International Environmental Agreements: Politics, Law and Economics

, Volume 12, Issue 4, pp 375–389

Geoengineering and global warming: a strategic perspective


    • Department of Political ScienceColumbia University
Original Paper

DOI: 10.1007/s10784-012-9167-0

Cite this article as:
Urpelainen, J. Int Environ Agreements (2012) 12: 375. doi:10.1007/s10784-012-9167-0


If major emitters fail to mitigate global warming, they may have to resort to geoengineering techniques that deflect sunlight from planet Earth and remove carbon dioxide from the atmosphere. In this article, I develop a strategic theory of geoengineering. I emphasize two key features of geoengineering. First, whereas emissions reductions can be mandated now, geoengineering techniques are only available in the future. Second, major powers can unilaterally implement geoengineering projects that may hurt other countries. My game-theoretic analysis demonstrates that unilateral geoengineering presents a difficult governance problem if it produces negative externalities in foreign countries. Interestingly, countries may be tempted to reduce emissions now, so as to prevent a harmful geoengineering race in the future. The theoretical results can help scholars and policymakers understand the relationship between climate mitigation and geoengineering.


Global warmingClimatepolicyGeoengineeringInternational cooperationStrategyGame theory

1 Introduction

Climate change is now widely recognized as a serious threat to the human civilization (Schneider 2004; Stern 2006). However, humanity has failed to mitigate global warming by reducing carbon emissions. Consequently, several scholars have proposed that alternate solutions should be seriously considered. In particular, the notion of geoengineering—modifying earth systems to mitigate climate change instead of reducing greenhouse gas emissions—has recently gained currency. 1

Geoengineering is an enterprise fraught with difficulties. As Victor (2008, 323) writes,

The option of geoengineering is ridden with danger. All the most promising geoengineering methods have likely side effects that are worrisome. The unknown harms from large-scale tinkering with the planet could be even more grave than the predictable effects.

In addition, geoengineering is something that major powers can do unilaterally (Barrett 2008b; Blackstock and Long 2010; Schelling 1996). If the United States prefers to modify the climate, it may do so without the approval of other countries. Thus, the United States could implement geoengineering projects that protect Americans from the adverse effects of climate change but actually hurt the Indian or Chinese people. Similarly, China or India could unilaterally implement geoengineering projects that harm North America.

In this article, I begin to develop a strategic theory of geoengineering. The purpose of the theory is to examine the strategic logic of unilateral geoengineering and investigate how it influences mitigation efforts. If countries expect unilateral geoengineering to be feasible in the future, how does this expectation shape their incentives to reduce their emissions? In my model, countries first decide on emissions reductions and can then complement mitigation strategies with geoengineering. In addition to being costly, geoengineering by country A is assumed to be potentially harmful to country B. Thus, a governance failure may occur because each country implements geoengineering projects without due regard for the negative externalities that they produce in other countries.

What can game theory teach us about geoengineering? While game-theoretic models are based on strong simplifying assumptions, they can complement informal studies of geoengineering in several ways. 2 First, the assumptions of game-theoretic models are fully transparent (Morton 1999, 62). Therefore, game-theoretic models allow scholars and policymakers to test hypotheses and evaluate policy prescriptions in detail. Second, game-theoretic models are useful for incorporating strategic factors into studies of social behavior (Barrett 2003; Bates et al. 1998). Countries’ geoengineering decisions are strategic, so it is important to account for strategic considerations in the design of institutions for geoengineering.

The analysis produces several interesting results that can inform a governance analysis of geoengineering. First, if countries worry that unilateral geoengineering has serious adverse effects, they may strategically increase efforts to reduce emissions. By mitigating global warming, a country reduces other countries’ incentive to implement geoengineering. Thus, in addition to preventing climate change, emissions reductions have the extra benefit of avoiding unilateral geoengineering. Unfortunately, countries continue to emit too much carbon in equilibrium, because they continue to fail to fully internalize the global benefits of emissions reductions, both from preventing climate change and avoiding a harmful geoengineering race.

Second, I also investigate whether countries should collectively limit geoengineering in the future. If the negative externalities from geoengineering are substantial enough, then the option to modify the climate is actually collectively harmful. Equally important, if the negative externalities are not comparable to the damage caused by global warming, then geoengineering unambiguously benefits all countries. In these circumstances, it may nonetheless be useful to develop international norms and rules for geoengineering, as Victor (2008) has previously proposed.

This article contributes to the burgeoning literature on geoengineering in several ways. First, it provides a strategic perspective to geoengineering. Many recent studies have recognized the importance of strategy in countries’ geoengineering decisions (Barrett 2008b; Royal Society 2009; Bracmort et al. 2011; Victor 2008; Virgoe 2009), but these studies do not offer a precise characterization of the strategic logic of geoengineering. 3 A game-theoretic analysis lays a solid foundation for understanding the role of strategy in the governance of geoengineering. Second, a game-theoretic analysis can characterize the conditions under which the expectation of geoengineering increases and decreases countries’ incentives to reduce their carbon dioxide emissions. While both scenarios are plausible, my game-theoretic analysis can help scholars and policymakers examine the likelihood of the two opposite scenarios. As I detail below, such information can be useful for designing international institutions in view of both mitigation and geoengineering.

While this article focuses specifically on geoengineering, the results apply to a much wider range of strategic situations in global governance. Indeed, my model captures the core of any strategic situation such that countries must choose between (i) preventive action, such as emissions reductions, and (ii) treatment of symptoms, such as geoengineering. The core insight into the strategic analysis is that, as long as treatment strategies produce negative externalities, a good case can be made for favoring prevention. Additionally, the relative importance of these two strategies is strategically determined, so standard cost-effectiveness analyses may produce insights that are actually highly misleading.

The remainder of the article is organized as follows. First, I motivate the analysis by discussing mitigation and geoengineering in the strategic context of global warming. Second, I present the formal model. Third, I analyze the resulting strategic interactions and conduct a governance assessment. Fourth, I analyze three important model extensions: discounting the future, uncertainty about the costs of geoengineering, and variation in the domestic distribution of damages. Finally, I offer concluding remarks.

2 Geoengineering: a solution to climate change?

The definition of geoengineering that I use subsumes several different techniques that can be used to manipulate the climate without actually reducing carbon dioxide emissions. One prominent class of geoengineering techniques is based on deflecting sunlight away from planet Earth (Govindasamy and Caldeira 2000). For instance, sunlight could be deflected by launching gigantic satellite mirrors into the orbit or shooting tiny particles into the upper atmosphere (Crutzen 2006). Geoengineering also includes direct extraction of carbon dioxide from the atmosphere using techniques such as air capture or iron fertilization of oceans (Blackstock and Long 2010; Eisenberger et al. 2009).

Geoengineering can be considered an imperfect substitute for emissions reductions. As Barrett (2008b, 47) writes, it is a “stopgap measure.” Instead of tackling the root cause of climate change, namely burning fossil fuels, geoengineering treats the main symptom of temperature increases. Some geoengineering techniques do not remove any carbon dioxide from the atmosphere, while others do so and sequester it underground or below the sea surface. In any case, if major emitters were to significantly reduce their greenhouse gas emissions, geoengineering would play a minor role at best in the emerging climate regime.

Another notable feature of geoengineering is that the techniques remain unproven (Victor 2008). It may be a while until geoengineering techniques are actually feasible, whereas significant emissions reductions are attainable with current energy technologies (Stern 2006). Thus, geoengineering is best seen as a future technique to reduce the damage caused by global warming. In my formal model, I capture this idea by assuming that emissions reductions are temporally prior to geoengineering choices.

What is more, geoengineering may produce unintended consequences and harmful side effects. According to Crutzen (2006), injecting particles into the upper atmosphere may cause ozone depletion. Govindasamy and Caldeira (2000) note that geoengineering may also reduce global warming unevenly, so that some countries benefit more than others. Geoengineering may also significantly alter ecosystems if flora and fauna must adjust to an environment with less light and more carbon dioxide (Stanhill and Cohen 2001). Equally important is the fact that geoengineering is a planetary experiment without precedent. Given the magnitude of the resulting changes, it appears highly probable that it will have genuinely unforeseen consequences that surprise even the most astute observer.

The potential for unintended consequences and harmful side effects does not necessarily create a difficult governance problem, but the issue is further complicated by the fact that individual countries can unilaterally implement major geoengineering projects (Schelling 1996). While geoengineering may have harmful consequences, the direct cost of implementing a geoengineering project is much lower than that of aggressive mitigation policies (Barrett 2008b). Thus, if a major power expects to benefit from a particular geoengineering technique, the major power can easily implement a geoengineering project without external assistance.

These dynamics create a difficult strategic problem. If major powers begin to unilaterally implement geoengineering projects, they may hurt other countries in the process. Perhaps paradoxically, geoengineering may well resemble carbon dioxide emissions in that much of the true cost is being paid by foreign countries. Consequently, it may be that the world will see too much geoengineering, as countries race to implement geoengineering projects that benefit them despite high costs to others.

Millard-Ball (2012, 1052–1053) provides an excellent summary of the potentially harmful side effects that unilateral geoengineering may have on other countries. I summarize the most telling examples here. First, using atmospheric aerosols to reduce the global mean temperature could increase ozone depletion (Tilmes et al. 2008). This inflicts damage on vulnerable countries such as Australia and New Zealand. Second, a country could use a space deflector to reduce the global mean temperature while hurting forestry and agriculture in high latitudes due to less sunshine (Naik et al. 2003). This would hurt countries located in high latitudes and dependent on primary production. Finally, stratospheric sulfur injections could produce large changes in regional climates while effectively reducing the global mean temperature (Brovkin et al. 2009). For example, simulations by Brovkin et al. (2009, 252) suggest that “[i]n southern Europe and subtropical North America, summer aridity is increased.” Both areas already suffer from droughts, so the damages could be considerable.

The problems of geoengineering have been recognized in the literature. Victor (2008) argues that while geoengineering should remain on the table, it is important to begin developing international norms and legal rules to govern its usage in the future. Barrett (2008b) proposes a thorough scientific review of geoengineering by the Intergovernmental Panel on Climate Change and the formation of a new international agreement to govern geoengineering. Others, including Crutzen (2006), worry that emphasis on geoengineering may deflect attention from emissions reductions that would be otherwise feasible. Blackstock and Long (2010) emphasize the importance of transparency in a global geoengineering regime.

Virgoe (2009) reviews various policy approaches to governing geoengineering in view of promoting “science-based multilateral decision making.” He considers options such as a United Nations convention, unilateral action by a major power, and geoengineering by a smaller group of states. While he does not offer hypotheses concerning the probability of any three options, he does note that all are plausible and should considered in geoengineering governance. In a report to the United States Congress, Bracmort et al. (2011) review geoengineering options and the role of existing legal instruments for geoengineering. However, they refrain from offering any specific policy recommendation. A comprehensive overview of geoengineering by the Royal Society (2009) concludes with a discussion of governance options. While the report notes that “geoengineering must not divert resources from climate change mitigation or adaptation” (Royal Society 2009, 54), it does not contain other detailed predictions or specific policy prescriptions.

Although the extant literature lays a foundation for a research program on geoengineering governance, the current debate suffers from the absence of a fully strategic theory of geoengineering. Extant studies provide excellent overviews of the problem, but they are not based on detailed strategic models of how governments enact policies for emissions reductions and geoengineering. Their policy prescriptions are not based on explicit assumptions and propositions, so it is not possible to evaluate the conditions under which these policy prescriptions would apply. The utility of any given policy could depend on external conditions in a complex fashion, so a strategic theory of geoengineering can help policymakers evaluate what kind of policies the current circumstances warrant.

To be useful, a fully strategic theory of geoengineering should incorporate three key elements. First, geoengineering is a future technique, so it will be in demand especially if emissions reductions fail. Second, geoengineering is unilaterally feasible and may produce harmful side effects. Finally, the choice of mitigation policies will to some degree reflect the expectation of geoengineering. If these elements are included, a strategic analysis may shed light on the relationship between emissions reductions and geoengineering and help policymakers formulate a credible governance strategy.

One important article that adopts a strategic approach is Millard-Ball (2012). In his model, countries are considering the formation of a mitigation treaty. He shows that in some circumstances, the prospect of geoengineering can increase participation in the mitigation treaty because countries worry about the dangers of geoengineering. My model differs from his in four ways. First, he assumes that geoengineering is a binary decision, whereas I allow for realistic variation in the intensity of geoengineering. Second, his focus is on treaty formation, while I focus on countries’ mitigation decisions. Third, in his model, geoengineering itself is not a strategic game, while in my model, countries choose geoengineering policies in view of other countries’ geoengineering policies. Finally, my model produces more precise predictions regarding the conditions that allow the prospect of geoengineering to increase and decrease mitigation efforts.

3 Model

In the model, countries i = AB first decide on mitigation policies and then on geoengineering. As I explained in the previous section, mitigation policies are a mutually profitable public good because they reduce the rate of global warming. By contrast, geoengineering is a mixed bag. On the one hand, it does reduce the rate of global warming. On the other hand, it may have unintended consequences that hurt some countries more than others.

The model presented in this section is a particularly simple one. Below, I also consider three important extensions. First, I allow countries to discount future payoffs. Second, I examine the consequences of uncertainty about the negative side effects of geoengineering. Finally, I allow damages to be asymmetric within countries. It turns out that none of these extensions compromise the main results.

3.1 Sequence of moves

Formally, the sequence of moves is the following.
  1. 1.

    Each country i selects emissions \(E_{i}\in[0,\infty). \)

  2. 2.

    Each country i selects geoengineering \(G_{i}\in[0,\infty). \)

In the simplest possible fashion, this sequence captures the idea that emissions reductions require a sustained effort while geoengineering will produce instantaneous results. For instance, the choice of emissions Ei can be conceived of as capital investments that shape the carbon intensity of country i for decades to come. By contrast, the choice of geoengineering Gi can only be made in the future because the technology is not currently available.

To keep the model simple, I assume there are only two countries. The main insights remain unchanged if new countries are added to the game. Predictably, cooperation failures grow worse because each country faces a greater temptation to cheat (Olson 1965). The qualitative strategic logic remains intact, however, so I use a two-country model.

I also assume that the countries are symmetric. In reality, of course, major emitters are both economically and politically very different. Some are poor (China and India), while others are wealthy (Europe and the United States). Again, however, adding asymmetries to the model does not yield new qualitative insights. Thus, I assume symmetry to reduce notational burden.

3.2 Payoffs

The payoff to country i is chosen as follows:
$$ U_{i}=B\cdot {\hbox{ln}}(E_{i})-C\cdot(E_{i}+E_{j}-G_{i}-G_{j})^{2}-G_{i}^{2}-x\cdot G_{j}^{2}. $$
In the first term, B > 0 measures the economic benefits of burning fossil fuels. I use the logarithm ln(Ei) to ensure that these benefits are decreasing in scale. In the second term, C > 0 measures the damage caused by global warming. Realistically, this damage is increasing in scale, as positive feedback loops and nonlinearities begin to destabilize the climate system (Schneider 2004). The damage is increasing in total emissions, Ei + Ej, and decreasing in total investments in geoengineering, Gi + Gj. Finally, geoengineering is also costly. Country i pays a cost but also inflicts damage on country j. Parameter x > 0 measures the sensitivity of the foreign country j to geoengineering projects by country i. If x < 0, the negative externality to the foreign country is less costly than domestic implementation. But if x > 0, exactly the opposite is true. The main result would hold for many other functional forms, as long as standard technical properties continue to hold.4

3.3 Information

This is a game of complete information, so all parameter values are revealed to each country i at the outset. In reality, the benefits and costs of geoengineering are obviously subject to great uncertainties. All main results would continue to hold if there was some uncertainty surrounding the efficacy and net cost of geoengineering. Since this extension to probabilistic geoengineering effects does not add any insight, I assume complete information instead.

4 Analysis

In a dynamic game of complete information, the appropriate solution concept is the subgame-perfect equilibrium. Through backward induction, I first find the optimal geoengineering investments as a function of previous emissions. Next, I characterize the strategic choice of emissions in the first period.

4.1 Second period

To begin with, suppose the total stock of emissions from the first period is EA* + EB*. Each country i selects geoengineering Gi to maximize its payoff from the game. Using standard optimization techniques, the relevant first-order condition is
$$ 2C\cdot(E_{i}+E_{j}-G_{i}-G_{j})=2G_{i}. $$
The left (right) side is the marginal benefit (cost) of slightly increasing geoengineering.

Importantly, this expression shows that each country i fails to internalize the impact of geoengineering on country j in at least two ways. First, it ignores some of the beneficial effect of geoengineering on temperatures. Second, it ignores the damage inflicted by geoengineering on the other country. This formulation captures the governance failure that geoengineering may entail (Barrett 2008b; Victor 2008).

The first-order conditions define a linear system of two equations. Since there are two choice variables, Gi and Gj, this system has a unique solution:
$$ G_{i}^{*}=\frac{C\cdot(E_{i}^{*}+E_{j}^{*})}{1+2C}. $$
This expression shows how geoengineering investments Gi* are related to past emissions, EA* + EB*, and the damage that global warming causes, C. Specifically, geoengineering is increasing both in emissions and the damage that global warming, as shown in Fig. 1. These findings are intuitive and thus inspire confidence in the results of the model.
Fig. 1

Geoengineering choices as a function of the marginal cost of global warming (left) and emissions (right)

The expectation of geoengineering will shape emissions levels in the first period. How do countries choose the level of emissions, given that they expect geoengineering Gi* in the future?

4.2 First period

In the first period, country i maximizes its payoff with respect to emissions Ei, correctly expecting geoengineering choices Gi*Gj*. The first-order condition is
$$ \frac{B}{E_{i}}=\frac{2C\left(1+C+Cx\right)}{(1+2C)^{2}}\left(E_{i}+E_{j}^{*}\right). $$
The left side is the benefit from increasing emissions, while the right side is the cost of doing so. Importantly, the cost is increasing in total emissions, Ei + Ej.
Solving the resulting system of two equations, each country i selects emissions Ei* as follows:
$$ E_{i}^{*}=\frac{\sqrt{B\cdot(1+2C)^{2}}}{2\sqrt{C\cdot(1+C+Cx)}}. $$
This expression has several notable features. First, unsurprisingly, equilibrium emissions increase in benefits B. Second, also predictably, equilibrium emissions decrease in the cost of global warming C. Finally, equilibrium emissions decrease in the damage caused by geoengineering to the foreign country x. These functional relationships are shown in Fig. 2.
Fig. 2

Equilibrium emissions as a function of the marginal cost of global warming (upper left) and the marginal benefit of emissions (upper right), as well as the negative externality from geoengineering (lower left)

Most important is the observation that emissions EA*EB* decrease in x. Why does country i reduce emissions when geoengineering is potentially harmful to country j? The reason is that if country i selects high emission levels, it will induce country j to invest heavily in geoengineering in the future, and this investment will in turn harm country i. Thus, counter to intuition, the prospect of a “geoengineering race” will actually help countries reduce emissions.

Finally, having solved for equilibrium emissions, it is possible to uncover the equilibrium levels of geoengineering. Recall that they depend on past emissions, and these are given by Ei*. Inserting EA*EB* in expression (3), the equilibrium geoengineering investment Gi* can be written as
$$ G_{i}^{*}=\frac{C\cdot\sqrt{B}}{2\sqrt{C(1+C+Cx)}}. $$
This expression shows that geoengineering increases with the benefits of emissions and the cost of global warming, while it is decreasing in the damage inflicted on the foreign country. Thus, implicitly, the two countries are governing geoengineering through mitigation policies that help avert a harmful geoengineering race.

4.3 Discussion

The present equilibrium analysis shows how mitigation policies and geoengineering interact. In general, the possibility of geoengineering increases the incentive to emit, because the damage can be partially treated in the future. However, the equilibrium analysis shows that as geoengineering begins to produce negative externalities for foreign countries, the incentive to mitigate global warming increases. Paradoxically, the most harmful forms of future geoengineering may be the most conducive to mitigation policies in the present.

5 Governance issues

As I wrote above, various governance approaches to engineering remain on the table. In the previous section, I characterized the strategic relationship between mitigation policies and geoengineering. But so far, I have not considered the possibility that geoengineering will be proscribed because it may prove harmful. If the two countries could proscribe geoengineering, would they have an incentive to do so?

5.1 Behavior with and without geoengineering

To do this, let us compare equilibrium behavior in the above game with another game, otherwise identical to the original game except that GA** = GB** = 0. This game can be thought of as being played in a world without geoengineering. In this game, the payoff to country i is
$$ B\cdot {\hbox{ln}}(E_{i})-C\cdot(E_{i}+E_{j})^{2}. $$
Differentiate this expression with respect to Ei for i = AB and use the resulting first-order conditions to obtain equilibrium emissions
$$ E_{i}^{**}=\frac{\sqrt{B}}{2\sqrt{C}}. $$
Comparing this equilibrium, Ei**, with the original equilibrium, Ei*, shows that geoengineering may even decrease total emissions when the foreign negative externality x is too high: we only have Ei** > Ei* whenever
$$ x\,<\,\frac{(1+2C)^{2}-(1+C)}{C}. $$
The very prospect of geoengineering may sometimes actually increase the incentive to enact mitigation policies. The reason is that if the negative side effects of geoengineering are truly horrendous, the two countries may be willing to significantly reduce emissions, if only to avoid a geoengineering race. Surprisingly, the prospect of geoengineering may in these circumstances facilitate the emissions reductions that are required to prevent climate change.

Perhaps more realistic is the possibility that the expectation of geoengineering increases emissions, as condition (9) fails to hold. If countries expect that geoengineering may bring some relief, while understanding that they are unable to individually prevent a harmful geoengineering race, it may be individually rational to select high emissions. Unfortunately, this strategy is often collectively irrational, as demonstrated below, because the high emissions both increase the damage caused by global warming and trigger a costly geoengineering race to offset these emissions.

The contingent effect of geoengineering on countries’ incentives to reduce emissions is a central contribution of this article. Previous studies have discussed both the possibility that geoengineering could reduce countries’ willingness to reduce their emissions (Royal Society 2009, 54) and the possibility that geoengineering could allow international cooperation on emissions reductions (Millard-Ball 2012, 1056). However, these claims are mutually contradictory, so they cannot hold at the same time. My game-theoretic analysis shows that the magnitude and asymmetries of negative externalities from geoengineering are central determinants of whether the possibility of geoengineering encourages emissions reductions

5.2 Payoffs with and without geoengineering

What remains is to consider the effect of geoengineering on expected payoffs. Having solved for equilibrium emissions and geoengineering levels, all that needs to be done is to insert the solutions to the payoff functions. Inserting the equilibrium levels of emissions EA**EB** in expression (7), the payoff without the possibility of geoengineering is exactly
$$ -\frac{1}{2}B\cdot\left(2+{\hbox{ln}}(4)+{\hbox{ln}}(C)-{\hbox{ln}}(B)\right). $$
The payoff with geoengineering is, using equilibrium emissions EA*EB* and geoengineering GA*GB* in expression (1), given by
$$ -\frac{B\cdot(1+C+Cx)}{1+2C}-\frac{1}{2}B\cdot\left({\hbox{ln}}(4)+{\hbox{ln}}(C)-{\hbox{ln}}(B+2CB)\right). $$
The marginal in favor of geoengineering is therefore
$$ -\frac{B\cdot(1+C+Cx)}{1+2C}+\frac{1}{2}B\cdot\left(2-{\hbox{ln}}(4)+{\hbox{ln}}(B+2CB)-{\hbox{ln}}(B)\right). $$
The first term is clearly negative, while the second term is unambiguously positive. Equally clear, there is a cutoff \(\tilde{x}\) above (below) which geoengineering is harmful (beneficial) to the two countries.

We now observe the following logic. As the negative externality coefficient x increases, countries reduce emissions to avoid harmful geoengineering races. However, neither country fully internalize the harmful effects of a geoengineering race, so each fails to reduce emissions enough. If the negative externality is small enough, then the benefits of geoengineering outweigh the costs. But if the negative externality is large enough, both countries are worse off than without geoengineering.

It is important to recognize that proscribing geoengineering is usually not the optimal strategy. Instead, the two countries would be better off if they could agree on limited geoengineering. But even in this case, the main insight from the strategic analysis would hold: the key to assessing the urgency of developing governance rules for geoengineering is not the fact that they produce harmful side effects, but rather the fact that countries can unilaterally choose geoengineering projects that are costly for foreign countries. This problem is in the core of the governance dilemma that geoengineering may prompt. A future institutional solution to the governance of geoengineering should be based on a careful review of the magnitude and asymmetries of negative externalities from geoengineering, because these negative externalities determine how stringents the legal constraints on geoengineering should be.

6 Model extensions

The model presented above can be extended in three ways without changing any of the main results. First, what if governments discount the future cost of emissions and geoengineering? Second, what if uncertainty surrounds the negative effects of geoengineering? Finally, what if damages are asymmetric within countries?

6.1 Discounting the future

If governments discount the future cost of climate changes from emissions and geoengineering, their incentive to emit in the beginning of the game increases. However, except for this increase, the strategic logic of the model continues to hold. The relative magnitude of geoengineering damages and the cost of climate change determine whether the prospect of geoengineering induces countries to reduce their emissions or not. All hypotheses from the original model remain valid.

6.2 Uncertainty

Suppose uncertainty surrounds the costs of geoengineering. This uncertainty may reduce countries’ expectations about the costs of geoengineering and change their choice of emissions in the beginning of the game. The exact effect of uncertainty on emissions Ei*Ej* and geoengineering Gi*Gj* is complex, but the strategic relationship between the choice of emissions reductions and the expectation of geoengineering remains unchanged. Indeed, the hypotheses derived from the original model remain valid.

6.3 Asymmetric damages within countries

Damages from geoengineering could be asymmetric within countries, so that some groups within a country win, while others lose. These asymmetric damages mean that countries’ willingness to engage in geoengineering depends on how much their governments value the well-being of these different groups. While changes in different groups’ sensitivity to geoengineering have complex effects on emissions Ei*Ej* and geoengineering Gi*Gj*, the original strategic logic remains intact. Consequently, the hypotheses from the original model are still valid.

7 Conclusion

Global warming is perhaps the most difficult cooperation problem that the world has ever seen (Barrett 2008a; Stern 2006). The recognition of this basic fact has spawned a huge body of literature on possible solutions. Given that the primary cause of global warming is the burning of fossil fuels, most of these solutions focus on emissions reductions. However, the abject failure of world leaders to solve the problem through mitigation has recently strengthened the calls to consider alternative solutions.

One such solution is geoengineering, and several prominent scholars have indeed recently focused systematic attention on it (Barrett 2008b; Victor 2008). These scholars show that the demand for strategic and governance analyses is urgent. Geoengineering allows unilateral implementation, and this could be a virtue if collective action to reduce emissions proves overly difficult (Schelling 1996). But it could be dangerous, because major powers can implement geoengineering techniques without concern for consequences to people in other countries.

In this article, I began to develop a strategic theory of geoengineering in the context of international climate policy. My model emphasizes the negative externalities of geoengineering in foreign countries, as well as the fact that geoengineering is undoubtedly a technique of the future. It cannot substitute for emissions reductions at the time I am writing this article in the summer of 2010, but it may well do so in the future. Thus, present strategies to reduce emissions should account for possible geoengineering techniques.

My equilibrium analysis uncovers how strategic interactions between countries mediate the role of emissions reductions and geoengineering. Quite surprisingly, I have found that if geoengineering produces severe negative externalities, it may spur deeper emissions reductions in the present. If countries recognize the danger of a harmful geoengineering race—a plausible contingency—they have incentives to prevent it by reducing emissions now. Unfortunately, this incentive is partial at best, as countries fail to internalize the foreign benefits of limiting geoengineering.

I also conducted a basic governance analysis, comparing behavior and payoffs to the two countries with and without geoengineering techniques. I found that even unrestricted geoengineering can be globally beneficial if the negative externality that hurts foreign countries is not overly severe. However, in the opposite contingency, powerful restrictions on geoengineering may be in order. While this governance analysis is obviously too abstract to guide the design of international and domestic institutions for geoengineering, it is much more rigorous than the variegated verbal arguments that the literature currently offers. Thus, it is my hope that it will provide a robust foundation for more detailed investigations in the future.

Much remains to be done. I have used a simple and abstract representation of the negative externalities that geoengineering may produce. Including details from the environmental sciences, as well as economics, would improve the accuracy of the results and the aptitude of the policy prescriptions. It is as though we now have the basic building blocks for a comprehensive governance theory of geoengineering, nobody has yet managed to combine them. I believe that such interdisciplinary assembling is the next step toward a better understanding of geoengineering as a governance problem.

Yet it may not be too early to offer some preliminary policy guidance. Perhaps most importantly, the present analysis shows how the expectation of geoengineering may condition the optimal strategy for emissions reductions. If the magnitude and asymmetries of the negative externalities from geoengineering are large, then geoengineering builds political will for emissions reductions. In the opposite case, geoengineering will reduce countries’ incentives to reduce emissions. Further scientific research on the consequences of geoengineering is necessary, and a sharp focus on the negative externalities and their distribution across countries would be particularly welcome.

If world leaders decide to develop international institutions to govern geoengineering, they should not miss the opportunity to exploit this connection for mutual gain. Any rule for future geoengineering will indirectly impinge on present mitigation policies, so it is essential to account for this effect in policy formation. Given the deep uncertainties that surround the consequences of geoengineering, as well as the interactive relationship between geoengineering and emissions reductions, the idea of choosing between mitigation and geoengineering is fraught with danger. Ideally, the present article may inform robust strategies for a comprehensive policy portfolio to prevent dangerous climate change.


This broad definition of geoengineering includes both solar radiation management and extraction of carbon from the atmosphere.


The fact that a model relies on simplifying assumptions does not mean that the model is not useful. After all, every model is based on simplifying assumptions. For example, although computational models of climate change are highly unrealistic, many people find them useful for understanding climate change.


For an important exception, see Millard-Ball (2012). I discuss this article in the next section.


Mathematically, the benefit function must be strictly concave while all cost functions must be strictly convex. Additionally, emissions and geoengineering must have opposite and interactive effects on the total cost.


Copyright information

© Springer Science+Business Media B.V. 2012