Climate change is a difficult problem to address because it is global, requiring concerted action by multiple sovereign countries. National sovereignty implies that international agreements require actions be in each signatory country's national self-interest. This feature has lead a number of scholars to conclude that such agreements are either implausible or ineffective. These problems are evident in the two significant agreements have emerged from international negotiations: the Kyoto Protocol in 1997, and the Paris Agreement in 2015. Among the features distinguishing these agreements, one stands out: the Kyoto Protocol entailed emissions reductions from developed countries, while the Paris Agreement on Climate Change involved all countries. Accordingly, participants in the Kyoto agreement were relatively similar in size and their level of economic development, while participants in the Paris Agreement were diverse. 

The pessimistic message from the literature described above involves symmetric players; allowing for asymmetric countries seems likely to further undercut the efficacy of forming an international agreement. I explore this aspect in this paper. To this end, I analyze data from an experimental analysis based on a relatively simple linear-quadratic payoff structure, where players are of two types: ``large countries'' and ``small countries.'' Large countries have payoff functions that yield larger rewards; this can be interpreted as resulting from lower abatement costs or lower marginal damages from emissions. In this way, large countries are emblematic of developed countries, while small countries represent developing countries. I find that subject choices in the presence of such heterogeneous payoffs are significantly closer to the (non-cooperative) Nash equilibrium than are choices in a symmetric structure. Moreover, the observed market shares for small countries exceed the shares such countries obtain in the Nash equilibrium. Indeed, the evidence from these experimental data shows that the small agent making output choices chooses outputs slightly above the Nash level and the large agent chooses outputs below the Nash level. In other words, it is the small producer that spoils the cooperative efforts of rivals.

While this experimental outcome cannot be explained by standard theoretical treatments – those generally allocate a larger share of combined emissions to the larger player – it is consistent with the Equity, Reciprocity and Competition model (Bolton and Ockenfels, 2000): while the typical small country's long-run output is close to its Nash level, the typical large country's output is smaller than its Nash level. This behavior is broadly consistent with features of the two climate treaties discussed above: In the Kyoto Protocol, large countries bore the load of curtailing emissions, suggesting that small countries – whom one might anticipate would best-respond to large countries' behavior – would likely increase their emissions. Under the Paris Agreement, countries are free to suggest emission reductions (through so-called ``Intended Nationally Determined Contributions''). One concern expressed by the current administration of the United States (US) is that smaller countries would ``take advantage of the US," presumably by raising their emissions in response to US reductions. The finding is also broadly consistent with casual empirical evidence taken from the numerous attempts -- via the annual ``Conference of the Parties'' that have taken place since 1995 -- to negotiate an international agreement limiting carbon emissions. For a long time, smaller (underdeveloped) countries insisted that larger (developed) countries bear most of the brunt in reducing emissions. My findings extend the commonly-found result in the theoretical literature that negotiations in similar repeated games (but with symmetric players) generally do not offer much hope for meaningful agreements, unless the effects are modest.

Reference

Bolton, G. E. and Ockenfels, A. (2000). ERC: A Theory of Equity, Reciprocity, and Competition, The American Economic Review 90: 166–193.