9 50.6% 39% 61% Canada (2) 32.341 36.5 1.8% 1129.0 370.7% 67% 33% 48.2 2.0% 1489.5 400.7% 75% 25% Oceania (19) 27.359 33.8 1.7% 1235.0 405.5% 45% 55% 43.8 1.8% 1601.5 430.8% 57% 43%

*F. S. U., Former Soviet Union; M. E. & N. Af., Middle East & North Africa; S. Africa, Southern Africa; S. E. Asia, Southeast Asia; Temp. S. Am., Temperate South America; Trop. S. Am., Tropical South America. Regions deﬁned in Supporting Information.

tages will change and thus so will the relative importance of the 51). Although knowledge is growing rapidly, only one detailed two GHGs looking forward. global assessment of these effects has been published to date as part Major anthropogenic sources of CH4 include ruminant live- of the World Health Organization’s (WHO) Comparative Risk stock, flooded rice cultivation, waste management, and fossil fuel Assessment (CRA) Project (52). Estimates were made of disability- systems (45–47). Although changes in emissions from livestock adjusted life years (DALYs)§ in 2000 due to premature death and and rice cultivation may require changes in consumer food illness or injury by age, sex, and 14 world regions (defined in Sup- preferences, poor waste management and leaks from fossil fuel porting Information) as a result of anthropogenic climate change systems, which accounted for nearly one-half of global CH4 (54). In the target year of 2000, the overall impact of 150,000 emissions in 2005, can be addressed with little impact on con- premature deaths annually worldwide (0.4% of global lost DALYs) sumption patterns. Current inventory methodologies are not is relatively small by comparison with other global risk factors. It is comprehensive, so there are probably additional CH4 sources the future expression of ill health from climate change (avoidable that are not yet accounted for, such as abandoned landfills and risk), however, that is the main worry, rather than what has hap- oil/gas wells (48). Leaks from fracking operations may also add pened so far (attributable risk). to future CH inventories. Substantial reductions in the major 4 We use the WHO CRA results in our analysis, nevertheless, CO (f) sources, namely fossil fuels, however, require substantial 2 because it provides the only currently available consistent set of changes in energy systems, portions of which can be addressed in the near-term, whereas others will require long-term infrastruc- health effect estimates that allow comparison across regions and ture changes and investments. risk factors in an equivalent manner. In addition, it seems rea- As demonstrated above, the bulk of warming and thus sonable to expect that the future patterns of impacts would be similar across the world, given that most of the risk will likely be INDCO2(f)+CH4 is due to CO2(f), so CO2 emissions from fossil fuels therefore must be aggressively reduced to address climate exerted as exacerbation of local baseline health conditions. The downward trending line in Fig. 4 is taken directly from the change over the long run. Controlling CH4 emissions could be considered “low-hanging fruit” because it includes many actions CRA (54) and shows that the “experienced” health burden from that do not directly impact personal lifestyles and are amenable climate change declines with increasing economic development ¶ to regulation, although at some cost. Moreover, control of CH4 (per capita GDP-PPP) across the 14 CRA regions . This trend in and other shorter-lived CAPs, such as black carbon, offers a way experienced health burden is to be expected in that the poorest to achieve significant protection of the climate in the next few parts of the world are less able to protect themselves from en- years while society identifies and implements ways to achieve vironmental stresses in general and, partly as a result, experience the major reductions in CO2(f) that will be required (44, 49). much higher levels of ill health from them. The upper line shows the “imposed” health burden or the same total impact distributed Distributions of Impacts and Accountabilities: The Example of Health. according to the 2005 per capita INDCO2(f)+CH4 for each region, IND can also be used to compare the distribution of accountability with the impacts of climate change. A major concern about climate change, for example, is the potential impacts on human health, §The DALY is a measure of lost healthy life years, a more accurate measure of lost health through shifts in environmentally-mediated infectious disease pat- than deaths alone as it accounts for both the degree of prematurely in the deaths as well terns, extreme weather events, damage to agriculture, changes in as the severity and duration of nonfatal injuries and illnesses (53). ¶ water availability, heat stress, sea level rise, and other routes (50, For a CO2(f)-only analog, see ref. 55.

E2870 | www.pnas.org/cgi/doi/10.1073/pnas.1308004110 Smith et al. Downloaded by guest on September 27, 2021 PNAS PLUS SCIENCE SUSTAINABILITY Fig. 3. Per capita INDCO2(f) (red squares) and per capita INDCH4 (blue circles) by per capita income (GDP-PPP). Linear regression lines are dashed for INDCO2(f) 2 2 (r = 0.55) and dotted for INDCH4 (r = 0.085). Data are limited to countries with populations greater than one million (n = 153) and are from Dataset S1.

which trends in the reverse direction, i.e., richer regions impose change impacts resulting from modeling. It has the advantage also more risk than poorer ones because of their natural debts. that individuals and organizations can duplicate and update the The embedded table in Fig. 4 compares the ratio of imposed index themselves without waiting for new modeling results, which to experienced impacts across regions. The poorest regions im- may not be easily interpreted by nonexperts or differences pose ∼10% of the risk they experience, whereas the richest im- among models that cannot be quickly resolved even by experts. It pose ∼30,000% (300 times) the risk they experience, a difference is thus substantially more transparent and meets the criterion of of a factor of ∼2,000. This is the basis of the often published being a “good enough” tool (66). With the addition of the second maps of global health inequity from climate change (56). It is major GHG, CH4, the IND is even more reflective of the full noteworthy that all regions above a per capita GDP-PPP of about physical reality and provides signals related to control of shorter- $2,500 were imposing more health impact than they were experi- lived pollutants that would be lost in a CO2(f)-only index. encing early last decade. INDCO2(f)+CH4 reveals that the bulk of accountability remains with the more developed countries, although there is a shift Conclusion in the balance of accountability toward less developed coun- A number of analysts have examined the implications of alter- tries compared with a CO2(f)-only perspective because the CH4 native ways of framing emissions to design strategies that achieve proportion of a country’s climate debt is roughly an inverse equitable climate control across countries without considering function of income. In contrast, compared with wealthier econ- cumulative emissions (57–62). More sophisticated elaborations omies, CH4 reductions in poorer parts of the world can result in of accountability have also been proposed, such as those related comparatively rapid reductions in these countries’ climate debts, to actual cumulative RF (63) or based on weighted temperature as CH4 is often a significant portion of the debt, while having a changes (64). major impact on global warming within this century. Some of these latter approaches contain within them a re- Use of IND to allocate accountability in international nego- maining emission calculation, similar to natural debt, but are tiations for action on GHG mitigation and adaptation could elaborated in substantially different units and are not easily help address the “differentiated responsibilities” portion of the modified to include non-CO2(f) CAPs. They are also more dif- UNFCCC. By itself, however, it is not enough to determine ficult to associate in discrete ways with individual countries and payments or obligations. Under the “respective capabilities” clause thus to associate with other country characteristics such as in- of the UNFCCC, actual payments for remediation or adaptation in come, population, and human development indices. In addition, an international regime would require implicit or explicit assess- they tend not to be transitive, i.e., not able to be directly added, ment of ability to pay and additional factors including safeguards subtracted, etc. Although in a sense more reflective of reality against promoting population growth (5, 20). than natural debt, they often are further along the causal chain Although CO2(f) and CH4 are the GHGs that contribute the from emissions to effect and thus subject to substantially more most to RF, even more accurate signals might be given for uncertainty and require more assumptions due to the complex mitigation if a more full accounting were available by including and evolving models on which they rely (65). N2O, the F gases, and other CAPs such as black carbon. Even The IND therefore represents a compromise between the in its present form, however, the IND is a useful approach for oversimplification and somewhat misleading nature of current or attributing RF from human emissions of CAPs that provides cumulative emissions as the accountability metric and the com- a window into the way CH4 reductions can help slow warming plex, difficult to explain, and often-changing expressions of climate over the coming decades, thus allowing significant progress to

Smith et al. PNAS | Published online July 11, 2013 | E2871 Downloaded by guest on September 27, 2021 Fig. 4. Per capita GDP versus per capita imposed and per capita experienced health impacts from climate change. Health impacts from climate change compared with the accountability for climate change across 14 world regions (deﬁned in Supporting Information) ranked according to income (GDP, PPP adjusted). The regions and health metric (DALYs per capita) are as used in ref. 52. Climate change health impacts are from ref. 54. The lower line ﬁtted to the data represents experienced risk (r2 = 0.67), i.e., the health burden from climate change in each region. The upper lines represent imposed risk (r2 = 0.64), i.e., the total

health impact parsed by region according to INDCO2(f)+CH4. The imbedded table shows the ratio of imposed to experienced risk from climate change by income, which varies by 3 orders of magnitude. Note the logarithmic scale for both axes. Data are from Dataset S1 and McMichael et al. (54).

’ – be made while CO2(f) emissions are being addressed for the both with Annex I countries 1990 2005 data from the UNFCCC (69). Addi- longer term. tionally, population data were drawn from the United Nations Department of Social and Economic Affairs, Population Division (70), and economic data Methods from the World Bank (71) and The World Factbook (72).

Overview. IPCC’s AR4 (8) presented an estimate of the 2005 global atmo- The CDIAC CO2(f) dataset included emissions from fossil fuel combustion, spheric RF across all carbon-derived CAPs based on anthropogenic emissions cement manufacture, and gas ﬂaring in oil ﬁelds, corresponding to Common since 1750 (Table 2.13 and Figure 2.21). These RF estimates implicitly account Reporting Framework (CRF) categories 1A, 2A1, and 1B2C1, respectively (73). for the depletion of emissions over time by natural processes, as well as To construct complete and consistent time series of CO2(f) emissions, direct and indirect effects from each CAP. Thus, expressed as RF, the total adjustments to the CDIAC dataset were necessary to account for, most no-

global natural debts of the remaining CO2(f) and CH4, due to anthropogenic tably, changes in the boundaries of countries (see Supporting Information 2 emissions, are 1,123 and 856 mW/m ,INDCO2(f) and INDCH4 respectively, or for details on this and additional minor adjustments). 2 1,979 mW/m from both GHGs combined (INDCO2(f)+CH4,seeSupporting Over 1950–2005, 30% of countries in the IND database experienced Information for further details). a boundary change unaccounted for by the CDIAC dataset. For unifications, To attribute this IND among 205 countries and dependencies, collectively we merged the time series of the component countries. For partitions, we representing over 99% of the world’s population and economy, we devised used cumulative emissions during the first 5 y postpartition of each com- fi a ve-step procedure, executed in parallel for each GHG, to apply to each ponent country to proportionally weight the attribution of emissions pre- fl country in the analysis (see owchart in Supporting Information). For partition. Fourteen percent of all country-years were adjusted in one these “ ” simplicity, we use the term countries to refer to all 181 countries and two ways. 24 dependencies included in our IND database. Together, these 205 political The EDGAR CH dataset included emissions from energy, industry, agri- entities comprise ∼90% of those covered by the United Nations De- 4 culture, waste, and wildfires (forest and grassland), corresponding to CRF partment of Social and Economic Affairs, Statistics Division. In an ensuing categories 1, 2, 4, 6, and 5B, respectively (73). EDGAR’sCH4 dataset covers final step, we combined each country’sINDCO2(f) and INDCH4 to generate 1970–2005, accurately mapping these historical emissions, with no adjust- its INDCO2(f)+CH4. ment necessary, to those countries that existed in 2005. – Step 1. We developed a database of country-level time series (1950–2005) of To estimate CH4 emissions for 1950 1969, we extrapolated individual countries’ full 1970–2005 time series back to 1950 by least-squares linear original emissions of CO2(f) and CH4. “Original emissions” refers to the amount of GHG emitted by a country during a particular year without ac- regression. If an extrapolated trend became negative, emissions for that and counting for any subsequent depletion. The database was derived from all prior years reverted to zero. As we explain next, the relatively short datasets that, in addition to meeting our geographic and historical lifetime of CH4 emissions means that only a negligible fraction (1.2%) of the requirements, met the following criteria: (i) free and publicly available, (ii) global total original emissions from 1950 to 1969 remained in the atmo- frequently updated, (iii) having well-documented methodologies with esti- sphere by 2005. mates of uncertainty, and (iv) widely referenced. Step 2. Given these criteria, we drew on datasets for CO2(f) from the Carbon We calculated the amount of original emissions from each country for Dioxide Information Analysis Center (CDIAC) (67) and for CH4 from the each year that remained in the atmosphere in 2005, given depletion over Emission Database for Global Atmospheric Research (EDGAR) (68), updating time. The phrase “remaining emissions” refers to such amounts. We based our

E2872 | www.pnas.org/cgi/doi/10.1073/pnas.1308004110 Smith et al. Downloaded by guest on September 27, 2021 calculations on the impulse response functions or lifetimes, which model the the post-1950 period is therefore used as an estimate of the full distribution PNAS PLUS

decay of GHGs over time, for CO2 and CH4 as shown in the equations below: of remaining CO2(f) and CH4 in the atmosphere in 2005. With all IND measures expressed in the same units as RF, we summed each ð Þ¼ : ð−t=1:186 yearsÞ CO2 fraction remaining at time t years 0 186e country’sIND and IND to compute a combined IND or IND . [1] CO2(f) CH4 CO2(f)+CH4 þ0:338eð−t=18:51 yearsÞ þ 0:259eð−t=172:9 yearsÞ þ 0:217; LUCF. Our LUCF dataset was derived from the most recent update to Houghton ð−t=8:7 yearsÞ fl CH4 fraction remaining at time t ðyearsÞ¼e : [2] (74, 75) and estimates net CO2 uxes resulting from human-induced LUCF at a region-level (defined in Supporting Information). Once again, we focused Eq. 1 is the updated version of the impulse response function from the Bern only on the 1950–2005 portion of time series. The methods for calculating Carbon Cycle Model as recommended in AR4 (8). Eq. 2 uses the global CH4 the annual LUCF fluxes of CO2 are described elsewhere in greater detail (e.g., lifetime reported in AR4 (8). ref. 76). Here, we note that the analysis (i) draws on a vast array of land use To illustrate the use of these equations and the different decay dynamics statistics from agencies and researchers; (ii) models carbon fluxes for multiple of CO (f) and CH , consider 1,000 tons of each GHG emitted in 1990. The 2 4 native ecosystems per region; (iii) accounts for changes in living and dead amount of these original emissions still present in the atmosphere in 2005 carbon (above and below ground), harvested wood products, and soils; and would be 605 tons of CO (f) and 178 tons of CH . If these 1,000 tons of each 2 4 (iv) incorporates time lags for the decay of biomass and soil carbon and GHG had been emitted in 1950, then remaining in 2005 would be 423 tons regrowth of secondary forests following wood harvest and agricultural of CO2(f) and 2 tons of CH4. abandonment. Thus, the net flux of CO2 for a given region-year may be ei- As a result, although our database only extends back to 1950, for CH4 in particular there would be little difference (estimated <0.1% at the global ther positive (net CO2 source) or negative (net CO2 sink). The LUCF dataset fl level) in accounting for remaining emissions in 2005 were original emissions does not include uxes of CO2 from ecosystems undisturbed by human ac-

before 1950 included. A comparatively greater fraction of CO2(f) remaining tivity. Nor does it include the effects of environmental change (e.g., CO2 ﬂ emissions in 2005 were originally released before 1950, but this fraction is fertilization, climate change, nitrogen deposition) on CO2 uxes. nonetheless not vast, given the rapid rise in emissions during recent decades To harmonize the different spatial scales of our datasets (country-level ’ (67). Applying Eq. 1 to CDIAC s global time series of original emissions dating versus region-level), we added INDCH4 for all countries within a region to

back to 1751, we estimate that 87% of CO2(f) still circulating in the atmo- arrive at region-level INDCH4 values. However, for CO2(f) we followed a sphere during 2005 was emitted subsequent to 1950. slightly different process to accommodate negative values for LUCF. Within

each region, we added member countries’ yearly CO2(f) original emissions Steps 3–5 and Final Step. With our database elaborated to include remaining to create region-level time series. Next, we combined region-level CO2(f) emissions, we summed each country’s remaining emissions from each and LUCF time series, adding or subtracting as warranted, to generate a year to calculate country-level total remaining emissions (step 3). Next, SCIENCE uniﬁed CO2,orCO2(f)+LUCF, time series. With these CO2(f)+LUCF time-

’ SUSTAINABILITY we divided each country s total remaining emissions by the global total series of original emissions, we then followed the procedure outlined of remaining emissions (all countries, all years), yielding a “percentage of above in steps 1–5, to eventually generate region-level INDCO2(f)+LUCF world” for each country (step 4). We then multiplied these percentages values. Last, summing each region’sINDCO2(f)+LUCF and INDCH4 yielded by the RF for the corresponding GHG to compute country-level INDCO2(f) its INDCO2(f)+LUCF+CH4. or INDCH4 (step 5). We thus parse the total global natural debts of CO (f) and CH across 2 4 ACKNOWLEDGMENTS. We appreciate assistance with data from Evan countries believing that the practical and theoretical difﬁculties, in addition Haigler, Paul Lefebvre, Ray Liu, Gregg Marland, and Seth Shonkoff; detailed to the vexing complications discussed above (Natural Debt, Review of His- comments by Arnulf Grübler and Richard Norgaard; and insights from Tom torical Accountabilities, and International Natural Debt), of determining and Athanasiou, Paul Baer, Paul Ehrlich, Aslam Khalil, Hal Levin, Tony McMichael, assigning emissions previous to 1950 would outweigh any minor improve- Margaret Torn, and Uthara Srinivasan. We appreciate partial funding by the ment in nominal accuracy that might result. The distribution derived from California Air Resources Board (11-302).

1. United Nations (1992) Framework Convention on Climate Change (United Nations, 17. Project Team of the Development Research Center of the State Council (2009) New York). Greenhouse gas emissions reduction: A theoretical framework and global solution. 2. Grübler A, NakicenovicN(1994)International Burden Sharing in Greenhouse China’s New Place in a World in Crisis, eds Garnaut R, Song L, Wood WT (Australian Gas Reduction (International Institute for Applied Systems Analysis, Laxenburg, National Univ Press, Canberra, Australia), pp 389–408. Austria). 18. Friman M, Linner BO (2008) Technology obscuring equity: Historical responsibility in 3. Baer P, et al. (2000) Climate change: Equity and greenhouse gas responsibility. Science UNFCCC negotiations. Clim Policy 8(4):339–354. 289(5488):2287. 19. Höhne N, Blok K (2005) Calculating historical contributions to climate change— 4. Cazorla M, Toman M (2000) International equity and climate change policy. Climate discussing the “Brazilian Proposal.” Clim Change 71(1–2):141–173. Issues Brief (Resources for the Future, Washington, DC), No. 27. 20. Baer P (2006) Adaptation to Climate Change: Who Pays Whom? Fairness in 5. Hayes P, Smith KR, eds (1993) The global greenhouse regime: Who pays? Science, Adaptation to Climate Change, eds Adger N, Huq S, Paavola J, Mace MJ (MIT, Economics and North-South Politics in the Climate Change Convention (Earthscan Cambridge, MA). Publications/United Nations Univ Press, London). 21. Beckerman W, Pasek J (1995) The equitable international allocation of tradable 6. Ridgley MA (1996) Fair sharing of greenhouse gas burdens. Energy Policy 24(6):517–529. carbon emission permits. Glob Environ Change 5(5):405–413. 7. Sagar AD (2000) Wealth, responsibility, and equity: Exploring an allocation framework 22. Neumayer E (2000) In defence of historical accountability for greenhouse gas for global GHG emissions. Clim Change 45(3):511–527. emissions. Ecol Econ 33(2):185–192. 8. Solomon S, et al. eds (2007) Climate Change 2007: The Physical Science Basis. Contribution 23. Miller D (2009) Global justice and climate change: How should responsibilities be of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel distributed? The Tanner Lectures on Human Values (Tsinghua University, Beijing). on Climate Change (Cambridge Univ Press, New York). 24. Arrhenius S (1896) On the inﬂuence of carbonic acid in the air upon the temperature 9. Höhne N, et al. (2011) Contributions of individual countries’ emissions to climate of the ground. Philosophical Magazine and Journal of Science 41(251):237–276. change and their uncertainty. Clim Change 106(3):359–391. 25. Grubb M (1995) Seeking fair weather: Ethics and the international debate on climate 10. Shindell DT, et al. (2009) Improved attribution of climate forcing to emissions. Science change. Int Aff 71(3):463–496. 326(5953):716–718. 26. Grubb M, Sebenius J, Maglhaes A, Subak S (1992) Sharing the burden. Confronting 11. Shine KP (2009) The global warming potential—the need for an interdisciplinary retrial. Clim Change 96(4):467–472. Climate Change: Risks, Implications, and Responses, ed Mintzer IM (Cambridge Univ 12. Smith KR (1991) Allocating responsibility for global warming: The natural debt index. Press, Cambridge, UK). Ambio 20(2):95–96. 27. Bhaskar V (1995) Distributive justice and the control of global warming. The North, 13. Smith K (1996) The natural debt North and South. Climate Change: Developing the South and the Environment: Ecological Constraints and the Global Economy, eds Southern Hemisphere Perspectives, eds Giambelluca T, Henderson-Sellers A (Wiley, Bhaskar V, Glyn A (Earthscan, London), pp 102–117. New York), pp 423–448. 28. Smith KR (1977) The interaction of time and technology. PhD thesis (Univ of 14. Baer P, Athanasiou T, Kartha S (2007) The Right to Development in a Climate California, Berkeley). Constrained World: The Greenhouse Development Rights Framework (Heinrich Böll 29. Winkler H, Spalding-Fecher R, Tyani L (2002) Comparing developing countries under Foundation, Christian Aid, EcoEquity and Stockholm Environment Institute, Berlin). potential carbon allocation schemes. Clim Policy 2(4):303–318. 15. Srinivasan UT, et al. (2008) The debt of nations and the distribution of ecological 30. Rosa LP, Ribeiro SK (2001) The present, past, and future contributions to global

impacts from human activities. Proc Natl Acad Sci USA 105(5):1768–1773. warming of CO2 emissions from fuels. Clim Change 48(2):289–307. 16. United Nations Framework Convention on Climate Change (1997) Implementation of 31. Grübler A, Fujii Y (1991) Inter-generational and spatial equity issues of carbon the Berlin Mandate (United Nations, New York). accounts. Energy 16(11–12):1397–1416.

Smith et al. PNAS | Published online July 11, 2013 | E2873 Downloaded by guest on September 27, 2021 32. Subak S (1993) Assessing emissions: Five approaches compared. The Global Green- 54. McMichael AJ, et al. (2004) Comparative risk assessment: Climate change. house Regime: Who Pays? Science, Economics and North-South Politics in the Climate Comparative Quantification of Health Risks: Global and Regional Burden of Disease Change Convention, eds Hayes P, Smith KR (Earthscan Publications/United Nations Due to Selected Major Risk Factors, eds Ezzati M, Rodgers A, Lopez A, Murray C Univ Press, London), pp 51–69. (World Health Organization, Geneva), Vol 2, pp 1543–1649. 33. Corbera E, Schroeder H (2011) Governing and implementing REDD+. Environ Sci Policy 55. Smith KR, Ezzati M (2005) How environmental health risks change with development: – 14(2):89 99. The epidemiologic and environmental risk transitions revisited. Annu Rev Environ 34. Klein Goldewijk K, Beusen A, van Drecht G, de Vos M (2011) The HYDE 3.1 spatially Resour 30:291–333. explicit database of human-induced global land-use change over the past 12,000 56. Patz JA, Gibbs HK, Foley JA, Rogers JV, Smith KR (2007) Climate change and global – years. Glob Ecol Biogeogr 20(1):73 86. health: Quantifying a growing ethical crisis. EcoHealth 4(4):397–404. 35. Ruddiman RF (2006) Plows, Plagues, and Petroleum (Princeton Univ Press, Princeton). 57. Berk MM, den Elzen M (2001) Options for differentiation of future commitments in 36. Smith KR (1994) Pre-industrial missing carbon and current greenhouse responsibilities. climate policy: How to realise timely participation to meet stringent climate goals? Chemosphere 29(5):1135–1143. Clim Policy 1(4):465–480. 37. Marland G, Hamal K, Jonas M (2009) How uncertain are estimates of CO emissions? 2 58. Böhringer C, Welsch H (2006) Burden sharing in a greenhouse: Egalitarianism and J Ind Ecol 13(1):4–7. sovereignty reconciled. Appl Econ 38(9):981–996. 38. Andres RJ, et al. (2012) A synthesis of carbon dioxide emissions from fossil-fuel 59. Manne AS, Stephan G (2005) Global climate change and the equity–efficiency puzzle. combustion. Biogeosciences 9(5):1845–1871. Energy 30(14):2525–2536. 39. Olivier JGJ, Peters JAHW (2002) Uncertainties in global, regional and national emis- 60. Pan J (2003) Emissions rights and their transferability. Int Environ Agreement Polit sion inventories. Non-CO Greenhouse Gases: Scientific Understanding, Control Op- 2 – tions and Policy Aspects. Proceedings of the Third International Symposium, Law Econ 3:1 16. fi Maastricht, Netherlands, 21–23 January 2002, eds Van Ham J, Baede APM, Guicherit 61. Sugiyama T, Deshun L (2004) Must developing countries commit quanti ed targets? fl – R, Williams-Jacobse JFGM (Millpress Science Publishers, Rotterdam, The Netherlands), Time exibility and equity in climate change mitigation. Energy Policy 32(5):697 704. pp 525–540. 62. Toth FL (2001) Decision analysis for climate change: Development, equity and sustainability concerns. Int J Glob Environ Issues 1(2):223–240. 40. Marland G, Brenkert A, Olivier J (1999) CO2 from fossil fuel burning: A comparison of ORNL and EDGAR estimates of national emissions. Environ Sci Policy 2(3):265–273. 63. Labriet M, Loulou R (2003) Coupling climate damages and GHG abatement costs in 41. Moore FC, MacCracken MC (2009) Lifetime-leveraging: An approach to achieving a linear programming Framework. Environ Model Assess 8(3):261–274. international agreement and effective climate protection using mitigation of short- 64. Tanaka K, O’Neill BC, Rokityanskiy D, Obersteiner M, Tol RSJ (2009) Evaluating global lived greenhouse gases. Int J Clim Change Strategies Manage 1(1):42–62. warming potentials with historical temperature. Clim Change 96(4):443–466. 42. Smith KR, Balakrishnan K (2009) Mitigating Climate, Meeting MDGs, and Moderating 65. Smith KR, Ahuja DR (1990) Toward a greenhouse equivalence index—the total Chronic Disease: The Health Co-benefits Landscape. Commonwealth Health Ministers’ exposure analogy. Clim Change 17(1):1–7. Update 2009 (Commonwealth Secretariat, London), pp 59–65. 66. Socolow RH, Lam SH (2007) Good enough tools for global warming policy making. 43. Smith KR, et al. (2009) Public health benefits of strategies to reduce greenhouse-gas Philos Trans R Soc Lond A 365(1853):897–934.

emissions: Health implications of short-lived greenhouse pollutants. Lancet 374(9707): 67. Boden TA, Marland G, Andres RJ (2010) Global, regional, and national fossil-fuel CO2 2091–2103. emissions. Available at http://cdiac.esd.ornl.gov/trends/emis/tre_coun.htm. Accessed 44. Shindell D, et al. (2012) Simultaneously mitigating near-term climate change and April 15, 2011. improving human health and food security. Science 335(6065):183–189. 68. European Commission–Joint Research Centre/Netherlands Environmental Assessment 45. US Environmental Protection Agency (2006) Global Anthropogenic Non-CO2 Green- Agency (2010) Emission Database for Global Atmospheric Research, release version – fi house Gas Emissions: 1990 2020 (Of ce of Atmospheric Programs, Climate Change 4.1. Available at http://edgar.jrc.ec.europa.eu. Accessed April 15, 2011. Division, US Environmental Protection Agency, Washington, DC). 69. United Nations Framework Convention on Climate Change (2012) Greenhouse gas 46. Reay D, Smith P, van Amstel A, eds (2010) Methane and Climate Change (Earthscan, inventory data. Available at http://unfccc.int/ghg_data/items/3800.php. Accessed April London). 1, 2012. 47. Khalil MAK, ed (2000) Atmospheric Methane: Its Role in the Global Environment 70. United Nations Department of Economic and Social Affairs–Population Division (Springer, Berlin, DE). (2011) World population prospects, 2010 revision. Available at http://esa.un.org/unpd/ 48. Wunch D, Wennberg PO, Toon GC, Keppel-Aleks G, Yavin YG (2009) Emissions of wpp/Excel-Data/population.htm. Accessed June 5, 2011. greenhouse gases from a North American megacity. Geophys Res Lett 36(15):L15810. 71. The World Bank (2012) World development indicators: GDP, PPP (constant 2005 in- 49. Smith KR (2009) Methane first, OK? New Sci 202(2714):24–25. ternational $). Available at http://data.worldbank.org/. Accessed October 10, 2010. 50. McMichael AJ, Woodruff RE, Hales S (2006) Climate change and human health: 72. Central Intelligence Agency (2006) The World Factbook 2006. Available at www.all- Present and future risks. Lancet 367(9513):859–869. 51. Parry ML, Canziani OF, Palutikof JP, van der Linden PJ, Hanson CE, eds (2007) Climate countries.org/wfb2006/. Accessed October 10, 2012. Change 2007: Impacts, Adaptation and Vulnerability. Contribution of Working Group 73. Houghton J, et al., eds (1997) Revised 1996 IPCC Guidelines for National Greenhouse II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Gas Inventories (IPCC/OECD/IEA, Paris). fl – Change (IPCC) (Cambridge Univ Press, Cambridge, UK). 74. Houghton R (2008) Carbon ux to the atmosphere from land-use changes: 1850 52. Ezzati M, Rodgers A, Lopez A, Murray C, eds (2004) Comparative Quantification of 2005. TRENDS: A Compendium of Data on Global Change. Available at http://cdiac. Health Risks: Global and Regional Burden of Disease Due to Selected Major Risk ornl.gov/trends/landuse/houghton/houghton.html. Accessed March 31, 2012. Factors (World Health Organization, Geneva). 75. Houghton RA (2003) Revised estimates of the annual net flux of carbon to the 53. Murray CJL, Lopez AD, eds (1996) The Global Burden of Disease: A Comprehensive atmosphere from changes in land use and land management 1850–2000. Tellus B Assessment of Mortality and Disability from Diseases, Injuries, and Risk Factors in 1990 Chem Phys Meterol 55(2):378–390. and Projected to 2020 (Harvard School of Public Health on behalf of the World Health 76. Houghton RA (1999) The annual net flux of carbon to the atmosphere from changes Organization and the World Bank, Cambridge, MA), Vol 1. in land use 1850–1990. Tellus B Chem Phys Meterol 51(2):298–313.

E2874 | www.pnas.org/cgi/doi/10.1073/pnas.1308004110 Smith et al. Downloaded by guest on September 27, 2021