The Rationality of Climate Change: A game theoretic perspective

(1)

The Rationality of Climate Change

A game theoretic perspective

Master Thesis Philosophy University of Amsterdam

Jarno Hazekamp (10349243) Supervisor: Dr. Gijs van Donselaar Second Examinator: Dr. Jacques Bos

(2)

(3)

I would like to thank my supervisor Gijs van Donselaar for his professional supervision and personal support. I appreciate the time he spent to help me sort my thoughts, direct the project and finish the thesis. I want to thank him for his dedication in supervising my thesis as well as for the courses he gave, which I had the privilege to follow. I also want to thank friends and family for the support and encouragement. This thesis is dedicated to you and all (future-)existing beings on this planet.

(5)

Abstract

Climate change is one the world’s most prominent perils. While the causes and consequences of climate change are currently recognized by most nations, the situation is still not amended and emissions keep rising. Using game theory, I assess the rationale behind this problem. I pursue this using the following steps. First, I differentiate between the rational structure of adaptation and mitigation measures. Second, I try to find the right game theoretic structure representative of climate change as a collective action problem. Third, having established the structure of the problem, I use the Tit-For-Tat strategy to find a way out of the problem using the same approach of game theory. Lastly, in addition to conventional game theoretic

approaches, I look at the possiblity of a radical change in our conceptions of rationality and the consequences it has for our problem of climate change.

(6)

Introduction

The latest report from the Intergovernmental Panel on Climate Change (IPCC 2014) shows us that we should be concerned about climate change, as it threatens not only our way of life, but also life itself. The scope and magnitude of this global problem will eventually affect all of earth, but some regions will be more heavily burdened then others. Scientific recognition of the problem is increasing and has by now reached very high certainty rates as shown in the report. International efforts to abate the increasing threat inherent in climate change have yet only failed, while there certainly is sufficient international awareness of the problem. In order to combat climate change, multiple climate agreements have been signed by many

participating countries, such as the Bonn-Marrakesh agreement, the Kyoto protocol, the Copenhagen accord and most recently the Paris agreement. These agreements have the goal to reduce emissions in order for us to live sustainable, thereby decreasing the likelihood of catastrophic changes in our climate. Although the international effort to set up these

agreements give us hope for addressing the problem, the voluntary nature of the agreements have shown to be falling short of this promise, as the agreements turned out to be a sign of goodwill rather than true commitment.

Since the emergence of the problem, a lot of research has been devoted to climate change. Most of this research is empirical and focuses on the phenomenon itself. Less, but still a substantial amount of attention has gone to the issue of failure of international efforts to combat climate change. How can it be that (almost) everyone knows that their emissions are causing a problem that will in time be so costly that it threatens even human lives, but they still continue to emit the same amount every day, sometimes even more? It seems that there is something going wrong in our deliberation, for we recognize the problem, we know what causes the problem, but we do not alter our behavior.

Method and Relevance

To find out why we cannot, as rational beings, work our way out of the climate change problem, game theorists have tried to find the underlying structure of the situation in order to be better able to amend it. The game theoretic approach to assess certain situations is

sometimes contested within philosophical debate, mostly for expressing rationality as methodological individualism using preference maximization, a notion which can be

criticized for being unethical. However, without having to refer to ethics, the game theoretic approach can legitimize itself in virtue of other advantages. As Smead and Sandler point out, the primary advantage of game theory lies in its ability to simplify strategic interactions so

(7)

that their underlying principles can be understood (Smead and Sandler 2013). But they think we can use the approach even further within ethics, as game theory can be used to identify key features of the problem, characterize possible solution spaces and explore possible strategies for getting us to morally desired solutions (Smead and Sandler 2013: 15). Stephen Gardiner argues as well that game theory can contribute to finding solutions for certain

problems, especially in the case of climate change (Gardiner 2011: 50 – 63). For in the case of climate change, game theory can, first, shed a light on the structure of the situation and

formalize the complexity of the issue. If we can succeed in modelling the situation of climate change into a game, we do not only know why we are in the problematic situation we find ourselves in, but we can also, second, use this game theoretic model to find possible solutions. Knowing the precise structure gives us valuable insights into what kind of measures we have to take in order to make progress. If we know why we keep failing to deal with the issue of climate change, we have a better chance of changing the outcome. Thus, the main advantages of a game theoretic analysis of climate change are its ability to clarify the underlying structure of the situation and find out which adjustments to that situation are necessary to solve the problem, thereby aiding to end our inability to act properly.

Thesis

This thesis is a contribution to the efforts of using game theory to find new ways to effectively combat climate change. The research question of this thesis is as follows: using game theory, how can we best describe the structure of the situation of climate change and what do we need to change in order to reach a solution? The inquiry includes questions on how we can best represent the issue of climate change, which solution models are feasible and whether a radical break in our conception of rationality can solve our problem.

In this effort I focus on states as the relevant actors on the international stage1_{. The}

game theoretic approach is limited in its ability to encompass complex relations between large groups, which is why we need to make a choice as to who we see as the relevant agents. In our case, states have the largest impact on the environment and the ability to change their emissions, as well as the (assumed) legitimacy to act as representatives for their citizens. Thus, when referring to a ‘player’ in this thesis, I refer to any country, more specifically to the effective government of that country, that has the power to formulate and effectuate climate

1_{In this thesis I use the terms ‘states’, ‘nations’ and ‘countries’, interchangeably to refer to the relevant agents in}

the climate change problem approached through a game-theoretic perspective. The effective agents are the governments representing their nation’s citizens.

(8)

policy.

The thesis has the following outline. In the first chapter, I make a distinction between adaptation and mitigation measures and argue that the rationale behind taking these measures is essentially different. While mitigation measures are part of a collective action problem, adaptation measures are not. Based on Gijs van Donselaar’s work on international aid (‘Rational Helpers’, unpublished), I make the argument that the rational structure of taking adaptation measures requires that each country takes these measures and helps less developed countries take these measures right here and now up to the point they think is fair for

themselves. In the second chapter, I analyze the structure of the collective action problem of international cooperation on taking mitigation measures to combat climate change. Using Gardiner’s theory on climate change as ‘a perfect moral storm’ (Gardiner 2006, 2011), I explain the distinctive features of the situation and assess different game theoretic approaches to modelling the problem into a game. The first part of the chapter is dedicated to showing why we should treat climate change as a version of a Prisoner’s Dilemma. I also discuss different models, such as the Battle of the Sexes and the Stag Hunt, but argue, in line with Gardiner, that these models do not adequately represent the situation. In the second part of the second chapter I set out to find a solution model able to get us out of the situation. Such a solution model should work not only in game theoretic analysis, but should also be feasible in real-world international affairs, which is a proper challenge to develop. The solution I propose is based on the Tit-For-Tat strategy, executed on a global scale. In defending my proposal I discuss the features of the strategy, the feasibility of an international performance of such a strategy, the necessary conditions and institutions, along with the greatest threats to the success of this strategy. I conclude that the execution of the strategy is possible, but that its weakness in dealing with players with enormous bargaining power could undermine its ability to solve the issue adequately. In the third and last chapter, I take a different approach to solving climate change, where I assess the advantages of a radical break in rational choice theory based on the work of David Gauthier and Edward McClennen. The first part of this chapter describes the way Gauthier deals with Kavka’s Toxin Puzzle, resulting in a different view on the connection between rational intention-forming and rational action-performing. This alternative way of viewing intentionality and rationality has massive implications for game theory in general and for our case of climate change, as it tells us that states are only rational when they actually execute the plans they formulated in the signed climate

agreement. The second part of this last chapter discusses the even further radicalization of rational choice theory by considering Pareto-Optimizing as the alternative for the

(9)

conventional preference-maximizing conception of rationality. Gauthier and McClennen both argue for this alternative conception and I use the work of McClennen to show what happens when we let go of the separability principle. I show that it hands us a way out of the climate change problem without having to resort to costly enforcement mechanisms such as Tit-For-Tat. I end the chapter with the conclusion that the Pareto-Optimizing account carries promise to solve our problem of climate change, but that the radical nature of the theory makes it vulnerable to refutation and such it will have to face other challenges that I am not able to discuss here. I conclude the thesis with a summary of my findings and a few final remarks.

Chapter I. Mitigation and Adaptation

The latest IPCC report (2014) shows us worrisome figures about climate change on earth. The certainty rates about climate change are higher and the consequences more severe than

reported in the earlier reports, and this is a continuing phenomenon since the very first report. To predict how climate change will be developing, the report includes into the calculations various possible measures of abatement. These measures can be divided mainly into two different strategies: mitigation and adaptation. Mitigation measures are those measures that reduce emissions, thereby trying to reduce the rate and scale of climate change. Adaptation measures are those measures that already account for future climate change and try to adapt our surroundings to it. They are meant to adapt to climate change, not prevent it.

Technological advancements and improved infrastructure are amongst the most important measures of adaptation. The report argues for the importance of both measures, mitigation and adaptation, as neither of them is sufficient in itself to overcome climate change. While it is not certain that a combination of both measures is sufficient, it is agreed that it is necessary to even stand a chance. Mitigation is needed to prevent climate change reaching (even higher) critical levels, while adaptation is needed because climate change has already reached such levels that we need to adapt to its consequences. People will be enduring the effects of past emissions because of the time-lagging effect of climate change, which holds that the causes and consequences of climate change are apart more than 100 years. My claim is that there is a difference in rationale behind these distinct abatement measures that we need to acknowledge in order to tackle the problem in the best way. While mitigation measures are part of a collective action problem, adaptation measures are not. In this chapter, I will show why adaptation measures are not subject to the collective action problem and why there is nothing rationally withholding us from implementing and executing adaptation measures right away.

(10)

‘Rational Helpers’

The idea behind the different rationale of adaptation measures is based on Gijs van

Donselaar’s paper ‘Rational Helpers’ (unpublished). The paper views the rationale behind the reluctance of affluent states to relieve global poverty. Van Donselaar argues that “nothing in the interactive structure of the cooperative effort that is required to relieve global poverty can justify that affluent states would be reluctant in taking on at least their proportional share in that effort” (17). I will show, first, how Van Donselaar reaches this conclusion, shortly describing the rationality involved. Second, I will argue that this same rationale applies for adaptation measures against climate change and why it doesn’t apply to mitigation measures.

So how is Van Donselaar’s argument structured? The article is focused on the problem of global poverty, in particular on (avoidable) death by starvation. The underlying assumption is that people (and states as a collective of people) have at least a modest moral motivation and prefer less global poverty over more global poverty. The main claim is that affluent nations should make their proportional contributions to reduce poverty, independent of what other nations do. The argument that grounds this claim is that reducing global poverty is not a collective action problem and does not suffer from the inherent tendencies toward

sub-optimality such as in the Assurance Game or Prisoner’s Dilemma. To see why this is so consider the following logic: if there is a certain percentage of annual national budget that a nation considers fair as their share of global poverty relief, then they should follow through on spending that budget for poverty relief, no matter what other nations do, because their

respective contribution, however restricted, will save at least some people’s lives, making it well worth the effort. The saving of lives is not dependent on the action of other nations. Because every life saved is worth the effort spent, and affluent nations can save lives independently of other nations, there is no requirement of the nations working together as a collective to reach the desired goal.

However, as the desired goal is to save as many lives as possible, for which there is some cooperation required, there remains a bargaining problem. Here, Van Donselaar makes the distinctions between ‘benign’ and ‘vicious’ bargaining situations. When, in a bargaining situation, the outcome of the bargaining is not completely dependent on whether the parties reach an agreement, the game is benign. To illustrate, Van Donselaar uses the example of dividing 100 euros, where the game is vicious when the 100 is taken away if the parties do not agree on a division. To make this game benign, we only need to change the conditions such that the parties will at least receive the shares they agree about. So, if it is a two-player situation and A thinks he should get 60 euros and B should get 40, while B thinks she should

(11)

get 60 euros and A should get 40, then there is nothing rationally withholding us from both receiving the 40 euros they would allocate to each other. Then, 80 percent of the outcome is agreed to and only the rest of 20 percent needs to be bargained for. If this applies for the situation above, it also applies for global poverty relief, because it is a situation of the same kind. If country A thinks her proper proportional share is to spend 5 percent of her annual budget on poverty relief, and B should spend 7 percent, while B thinks it should be the other way around, then both should at least spend the 5 percent they agreed to for themselves. The bargaining area is thereby reduced to the leftover 2 percent, while the largest part of the poverty relief can be set in motion right away.

Van Donselaar’s argument is clear and the logic involved is relatively simple. It shows that the belief that we find ourselves in a collective action problem concerning international aid is false. However, global poverty is still leading to 28.000 deaths by starvation each day. This means that either too many people are subject to defeatism (thinking that a little help does not change anything), or they think waiting long enough will cause others to take on more burdens, or they simply do not care about those starving. Or maybe they think that they are only blameworthy when there are others who can rightly blame them, which, by collective inaction, leads to no one being blameworthy. Each of these, as Van Donselaar notes, is a false conviction, but whichever one it is, we can conclude that there is something withholding states from acting right here and now.

Help to Adapt

I argue that the rationale explained above to reduce global poverty is the same as the rationale behind adaptation measures against climate change and that the same conclusions apply. To see this, consider the following arguments. First, while mitigating climate change is a

collective action problem, which I will address further in the next chapter, adapting to climate change is not. We can already make adjustments to combat current and future consequences of the rising temperature and each of these adjustments is a contribution to the safety and welfare of the people for which the adjustment is made. Think for example of adaptations in infrastructure, water management, purification of air, etc. Second, affluent nations have enough knowledge and budget to make adaptational changes on their own, without the need for other nations to contribute. Third, the poorest countries are mostly the ones that will be hit the hardest by climate change, because of their geological location and their inability to take adaptation measures on their own (IPCC 2014: 13-16). Their situation is comparable with the situation of death by starvation in the sense that the people who live there are in real danger

(12)

and might be subject to death if there are no changes made. Those countries cannot solve these problems themselves, because they do not have the means (although there might be cases in which there are enough means in the financial sense, but where political instability or corruption prevents the correct measures to be taken). Fourth, we can assume some moral responsiveness for the people living in affluent nations, meaning that they prefer less deaths by global climate change over more deaths. This would lead them to pursue action right away when possible. However, even stronger claims could be made. There are those who argue that the dangers of climate change are caused by the development of the current affluent nations, which in turn gives them the duty to act according to this responsibility, meaning they should help alleviate the threat in poorer, less developed countries as much as possible (for example see Neumayer 2000).

The rationale behind adaptation measures against climate change are the same as for relieving global poverty. If country A thinks her proper proportional share is to spend 5 percent of her annual budget on helping vulnerable nations make adaptational changes to climate change, and B should spend 7 percent, while B thinks it should be the other way around, then both should at least spend the 5 percent they agreed to for themselves. Each country has their own expertise which they can put to use. The Netherlands for example have excellent knowledge of water management, which they could use to prevent floods in

vulnerable countries such as Bangladesh (for an assessment of the flood-vulnerability of Bangladesh, see Brouwer et al. 2007). The Netherlands can make this contribution on their own, without any help from other countries. There is no necessity for the affluent and capable nations to agree among themselves on the fair amount of contribution for each other, as they can rationally execute the amount they find fair and reasonable for themselves, while leaving the remainder as a bargaining situation. This means that there is still a (possibly large) part which needs to be solved through bargaining. The point is, however, that we do not have to wait for that bargaining to be successful in order to bring up the action today. If all nations contribute today that part they think is fair for themselves, an expectably large contribution worldwide would be made to adapt vulnerable nations to the dangers of climate change.

The rationale I explained is mostly focused on intragenerational help with adaptation measures. Next to this, I want to note that there is also an intergenerational dimension that requires some attention. In the next chapter we will see how the intergenerational dimension brings with it far-reaching complications for mitigation measures. For adaptation measures, it does not. For one thing, the rationale yet explained does already carry with it some notions of intergenerational concern. Climate change has yet to bring its most destructive effects, and

(13)

future generations will carry the largest burden. The adaptation measures we do today are mostly to prevent excessive damage in the future. It is a matter of preparing for what is coming. This means that we already take into consideration the fate of future generations in this matter. For example, in the Netherlands we construct our delta works with the idea that for many years coming they will have to withstand. But the rationale includes this

intergenerational dimension for other countries as well. If we are building dykes in Bangladesh, this is not only for the current generation that lives there, but also for future generations that are going to live there. The same rationale applies, assuming that we, as humans, also have some moral responsiveness to future generations. There is no conflict between generations, because adaptation measures taken now will also already improve the situation of the future generations, although more and heavier measures may need to be taken to secure the position of those future generations later.

I have argued that adaptation measures to combat climate change should be made right away and that there is nothing rationally withholding us from doing so. However, the fact that this is not yet happening leads us to the same conclusions as those of Van Donselaar, meaning that there is something else, probably a false conviction, that is withholding the affluent nations to act accordingly. Hopefully they will come to see the falseness in their convictions or take responsibility for their moral position. What I have not yet shown, is why mitigation measures against climate change are part of a collective action problem and what this means. This will be my endeavor in the following section.

Chapter II. Mitigating Climate Change

In this chapter I will consider the problem of climate change and explain why mitigating climate change is a collective action problem, after which I make a case for a solution model that stays within the conventional conception of rationality. In this endeavor I will look carefully at the theory of Stephen M. Gardiner, who calls the case of climate change the ‘perfect moral storm’. After explaining Gardiner’s view on climate change, I address some of his critics and an alternative approach to modelling climate change. Lastly, I will show the advantages of an international Tit-For-Tat strategy on climate policy and assess its feasibility.

II.I Modelling the Problem

‘A Perfect Moral Storm’

In the first section of this chapter, I investigate the characteristic features of climate change. In doing this, I will look mostly to Gardiner’s work, who sets climate change apart as the

(14)

‘perfect moral storm’. Gardiner’s first comparison of climate change to a perfect storm was in an article in 2006. The peculiar features of climate change make it hard for us to address the problem properly. Gardiner identifies three different dimensions that together form the perfect storm: the global storm, the intergenerational storm and the theoretical storm (Gardiner 2006: 398). For the first two of these, there are three characteristics of climate change responsible, namely dispersion of causes and effects, fragmentation of agency, and institutional

inadequacy (399)2. Dispersion of causes and effects in the spatial dimension is a well-known issue. If I emit right here and now, I do not know exactly how my emission will contribute to climate change. The specific consequences are unknown and no single emission is on its own responsible for a certain outcome. This is problematic because it gives us no direction of responsibility: my emission from driving a car on Sunday will not lead to a flood in Bangladesh.

Fragmentation of agency in the international situation, where the capable institutions are mostly recognized as states, leads to a situation much like the Prisoner’s Dilemma (PD). Gardiner uses Hardin’s ‘Tragedy of the Commons’ (Hardin 1968) to explain the situation. A tragedy of the commons is a prisoner’s dilemma involving a common resource, such as with the case of climate change, which leads to a collective action problem. Gardiner uses the example of overpollution:

(PD1) It is collectively rational to cooperate and restrict overall pollution: each agent prefers the outcome produced by everyone restricting their individual pollution over the outcome produced by no one doing so.

(PD2) It is individually rational not to restrict oneʼs own pollution: when each agent has the power to decide whether or not she will restrict her pollution, each (rationally) prefers not to do so, whatever the others do. (Gardiner 2006: 400).

To solve the tragedy of the commons, Hardin himself proposes a system of ‘mutual coercion, mutually agreed upon’ (Hardin 1968: 1247). This brings Gardiner to the third point, that of institutional inadequacy. In theory, the problem can be solved. In practice however, we lack the global governance that is required to put into effect such a system of mutual coercion (Gardiner 2006: 401).

Further complicating factors to the PD are cases such as the knowledge or reasonable

2_{From here on, when I refer to the same source multiple times, I abbreviate the reference to contain only the}

page numbers. When I refer to a different source, the complete reference will be made, after which it will be just page numbers again. I use this method to keep the thesis readable.

(15)

expectations of some countries to do relatively better under climate change than others, which gives them an incentive to contribute less to the solution (or, in other words, deny full

cooperation). Also, considering the fact that our civilizations are built on the sources that cause climate change, mitigating climate change will require severe and radical economical and cultural transformation. Together with the scientific uncertainty that still remains today, this brings forth a status quo bias. It is easier to give in to our uncertainties rather than transforming our economies (402).

Lastly, the skewed vulnerabilities, which I also discussed in the first chapter, are making things worse. Although the developed nations are mostly responsible for the climate change, they will not be burdened by it the most. The poorer and less developed countries will be so burdened, but do not have the power to force the developed countries into helping them. Also, if the developed countries decide to take responsibility for their past emissions, they might fall into a moral swirl of neglecting global poverty, human rights violations and other things for which they might have been responsible (402). Here, the situation that Van Donselaar sketches applies: under the affluent nations there is an equilibrium of avoided tu-quoques (Van Donselaar: 19). If they collectively deny responsibility, then there is no one blameworthy.

On the intergenerational dimension the dispersion of causes and effects also applies. There is a time-lag between emission and consequence. The carbon dioxide we emit now will only cause climate change far later, and it will stay in the air for longer periods of time. Climate change is therefore a resilient phenomenon, whose effects are backloaded, which means it is substantially deferred (Gardiner 2006: 403). This feature reinforces the problems already discussed.

The fragmentation of agency in the temporal dimension carries even worse consequences than it does in the spatial dimension. It brings forth a new and different collective action problem that is caused by the characteristic of climate change being a deferred phenomenon. Using cheap ways to grow economically fast will give current generations great benefits, while the consequences of these emissions will not be felt by the current generation. This causes an incentive to use these fossil energies. The next generation will feel the consequences of the past emissions, but they in turn will have the same incentive to use cheap and polluting methods to save their own welfare levels. This effect will iterate itself continually (404). Gardiner illustrates this problem by the ʻPure Intergenerational Problemʼ (PIP), where the generations do not overlap. In this case, the problem can be characterized as follows:

(16)

(PIP1) It is collectively rational for most generations to cooperate: (almost) every generation prefers the outcome produced by everyone restricting pollution over the outcome produced by everyone overpolluting.

(PIP2) It is individually rational for all generations not to cooperate: when each generation has the power to decide whether or not it will overpollute, each generation (rationally) prefers to overpollute, whatever the others do. (Gardiner 2006: 404).

Now in PIP1, the first generation is excluded. They have nothing to do with other generations as their life will not be affected by it. This causes an iteration of

current-generation-preference. Also, and even worse, in the intergeneration context there is no interaction between the parties involved. Future generations cannot coerce current generations to cut emissions. The possibility of an enforcement mechanism to hold each other in check is completely gone.

Lastly, the theoretical storm shows itself through our inadequacy to even theoretically come to a solution. The case of climate change could lead to moral corruption, for example by our selective attention to a certain facet of the problem (408). Focusing on the spatial

dimension tends to neglect the temporal dimension. Thus, if we, the current generation, keep focusing on the spatial dimension, we do not currently have to consider the temporal

dimension, resulting in a more lenient emission scheme for us today.

Climate Change: A Prisoner’s Dilemma

After writing his 2006 article, Gardiner published a book under the same name: A Perfect Moral Storm (2011). In this book, he worked out in more detail his thoughts about climate change, with a stronger focus on the game theory as well. In defense of his game theoretic approach, he offers us some deeper explanation as to why climate change should be represented as a PD. The key features of the PD are as follows:

a) The success of each actor in achieving his or her aim depends in part on what the other actors do;

b) The aims of the actors are moderately at odds, in the sense that they cannot all achieve the best possible result in terms of their aim;

c) Each actors’ individual pursuit of his or her aim results in a less successful outcome for that aim than would be secured by coordinated pursuit.

(17)

We can see that the same features apply to the climate change problem: we cannot solve it on our own (as a single state); we need to work together and each state that doesn’t cooperate is damaging the outcome; we cannot have our cake and eat it: solving climate change (if we can even speak of ‘solving’ the problem in such a way) will require a large sacrifice in our living standards and reduce economical growth; we have not yet developed adequate institutions to implement a mutual coercive pact amongst states. In this sense, the PD seems indeed to

adequately represent the problem.

Gardiner notes some important differences between the classical PD and the climate problem when we expand the dilemma to a multiple person game, which I have clarified in the table below.

Table 1. Differences between the classic PD and the climate problem in a multi-person game (from Gardiner 2011: 107, rewritten in a table by me).

Classic PD Climate problem

(Individuals & Firms, (I&F)) (PD3) Each agent must make only one

decision.

(I&F3) Each agent must make multiple decisions.

(PD4) The decision is ‘all or nothing’ and there are no intermediate options.

(I&F4) The decision at a particular time need not be all or nothing, since there are intermediate options.

(PD5) A single defection is decisive: it is enough to render the cooperation of other parties completely ineffective.

(I&F5) At a particular time, neither the defection of any one agent, nor even that of all agents, is decisive. (Currently we are collectively defecting, for we do not live at a sustainable rate and climate agreements are not lived up to. However, the world is not lost yet. So, while we are defecting now, this does not mean that future cooperation is ineffective.) Even so, such noncooperation does progressively erode the cooperative outcome.

(18)

(PD6) The decision must be made without any information about what the other agents have actually done.

(I&F6) Each agent could make future decisions in light of information about what others have done in the past.

(PD7) There are no independent

considerations to be taken into account.

(I&F7) There are independent

considerations to be taken into account.

As you see, there are significant differences between the two. Gardiner thinks they do not, however, undermine the point of the PD. This brings Gardiner to the next point, the Tragedy of the Commons. As we have seen, the central features of the Tragedy of the Commons (TC), are the same as those of the PD, such that:

(TC1) It is collectively rational to cooperate: each agent prefers the outcome produced by everyone cooperating over the outcome produced by no one cooperating.

(TC2) It is individually rational not to cooperate: when each agent has the power to decide whether or not she will cooperate, each (rationally) prefers not to cooperate, whatever the others do. (Gardiner 2011: 109).

However, there are also differences between the PD and the TC. The original TC was illustrated by Gardiner as a group of herdsmen that use a plot of land as a common resource. These herdsmen all have initially one animal grazing on the land. However, the herdsmen realize that they can individually add one more animal to the herd, giving them a higher expected payoff, while the cost, which is the degradation of the land by overgrazing, is shared with all the other herdsmen. Therefore, it is rational to add one more animal to the herd. But given this result, it is rational to add even more animals to their herd, and this rationale counts for all the herdsmen using the plot of land. It results in a collective overgrazing of the land, leading to the tragedy of the commons (Hardin 1968).

One difference between the PD and the TC is that in the original case by Hardin, the herdsmen make an iteration of marginal choices. Each time they make the consideration whether or not to add one animal to the herd. This makes it such that instead of PD3 and PD4, I&F3 and I&F4 hold. The other difference is that the effect of adding one animal to the herd is not decisive, such that defection is marginal as well. Each time they add one animal, the collective good erodes, but does not make further cooperation completely pointless. So I&F5 holds as well. Therefore, Gardiner calls the TC one that is evolving: ‘It is the cumulative effects of marginal decision making that results in collective ruin’ (Gardiner 2011: 110). The

(19)

result of the evolving nature of the tragedy is that the features change: the herdsmen can see what the addition of animals does to the pasture, which affect the incentives they face and gives rise to new options, such as trying to preserve what is left of the pasture. We should approach the TC as a problem of marginal choice, because first, in the real world people have budget constraints, and second, even if there are no budget constraints the added

environmental cost of each animal becomes critical at a certain point even if just one player adds them (112). Gardiner argues that this marginal approach gives us reasons to place restrictions on TC2. However, while these features give hope to finding a solution, there are also features that threaten this, such as the possibility for a race to the bottom, which only has to be initiated by one player and the rest will follow in order to maintain the status quo (113).

Optimistic Accounts of Climate Change

Gardiner also discusses alternative approaches to modelling climate change, such as the optimistic account. In this account, only partial cooperation is needed and agents involved act on internal motivation, to be understood as motivation internal to the game itself (86).

(Partial Cooperation) There is a number M (such that M < N) which is the minimum number of players whose cooperation is necessary if some situation, which is

dispreferred by all, is to be avoided (Gardiner 2011: 89).

Were the climate a ‘pure public good’ (which is both nonexcludable, meaning that no one can be prevented from consuming it, and nonrival, meaning that one person’s consumption does not affect another person’s consumption), it would be easy to solve in this way based on partial cooperation. Battling climate change would then be more like the example of building a lighthouse, where the necessity of it being there to guide boats is enough to make most agents contribute to it, even if some others choose not to contribute while entertaining the benefits anyway (85). It would represent what Russell Hardin calls a ‘step good’ collective action problem, which does not have the PD-like structure and relies on partial cooperation. (Hardin 1982: 55 - 61).

But our climate need not even be a pure public good in order to be solvable by optimistic accounts, which can be shown by the Battle of the Sexes (BoS) model. The BoS describes a situation between two agents that want to go on a date together, but disagree about the preferred venue. The choice in the example is between boxing and ballet, where both agents prefer going together over going alone.

(20)

(Fig. 1. From Gardiner 2011: 87, Table 3.1)

Emma

Boxing Ballet Ben Boxing 1st, 2nd 3rd, 3rd, Ballet -, - 2nd, 1st

In this game there are two Nash-equilibria, where an equilibrium point is ‘an n-tuple s such that each player’s mixed strategy maximizes his payoff if the strategies of the others are held fixed. Thus each player’s strategy is optimal against those of the others.’ (Nash 1951: 287). Here, both venues offer us an equilibrium point, but there remains a coordination problem as to which of these points to choose. Equilibrium selection is the main problem in this model.

Now, when the BoS is enlarged to encompass a group of agents, the conditions of partial and marginal cooperation come into play, as well as passive cooperation.

(Marginal Cooperation) If the number of others who are willing to cooperate is just short of M, then a given party prefers to cooperate, since each prefers to enjoy the benefits of cooperation and pay a share of the costs than to forego the benefits altogether (Gardiner 2011: 89).

(Passive Cooperation) Once the cooperating group is formed, the noncooperators prefer that it remain so, and succeed in its task. Hence, although they will not take on the costs of cooperation, they will also refrain from disrupting the efforts of the cooperating group (Gardiner 2011: 90).

Gardiner now wants to attack the optimistic account on two levels: practice and theory. To show why the optimistic account of climate change falls short, Gardiner discusses the Bonn-Marrakesh agreement, the Kyoto protocol and the Copenhagen accord, which are all based on optimistic accounts of climate change and are voluntary by nature. He argues that none of these agreements have shown to be sufficient in practice (Gardiner 2011: 94), for neither of them worked. Especially the lack of an enforceable global ceiling or strict national emission cap undermines the plausibility of a solution ever being achieved. The voluntary nature of such agreements is the main reason they do not work. We could note that, as the book is

(21)

published in 2011, the Paris agreement has followed since. But although there is some progress here when compared to the previous agreements (with the Paris agreement all countries were initially on board and financial support for less developed nations was

promised), the yet again voluntary nature of the Paris agreement has undermined its potential as a global solution to climate change. Pledges to reduce emissions, when not enforceable, can be easily overstepped when nations see the need to.

In addition to this practical failure, Gardiner wants to argue that also in theory the optimistic account fails. Partial cooperation is not enough to solve the problem of climate change (we need all countries to cooperate), while motivation internal to the game alone is not enough to incentivize nations to cooperate, which follows from the features that make climate change like a TC (95-101). The optimistic account of climate change, which can be

represented through the BoS game, is thus not the right way to represent the problem

according to Gardiner. Against this view by Gardiner and in support of an optimistic account which relies on game-internal motivation, we could use the first-mover example of Jean Hampton. The example goes like this: two soldiers are on a battlefield with the enemy approaching. They make an agreement to stay together and defend their post, which gives them a 50% chance to survive. They know that if one runs, while the other stays behind, the one who runs has 80% to survive while the one who stays behind has 10%. If they both run, they have 30% chance to survive. Now, if one of the soldiers has the urge to run away with the enemy advancing, he knows that if he runs, the other one will follow him. This knowledge is enough internal motivation to stick to the agreement and stay for both (Hampton 1986: 228). The example shows us that in some PD-like situations with contingent moves, the cooperative outcome can be realized without the need for a binding contract. Internal motivation, understood as motivation internal to the game itself, thus can be enough to prevent defection. For this to work, both transparency of action and knowing the reaction to contingent moves are essential.

The example is clear and the logic seems solid. But how would this look like in the context of climate change? I argue that in the case of climate change, it has proved to work the other way around. There has already been a first mover in the climate change situation, namely the past emissions of currently developed nations. Thus it is, in line with Hampton’s first mover example, also rational for other nations to follow the action of the developed nations, not restricting their emissions. To make the first mover example representative, we need to forget all past emissions (and their consequences) and start from the current situation as a new game. The rationality involved in that case would be: if country A violates the

(22)

agreed emission guidelines, other countries will follow this behavior, because the fact that they will succeed without A is extensively inhibited. A knows that the other countries know this and that they will follow with like behavior, also violating the guidelines. This will lead to an even worse situation for A than not violating the guidelines. Therefore, it is rational for A to stick to the guidelines without the need for signing a binding contract.

If this is the rationality involved in the climate change abatement agreements that were made in reality, such as the Kyoto Protocol and the Copenhagen Accord, why has none of these agreements proved to work? I think this is caused by the other complicating factors I discussed earlier in this chapter. They make the situation more like the Tragedy of the Commons, and less like the first-mover. The result is that the consequences of violating the climate guidelines are uncertain and tend to give a better expected outcome for such non-compliance, thereby decreasing the likelihood of sticking to the guidelines.

The other alternative and optimistic approach to climate change is the Stag Hunt model. The stag hunt game is one of coordinating cooperation. In the game, the players can choose to hunt stag or hunt hare. There are two stable equilibria, all hunt stag or all hunt hare. Getting stag requires cooperation and results in higher payoff, while getting hare can be done alone, but offers less payoff.

(Fig. 2. Stag Hunt)

Player B

Stag Rabbit Player A Stag 3, 3 0, 2 Rabbit 2, 0 2, 2

Gardiner reviews this alternative approach to climate change and rejects that it fits better than the PD (114, footnote 14). In the stag hunt, the cooperative outcome is stable, because there is no threat of defection. Defecting from cooperation does not give the defector a higher payoff. Instead, it gives her a lower payoff, yet with less risk. Defecting is thus not a dominant strategy, which it is in a PD. The dispreferred outcome is also stable, because there is no threat of a race to the bottom. Next to that, the dispreferred outcome is not tragic, as all can survive on the non-cooperative outcome (although one could argue that living on hare has a limit and eventually requires cooperation to get a decent meal as well). Based on these features, Gardiner argues that is does not represent the problem of climate change adequately

(23)

enough, because climate change does carry with it a threat of defection (which is the dominant strategy), a race to the bottom and a tragic outcome (114, footnote 14).

Gardiner does recognize that once we assume away the intergenerational storm, the stag hunt analysis has more potential. If the parties involved are seen as countries representing the interests of all their citizens across time, countries would not have the strong incentive to defect when other parties are cooperating. However, we cannot simply assume away the intergenerational storm, as history has shown that countries (their respective governments) work within a limited timeframe and have the tendency to take certain benefits today over ‘uncertain’ costs in 50 or 100 years.

This dismissal of alternative game theoretic models fueled a criticism of Gardiner by Smead and Sandler. In ‘Game Theory and The Ethics of Global Climate Change’ (2013) they argue that we need not only focus on cooperative games, but also look at the question of fairness of distribution in order to understand the relevant aspects of the climate change problem (17). Their central argument is that we need a pluralistic approach to climate change and use different games to explain different aspects. In this light, international negotiation should be represented by bargaining games, competition over resources by Chicken, and coordinating cooperation by Stag Hunt (17). Using the stag hunt analysis is important, then, to find out which of the equilibria we are most likely to achieve and how we are to get there. The game shows that even if cooperation is stable and efficient, we are still ‘stuck’ in the

suboptimal outcome, which is in itself equally rational (21). They further argue that the standard solutions to the PD are mostly those that change incentives so that the structure becomes more like a stag hunt. These external incentives try to make cooperation stable, but do not tell us how we get from non-cooperation to cooperation. Therefore, by dismissing the stag hunt so quickly and thereby the problem of equilibrium selection, Gardiner omits to describe an important aspect of climate change (21). Lastly, they argue that evolutionary game theory should play a larger role to the extent that we must acknowledge that in the real world, not all players are fully rational and that they change over time as the circumstances change (22). A lesson to be drawn from Smead and Sandler is that we need not only look to stable cooperative outcomes, but also to the conditions under which reaching and maintaining this outcome are possible.

Given these reasons, why should we still view the climate change problem as a PD instead of the alternative games? One reason is that, while PD can only be solved through external motivation, the alternative games can be solved both through internal and external motivations. This means that, even if in the end it shows to be a BoS or Stag Hunt after all,

(24)

finding external motivations is the safest way to go, because it works in both situations (Gardiner 2011: 106). The other reason is that even if the intragenerational problem

represented by the global storm is a BoS, the problem of intergenerational buck-passing (105) would likely prevent this from being solved. So Gardiner concludes first that it is unlikely that climate change is best represented by optimistic accounts, and second that even if it shows after a while to be better represented by an optimistic account, it would still be resolved by the external motivations necessary in a PD. Therefore, in the case of uncertainty, it is better to treat it as a PD (106). We can see that Gardiner holds a pessimistic view on climate change, which he himself acknowledges. In this pessimistic view, opting for the PD/TC-approach is the best way to go, because it calls for the harshest measures and efforts to secure cooperation and thus carries the least risk for failure.

Fairness in Bargaining

To go into the issues of fairness and equilibrium selection stressed by Smead and Sandler, we should consider the morally problematic aspect of climate change of skewed vulnerabilities. The nations that have contributed most to climate change have thereby created the means to take the necessary adaptation measures, while the less developed countries have not

contributed to climate change at the same rate, but find themselves in a geographically and economically worse position. This creates questions of responsibility and fairness. The game theoretic approach to solving climate change does not include notions of fairness - only bargaining problems and rational choice matter - but the skewed vulnerabilities that are at play do establish an unfair bargaining process. Gardiner argues that if the current state of international affairs is unjust, the bargaining approach to solving climate change between nations in their current relations could therefore also be inherently unjust (119). It could then be an example of what Henry Shue calls compound injustice: when powerful countries use the current structure of exploitation to advance their position further (Gardiner 2011: 119, he is referring to Shue 1996).

I agree that these skewed vulnerabilities pose a problem in the context of fairness within climate change. The situation of a TC is forward looking, ignores historical

responsibilities and assumes that everyone gains the same from cooperation, while in reality skewed vulnerabilities give a highly different payoff for nations. I do think, however, that what bargaining theory tries to do, finding out how rational parties would bargain for a certain distribution in a given situation, is not in itself dependent on the fairness of the initial

(25)

resulting outcome of the bargaining situation is also unfair given the unfair initial situation. While this might definitely be true, I cannot take such considerations of fairness of games itself into account in my current inquiry to find a rational solution to climate change within game theory. It would require a much more moralistic approach to the problem, which is what I have been trying to avoid by focusing on game theory and the concept of rationality. That being said, I have already used and will further use notions of fairness in this thesis in some way. In the first chapter that adaptation measures should be taken right here and now, given the rationale of each nation doing what they think is their fair share without awaiting the outcome of the complete bargaining problem. If you think doing A is fair, you have no reason not to do at least A. This could answer Shue’s worry about the almost exclusive focus on mitigation within the international arena, thereby neglecting the need for adaptation measures (Shue 1990). In the second part of this chapter, I will use notions of fairness as a tool for reaching an agreement to solve the problem. However, I want to make explicit here that I am not arguing that notions of fairness are rational in themselves.

Section Summary: Modelling the problem

To summarize this section, I have shown how Gardiner uses the rhetoric of the perfect storm to explain the peculiar and relevant features of climate change, which lead him to the

conclusion that the situation represents a PD in essence, with some features that make it like a TC. This pessimistic approach to climate change is worrisome for our hopes to find a solution, and alternative approaches have been voiced. Gardiner reviews the alternative approaches that rely on optimistic accounts of climate change and dismisses them for misrepresenting the situation. He argues that both practice and theory lead to more pessimistic accounts of climate change, which rely on strong external incentives instead of internal motivation. The history of climate agreements is evidence that internal motivation and voluntary agreements are not enough. While the first-mover example shows us that in some PD-like situations, internal motivation is enough to prevent defection, the specific and complex features of climate change are such that this is not one of those situations in which it works.

II.II Solution Models

This section is dedicated to finding the right solution model for climate change, one that stays within the conventional view of rationality. The solution models are based on the assumption that climate change is indeed best represented by a version of the PD, namely the TC. To start, there have been some proposed solutions, which are grounded on some specific

(26)

conditions. Elinor Ostrom for example explains the conditions under which we are more likely to find our way out of the TC:

Effective commons governance is easier to achieve when (i) the resources and use of the resources by humans can be monitored, and the information can be verified and understood at relatively low cost … ; (ii) rates of change in resources, resource-user populations, technology, and economic and social conditions are moderate … ; (iii) communities maintain frequent face-to-face communication and dense social networks—sometimes called social capital— that increase the potential for trust, allow people to express and see emotional reactions to distrust, and lower the cost of monitoring behavior and inducing rule compliance … ; (iv) outsiders can be excluded at relatively low cost from using the resource … and (v) users support effective monitoring and rule enforcement. (Dietz, Ostrom and Stern 2003: 1908, cited by Gardiner 2011: 116).

However, while these conditions can obtain in small communities, as Gardiner argues, they certainly do not obtain in the case of climate change (Gardiner 2011: 118), for climate change occurs under very different conditions, the ones I explicated in the first part of this chapter. The rest of this chapter is used to formulate a different solution model, one that I think holds the best promise of achieving the wanted results while still being feasible on a global scale.

Tit-For-Tat

The model I am proposing has not received much attention yet as a solution model to the climate problem as a PD. The proposal is based on the Tit-For-Tat strategy in iterated Prisoner Dilemma’s. This strategy became known by winning two game theoretic tournaments, where scholars tested their strategies in iterated PD-like situations. Anatol Rapoport submitted the Tit-For-Tat strategy and won two times. Robert Axelrod, together with William Hamilton, explained why the strategy was successful in Evolution of

Cooperation (1981). Tit-For-Tat is used in evolutionary game theory, which they use for explaining the process of evolution in biology.

In game theory, Tit-For-Tat is a strategy where the first move is cooperative, while the following moves are mirroring the move of the other player in the game. It is a ‘strategy of cooperation based on reciprocity’ (1393). Starting with cooperation, if the other player cooperates as well, both players continue to cooperate. If the other player defects, the Tit-For-Tat strategy punishes defection with defection. It thus takes into consideration the history of choices and bases the next move on the last choice of the other player, thereby achieving an evolutionary stable strategy. A stable strategy is seen as one by which ‘a population of individuals using that strategy cannot be invaded by a rare mutant adopting a different strategy’ (1393). Within the standard PD, such a stable strategy is defection.

(27)

(Fig. 3. From Axelrod and Hamilton 1981: 1392, reproduced by myself.) Player B Player A C Cooperation D Defection C Cooperation R = 3 Reward for mutual

cooperation S = 0 Sucker’s payoff D Defection T = 5 Temptation to defect P = 1 Punishment for mutual defection

When the PD becomes iterated, the viable strategies change. Defection is not the only stable strategy anymore. However, when the number of games is known to be a fixed number of encounters, defection is still the strategy to go for. This is shown through backward induction, as it is rational to defect in the last game, it is also rational to defect in the next-to last, and so forth until the very first game. Only when the number of games is not fixed, such as the real-world encounters of everyday life, we have different strategies to take.

I leave aside the evolutionary application of the strategy, as my main concern is its relevance for game theoretic approaches to climate change. First of all, Tit-For-Tat is useful for situations in which there is an unfixed number of iterated games continued over time. In the case of climate change, we are all stuck together in the same situation (at the moment no one can just stop playing the game by leaving this earth and live elsewhere), which proceeds over time and which has intervals in which all players can make decisions whether to

cooperate or defect. There is no known last game, and the payoff structure is like the PD, as we have seen earlier in this chapter. So it seems that the conditions hold under which the Tit-For-Tat strategy is a good strategy. To be able to define a good strategy, we can use the criteria that Axelrod and Hamilton use, namely robustness, stability and initial viability (1393). Robustness is the ability to deal with a variegated environment in which other players use a wide variety of other strategies. Stability is, as we have seen, the ability to resists

‘invasion’, in other words, it is an equilibrium. Lastly, initial viability is the ability to get foothold in a predominantly noncooperative environment (1393).

The robustness of the strategy is shown by the two tournaments in which multiple strategies were used and Tit-For-Tat won both times as producing the highest results. The features responsible for this success were identified as being benevolent (it never defects

Fig. 3 The Prisoner’s Dilemma game. The payoff to player A is shown with illustrative numerical values. The game is defined by T > R > P > S and R > (S + T)/2.

(28)

first), provocable (defection is punished by defection) and forgiving (it will reward

cooperation with cooperation after just one turn) (1393). The stability of the strategy follows from the proof that neither strategies that defect only nor strategies that alternate between defection and cooperation can invade Tit-For-Tat, given the features above and a sufficiently large probability of continuing interaction (see 1393 for a full formal proof). I should note that there is an argument, made by Jiborn and Rabinowicz (2003), that states that the backward induction rule counts also in indefinitely iterated PD-interactions if it can be defended for finite extensive-form games. This could prove to be undermining the plausibility of our application of Tit-For-Tat. However, as Jiborn and Rabinowicz point out, the argument for backward induction is a highly contested one in itself and I will not try to refute it here, for I do not have the required space. If I would go there, my argument against backward induction would be along the lines that in our case, the relevant agents are nations that can exist

indefinitely in theory and are thus not limited to the relatively short time our natural bodies give us, thereby reducing the certainty that there will ever be a last game, let alone when this last game will be.

Lastly, the initial viability of Tit-For-Tat in the original text is conditioned by

relatedness, an act of altruism and reciprocity. In the biological sense of evolution, relatedness of species causes them to have an interest not only in themselves, but in the well-being of the others too. This could cause an act of altruistic sacrifice, followed then by cooperation through Tit-For-Tat.

So how would this strategy play out in the case of climate change? Let me start with a few differences between our case and the original Tit-For-Tat application. First, in the case of climate change we are in a multi-personal situation of iterated interactions over time. How this will affect the strategy will come back later in this section. Second, the reason for the initial viability is slightly different. We cannot assume there is one first act of sacrificial altruism, but we also do not need to assume this. In our case, what we need is an incentive to initiate a cooperative move. This incentive is already given to us, namely the recognition of the immense danger that climate change proves to be. We need to cooperate. Continuous defection might well prove to be unstable in this case, as it could lead towards severe harm and ultimately death. For initial viability, we need only a sufficiently large number of players willing to engage in cooperation. If there is a sufficiently large amount so that unwilling players are immediately punished, the viability holds. Third, the punishment for defection is not the exact mirror of the punishable action. If one nation does not hold itself to the

(29)

emitting more, as this undercuts the point of the strategy. What we can do to punish defection is an act of exclusion. This entails that the strategy uses benefits of cooperation on other areas as well, such as trade, exchanging knowledge and diplomatic cooperation, from which players can be excluded as punishment of defection. When one nation defects from the agreement, the collective of other nations will exclude them from all other benefits of cooperation, thereby reducing the outcome of defection to suboptimal levels. This makes the Tit-For-Tat strategy insusceptible to exploitation in theory. It is why Tit-For-Tat is more resilient than the first mover approach, as the first mover approach works only one way: if player A does decide to flee from the battleground, player B should follow and there is no way for player B to get player A to stand their ground. If player B stops fleeing and decides to stand his ground again, the rational choice for player A is still to flee.

Implementation

The next question we have to address is how to effectively implement a Tit-For-Tat strategy on an international scale. First of all, institutions that can provide a high level of transparency of action need to be put in place. Without transparency of action, the Tit-For-Tat strategy fails. We need a high amount of transparency about the emissions and the efforts to stick to cooperative agreement for each country. To obtain this, we could put country-independent institutions in place that monitor emissions and environmental behavior on a global scale. We already have sufficient means to implement institutions like this as well as the scientific knowledge and technology in order to adequately monitor emissions and greenhouse gasses (one example is GHGSat). Second, we need transparency of intentions in the way that tit-for-tat works as a clear strategy, namely the transparency of the intention by other cooperating players to punish the non-cooperator through exclusion. This kind of transparency could be secured by signing a climate agreement that states to implement a Tit-For-Tat strategy. Third, we need initial agreement amongst (almost) all nations to execute the strategy. There need at least be enough consensus to implement it in order to put effective threat to recusants. This entails that we need to prevent coalitions of recusants being formed. To reach this initial willingness to cooperate we would at least have to assume that every nation is strongly convinced that climate change is indeed a catastrophic development that needs abatement and a change in our ways of life. Without such conviction, implementing the Tit-For-Tat strategy will be impossible.

We also, however, need to include into our distribution a notion of fairness. Earlier I shortly discussed the issue of fairness within bargaining theory and stated that for the

(30)

purposes of my thesis I could not go into the issue further. I now want to make an adjustment to that statement. Namely, if a global Tit-For-Tat strategy against climate change is going to work, all nations must agree on the conditions of the cooperation, which in turn must

incorporate some notion of fairness. For example, the difference in development per nation reflects to a certain extent that nation’s responsibility for the current rate of climate change and its ability to adapt to it. For any nation to accept any given outcome, this outcome must be seen to be (reasonably) fair. It is tricky to venture here, because notions of fairness open up all kinds of problems in a given distribution, but I do not think the Tit-For-Tat strategy is going to work if there is a large degree of unfairness involved in the distribution falling under the strategy.

I think we could make some basic statements about what is required if we want it to work in general: First, we assume some minimal degree of moral responsiveness in the same sense as I did in chapter 1. Otherwise, al notions of fairness are off the table. Second, the adaptation measures I wrote about in chapter 1 should be taken right now, in order to make every nation (at least to a high extent) ready for the consequences of climate change. This also increases the acceptability of any distribution of emissions to those countries who are most vulnerable. Third, there must be some degree of aid in the transition to greener energy for those less developed countries. If the more developed countries deny this, the less developed will have a larger incentive to increase emissions in order to gain economical growth. It also increases the sense of fairness of the distribution, given that the developed nations have caused the largest part of the problem.

Thus we have two possible approaches to bargaining theory. One incorporates a notion of fairness and the other bases the outcome solely on bargaining power, leaving notions of fairness aside. The question we should ask then is: is it rational to accept an unfair distribution if the unfair distribution is the only thing you can get and it gives you a better outcome than stepping out of the game? I think under the conventional conception of rationality the answer is ‘yes’, but the dichotomy need not be that strong in the case of climate change. Think, for example, what would happen if the less developed nations do not want to cooperate, because they think the terms are unfair. It would give them a reason to defect, thereby obstructing the goal for the others to achieve a solution. This means that they can transform their sense of fairness into bargaining power, threatening to defect if the more developed countries do not agree to a certain level of aid or advantages for the less developed countries. Fairness-sensitivity and bargaining power can both be used to reach a distribution of goods.

The Rationality of Climate Change: A game theoretic perspective

The Rationality of Climate Change

A game theoretic perspective

Table of Contents