Five years ago this week, I published a white paper entitled “How to Save the Planet: Be Nice, Retaliatory, Forgiving and Clear”. In it, I postulated that climate is a prisoner’s dilemma, but one one which will be played repeatedly, not just once. The difference is critical: instead of shooting for an unachievable top-down climate deal, we should focus on accelerating the inevitable emergence of domestic action on climate change. Since then we have seen the failure of Copenhagen, and the emergence of strong national responses. Is it time to look again at the game theory behind climate change?
Analysis of climate change from a game-theoretical perspective reveals an Iterated Prisoner’s Dilemma. As Robert Axelrod demonstrated in the Evolution of Cooperation (1985), such games are frequently characterised by the evolution of cooperative behaviour, independent of strong central authority. And indeed this is what we are already seeing in climate negotiations, with countries and regions increasingly committing to unilateral action.
The optimum strategy for an Iterated Prisoner’s Dilemma is to be Nice, Retaliatory, Forgiving and Clear. This provides a framework for the evaluation of strategies to date, which shows that no country or region has so far adopted an optimal strategy. TheUSneeds to start being Nice,Europeneeds to learn to Retaliate, and the developing world needs to Forgive. All players barEuropeneed to improve the Clarity with which they communicate their strategies.
The analysis also provides valuable insight into the optimal role of the UN. It should focus on its role as educator, coach and communications platform, rather than attempt to act as regulator and policeman. The UN should also find ways of breaking the negotiating process into smaller steps to encourage the emergence of sound national strategies.
For companies and investors, meanwhile, the lesson is that they should plan for a carbon-constrained future – irrespective of the outcome of upcoming negotiations.
As the Kyoto Protocol approaches the end of its working life, the only hope for the planet, according to its supporters, is to put everyone in a room – preferably somewhere exotic – and lock the door until a puff of white smoke announces the emergence of a new deal. There is a growing clamour for this to be done later this year in Bali, at the upcoming meeting of the signatories of the Kyoto Protocol. The logic is that there must be progress this year if the details of a successor regime are to be worked out by 2009, which would leave three years for implementation – the practical minimum if there is not to be a hiatus afterKyoto’s expiry in 2012.
The problem is that the world is beginning to realise that tackling climate change will be painful. It will require deep structural changes to our energy and transport systems, and changes in the behaviour of billions of consumers. And it will cost money. Even the Stern Review admitted that it will likely cost around one percent of GDP – a sum which doesn’t sound like much, until you put it in dollars, or realize that it is double the amount currently spent on development aid worldwide.
The US, with its expanding population, relatively consistent economic growth and extensive domestic coal reserves, sees a cap on carbon emissions as a threat to its competitiveness, and hence to its global hegemony. The developing world – led by China and India– denounces any calls for a cap on emissions as an effort by former colonial powers to hold back development. At a recent debate organized by the UN General Assembly, delegate after delegate stood up to insist that the developed world caused the problem, and the developed world must solve the problem – this despite the fact that China will this year become the world’s largest emitter of greenhouse gases. Europe, meanwhile, has been making encouraging though patchy progress towards its Kyoto targets, driven mainly by a one-off switch from coal to gas – leaving it cradled in Russia’s increasingly rough embraces – and partly by buying cheap carbon credits from the developing world. Canada, with its huge tar sand deposits is a key battleground in the fight against climate change has so far baulked at taking painful actions to meet its Kyotocommitments.
In this environment, the challenge of negotiating a successor treaty to Kyoto appears insuperable. But is there another way?
Is it possible that action to reduce emissions of greenhouse gases will simply emerge from each country’s unilateral response to the challenge of climate change? Impossible, cry the fans of Kyoto! How can we expect countries to make deep and painful cuts in emissions unless they first make a commitment to do so, binding in international law, and receive one in return from their economic competitors? It’s the Tragedy of the Commons, they explain. And indeed it is, but that does not mean that the only way to break the deadlock is through collective action mediated by the UN.
The Tragedy of the Commons occurs when a group’s individual incentives lead them to take actions which, in aggregate, lead to negative consequences for all group members. It is a multi-player version of the Prisoner’s Dilemma – a simple game which has been endlessly analysed since its first description in 1950 by Merrill Flood and Melvin Dresher of the Rand Corporation.
In the version of the game from which it got its name, the players are two prisoners, held in separate cells. Each has to choose between “cooperating” with the other (keeping quiet) or “defecting” (giving evidence against the other). Each makes the choice without knowing what the other will do. If both prisoners keep quiet, they are each sentenced to one year in prison. If one rats on the other, he or she goes free and the other gets 10 years. If they both rat on each other, they each get 5 years.
The problem arises because whatever your opponent does, defecting gives you a higher payoff than cooperating. Supposing your opponent is going to keep quiet, your best strategy is to give evidence and you go free. But if he or she is going to talk, you are better off returning the favour and getting five years in prison, rather than keeping quiet and getting ten years. In any game played rationally, both parties will defect. Each will end up in prison for five years, rather than one year: a worse outcome for both.
Understanding this simple game sheds light on many real-life situations. Two countries deciding whether or not to go to war are playing Prisoner’s Dilemma. So are two companies thinking about a price war. So are villages deciding how much water to extract from a limited supply. So is a couple, deciding whether to behave selfishly in a relationship. In short, it is one of the key games governing human interactions. And sure enough, in many cases we experience “defect-defect” type behaviour.
Climate Change as Prisoner’s Dilemma
At first sight, negotiations over climate change do indeed look like a classic Prisoner’s Dilemma.
The Stern Review even provides much of the data to populate the payoff table (see Table 2). If the world takes action now to tackle climate change it estimates that the cost should be up to 1% of annual per capita GDP; if the world does nothing, however, the cost will be in the range of 5% to 20% of GDP. So that defines what happens at the extreme of cooperative or non-cooperative behaviour.
If you are the only country not to take action, and all other countries do so, then you enjoy a “free-rider benefit”: climate change is mitigated, but you have not borne the cost of taking action – saving yourself the 1% of GDP that such action would have cost.
If, on the other hand, you take action and your trading partners do not, then you are a sucker. The Stern Review does not cover the cost of being a sucker to any single country, which will result from decreased economic competitiveness. The order of magnitude, however, can be estimated by comparing with the impacts of other known cost-side drivers of competitiveness such as energy costs or transport infrastructure, and may be as much as 3-5% of GDP.
Sure enough, combining these “scores” can lead to a payoff table that looks like a Prisoner’s Dilemma (see Table 2). It seems inevitable then that countries should choose to be selfish: the best national strategy, if others cut emissions, is to do nothing (defect); similarly, if other countries are not taking action, then it is pointless to be the only sucker to take action, and one should again do nothing.
Iterated Prisoner’s Dilemma and the Evolution of Cooperation
However, there is a flaw in this reasoning. Climate change negotiations do not constitute a one-off game of Prisoner’s Dilemma. In fact, negotiations over climate change take place repeatedly over a period of decades – they started before the Earth Summit in Rio in 1992 and will continue far out into the future. What does not make sense this year might make sense next year, depending on stages of economic development, political contingencies and the evolving scientific picture. Perhaps most importantly, countries can be expected to change their strategies over time in response to the actions of other countries.
So the interesting question is what happens if you play Prisoner’s Dilemma repeatedly over time with the same opponent?
In 1984 an American political scientist at the University of Michigan, Robert Axelrod, published a book called The Evolution of Cooperation, which contained the surprising answer: if you play a Prisoner’s Dilemma game, not once, but repeatedly, then what you are likely to see emerging is cooperative, rather than mutually destructive, behaviour.
Axelrod understood that there could be no one strategy for playing an Iterated Prisoner’s Dilemma since it was not possible to guess what other strategies one might come up against. What intrigued him was to identify the characteristics of strategies that did well when placed in a population containing a range of different strategies, and in particular under what conditions you might see cooperative strategies thrive.
He announced a competition, inviting the submission of computer programs to play Prisoners Dilemma, which he pitted against each other to see which strategies did best. Surprisingly, it was the simplest programme which won: called Tit-for-Tat, submitted by Professor Anatol Rapoport, a mathematical biologist from the University of Toronto.
Tit-for-Tat started each round by cooperating with its opponent, and then simply mirrored its opponent’s last move. Axelrod was struck by the elegance of the solution, but wondered if there was any way of beating Tit-for-Tat, especially if you knew that was your opponent’s strategy. He published the results, and invited another round of entries. Tit-for-Tat won again!
Axelrod decided to see how robust was Tit-for-Tat’s advantage. He ran simulations of the tournament, changing the mix of entries, to see if he could spot populations in which Tit-For-Tat did badly. Tit-for-Tat did well in almost all initial populations. He also ran an “ecological” simulation, whereby successful programmes multiplied and unsuccessful ones died out. Tit-for-Tat not only did well initially in a diverse field of strategies, but as unsuccessful strategies died out, it became even more successful. It was, in the parlance of the biological sciences, “evolutionarily stable”, unlike some other initially promising strategies, which died out when the unsuccessful strategies on which they preyed disappeared from the gene pool.
Axelrod went on to boil the essence of Tit-for-Tat down to four simple rules to follow when faced with an Iterated Prisoner’s Dilemma.
- Be Nice. Start by cooperating, and never be the first to defect. Otherwise you have no chance of getting into the zone where you both cooperate repeatedly and rack up the best outcome over time.
- Be Retaliatory. If the other player defects, inflict a cost on him or her which is at least as severe – otherwise you open yourself to exploitation.
- Be Forgiving. If your opponent mends his ways after defecting, restore cooperation as quickly as possible, so that you can both get back to scoring highly on each round.
- Be Clear. Since there is no way to beat the Nice, Retaliatory and Forgiving strategy, if your opponent knows you are following it, there is no incentive for him or her to seek advantage – it will only destroy his or her score as well as yours.
Axelrod’s Evolution of Cooperation explained the previously inexplicable: under what circumstances people should trust another player who can, in the short term, benefit at their expense. The implications are profound: in situation after situation where it looks as though central regulation is the only way to avoid the Tragedy of the Commons, we may instead see the Evolution of Cooperation. And indeed we frequently do – by and large, borders are respected, price wars averted, villages manage their water supplies, couples are considerate to each other and so on. Truly, this is an important piece of work.
Climate Change Negotiations as Iterated Prisoner’s Dilemma
Considering international climate change negotiations as an Iterated Prisoner’s Dilemma provides valuable insight into how countries should act, not just in the forthcoming negotiating session inBali, but in their everyday economic life, away from the negotiating table and over the long term. For instance:
- Be Nice. Sign up to unilateral cuts in emissions, as deep as your economy and financing capability will allow.
- Be Retaliatory. Single out players who have not begun to take action and, in cooperation with the others, find ways of hurting them until they do so. From a game theory perspective it doesn’t matter whether any retaliation takes place within the framework of an emissions treaty, or whether it is on a broader trade or diplomatic front – as long as it is clearly linked to the opponent’s actions on climate change.
- Be Forgiving. When recalcitrant countries do come in from the cold, no matter how obdurate they have been, welcome them with open arms. Fete them with lavish praise; provide them with finance and technology. Signal that good behaviour will be rewarded with even deeper cuts in your own emissions.
- Be Clear. Let everyone know in advance exactly how you are going to behave – that you will work with them if they take action on emissions, and that you will retaliate if they do not.
This analysis also highlights the fact that the negotiations themselves are not the game. Diplomats and politicians don’t reduce emissions, engineers and consumers do. Players will look for signals which cannot be falsified. These include meaningful changes to patterns of investment and tough regulations governing consumer behaviour. Populist statements, commitments too small to affect national budgets, programmes that bite only after the political lifetime of its author and other “cheap” signals will be seen as merely symbolic. They are worse than valueless to the process, since they serve only to reduce the Clarity of players’ true positions.
Evaluation of Country Strategies
Given this model of climate negotiations as an Iterated Prisoner’s Dilemma, how have the various players been doing so far, and how might they improve their game?
Europe has demonstrated its “Niceness” by unilaterally accepting, and doing the most to achieve, a cap on its carbon emissions underKyoto. Indeed, it has gone further, unilaterally agreeing to cut its emissions by 20% by 2020. The EU, however, suffers from structural problem which makes it unwilling ever to retaliate against those that take it for a ride. At the launch of the Euro, Greece’s central bank actually presented falsified economic data to ensure it was allowed to join; no sanctions were enacted. When a number of other countries failed to hit the targets of the Growth and Stability Pact, no action was taken. Fraud in EU programmes on an industrial scale goes largely undocumented and unpunished. The reasons for this weakness lie in the EU’s institutional structure, its political philosophy, and its history. When it comes to Forgiveness, the omens are good: the EU has already signalled that it will increase its target cuts in greenhouse gases to 30% by 2020 if other nations, including the US, China and India come on board – what is that but a Forgiving strategy. And it is not just Forgiving, but also Clear. Indeed one cannot fault Europe’s Clarity on the question of climate change and its strategy: it has always made clear that it will be a leading player in any negotiations and accept its obligations. Europe’s strategy, in summary, is to be Nice, Non-Retaliatory, Forgiving and Clear.
The problem is that in any game of Iterated Prisoner’s Dilemma against the EU, a strategy of defecting early and often will be the best one. To improve its strategy, Europe would need to identify ways in which it can punish nations that are not stepping up to the mark. These might include naming and shaming, carbon-related import taxes, or more general exclusions from the benefits of trade.
The US, on the other hand, at the federal level, has committed the cardinal error of not being Nice. Axelrod’s analysis shows that if you start a repeated Prisoner’s Dilemma series by defecting, you risk never getting into the high-scoring zone of mutual cooperation. Although you may reap a short-term benefit, you will lose in the long term. Luckily, at the state level – in California and in the New England states, in particular – this seems to be better understood. On the next two dimensions, being Retaliatory and Forgiving, the US has a strong hand. It has, in other spheres of international endeavour, only too frequently demonstrated its willingness to retaliate; but it has also demonstrated its willingness to forgive and reach out to former opponents – the help it provided to the formerSoviet Unionfollowing the collapse of communism comes to mind. On climate change, theUShas also signalled that it will be Forgiving: the Asia-Pacific Partnership on Clean Development and Climate Change, designed around large-scale technology transfer from theUS, can be interpreted as a reward for developing world countries which commit to cooperation on climate change.
Unlike the EU, theUS suffers from a lack of clarity, particularly as we go into an extended presidential election cycle. Its political system, with so many competing interests, and so hemmed in by checks and balances, can appear alarmingly opaque to outsiders. This is a handicap, perhaps one that constitutionalists willingly put up with, but which renders the achievement of global progress on emission reductions more fraught. So the US’s strategy can be characterised as Mainly Nasty, Retaliatory, Forgiving and Unclear. Playing against the US, one might be tempted to get one’s Retaliation in early.
To avoid this, US should be looking for ways to make a meaningful unilateral commitment to reducing its carbon emissions. Activity at state level, as well as the statements coming out of the G8 Summit at Heiligendamm, could already be seen as paving the way for a federal commitment. The Energy Bill that is currently deadlocked between Senate and the House of Representatives would be a good place to start: combining the most aggressive measures being suggested in one bill would send a clear message that the US is going to get serious on emission reductions and expects other countries to do likewise – with the veiled threat that defectors should expect to be punished. Another good signal would be to pass one of the many proposed bills on carbon caps being talked about on Capitol Hill.
It may be overly optimistic to expect much movement from the current administration, but the key point is that if the analysis of climate negotiations do indeed form an Iterated Prisoners Dilemma holds, then eventually we should expect policy makers, even at the federal level, to adopt a strategy that is in the US’s national interest and start taking meaningful action.
For the developing world –China, India and Brazil in particular – escaping from the Kyoto negotiations with no limits to their allowed emissions may look like a brilliant negotiating outcome, but it could end up proving very costly.
By demonstrating a refusal to be Nice, and by now refusing even to consider cooperation in setting carbon limits, developing countries are running a significant risk of Retaliation by the other players – as would be the lot of any player who sets out to play a Defect-only strategy in an Iterated Prisoner’s Dilemma.
Given the importance of energy in helping developing countries achieve increases in their standard of living, it is not surprising that they have hung back from taking action on emissions. And although their rhetoric on climate change has been Nasty, their actions have in fact displayed hints of Niceness. Almost all of them have started down the road of limiting growth in emissions, usually in the form of aggressive energy efficiency initiatives – despite refusing to sign up to a hard target – just as the Evolution of Cooperation suggests that they should.
In terms of Retaliation and Forgiveness, the Developing world is taking things to extreme when it uses climate change as a way to punish the Developed world for the sins of past colonialism and industrialisation. Its willingness to frame the debate in this way demonstrates an almost pathological lack of Forgiveness. Finally, developing countries’ militant rhetoric on climate change overshadows their very real initiatives to limit emissions and so harms the Clarity of their position.
In summary, most developing countries’ strategies can be characterised as Predominantly Nasty, Retaliatory, Stridently Unforgiving, and Insufficiently Clear. Rather surprisingly, it is developing countries whose strategy appears least well-adapted to the climate negotiation game. This is particularly worrying because, as noted by Sir Nicholas Stern and others, it is the developing world that is likely to suffer disproportionately from climate change.
The best strategy for the developing world would be to renounce the short-term advantages of non-cooperation, as well as the domestic political benefits of aggressive rhetoric, in favour of a clearer commitment to sharing in global efforts: in other words more emphasis on ‘common’ and less on ‘differentiated’ responsibilities. Otherwise it might find itself on the receiving end of either local retaliation for non-cooperation or, even worse, a complete breakdown in global attempts to mitigate climate change, which would have devastating regional consequences.
Implications for the UN
Thinking of climate change negotiations as an Iterated Prisoner’s Dilemma would also have profound implications for the role of the UN, and particularly the UNFCCC, keeper of theKyotoprocess and most likely of any successor regime.
Under this analysis, the UN should stop focusing on corralling all members of the international community into one room and trying to get them to make simultaneous, binding commitments on emission reductions.
Instead it should focus on encouraging the four key behaviours that make up good strategy. It should provide encouragement to nations to make meaningful unilateral commitments, whether as part of a treaty or not; it should help players who are making painful cuts identify appropriate mechanisms for retaliation against those who are taking a free ride; it should help put in place technology transfer and other programmes to ensure a warm welcome for late converts to the process; above all, it should provide a talking shop to improve the clarity with which countries communicate their commitments and their plans to deal with free-loaders.
The UN can also intervene to tweak the nature of the game in such a way that good strategies can emerge more easily. Instead of creating a one-off I’ll-show-you-mine-if-you-show-me-yours forum where all countries are expected to step up to the mark at once to make big commitments, the UN should try to break the process down into a larger number of small steps, lowering the cost for countries to experiment with cooperative strategies, and facilitating confidence-building (Forgiveness and Clarity). The UN can also use its own prestige and position to tilt the payoff table, providing some early rewards, either at the country or individual level, for those displaying Nice and Forgiving behaviour.
This is far from being an exhaustive analysis of the game theory of climate negotiations. In particular, a simple model of international emission reduction negotiations as an Iterated Prisoner’s Dilemma does not deal with the following factors, the impact of which would bear more analysis:
- Asymmetry. Not every country has the same to gain from mitigating climate change, nor do they have the same costs. In practice each country will have its own payoff table.
- Non-linear effects. Defection by one or both players increases the cost of climate change (even if only fractionally, for small countries), which should be captured in the payoff table, instead of expressing outcomes in the form of a range. Also, emissions and associated costs lie on a continuum which cannot be grouped into a simple two-by-two payoff table, and there may be significant threshold effects as global emission levels rise.
- Cumulative effects. The payoffs from successive rounds of climate negotiations are not independent, since carbon accumulates in the atmosphere and is only very slowly eliminated. In practice, therefore, the payoff table changes over time. The international process itself acts as a “lock-in” or ratchet, cementing unilateral commitments made by players, which has a cumulative effect on payoffs.
- Multi-player dynamics. In a simple Iterated Prisoner’s Dilemma model, each pair of players negotiates with each other successively, with outcomes independent of any other pairing. In practice, the outcome of each pairing does has an impact on other negotiations, by altering their payoff tables.
- Political factors. The payoffs for negotiators and politicians may not match those of their countries as a whole. For some, participation may be more important than achieving real progress. Others are determined to be seen by their domestic audiences to “win” negotiations.
- Alliances. The simple model does not allow analysis of the role of alliances between players. For instance, Europe might hold out the hand of Forgiveness to laggard countries while the US shakes the stick of Retaliation. Although political forces make it unlikely ever to be stated openly as strategy, such a strategy could be effective if consistently and rigorously implemented.
If the model of climate change mitigation as an Iterated Prisoner’s Dilemma holds, the good news is that the world is not going to wait for a single, binding, global deal in order to take action.
Countries may move quickly to adopt game theoretically optimal strategies, in which case carbon constraints will be imposed quickly and aggressively. If not, they may be implemented more patchily and slowly. But implemented they will be.
For companies and investors, the most important implication is that they should plan for a carbon-constrained future, whether the upcoming round of climate negotiations results in a major accord or not.
Axelrod, Robert. 1985. The Evolution of Cooperation. Basic Books, Inc, New York.
Axelrod, Robert. 1990. The Evolution of Cooperation (with foreword by Richard Dawkins). Penguin Books, London.