Research Collection

Doctoral Thesis

Negotiating climate changepositioning behavior, cooperation and bargaining success

Author(s): Weiler, Florian

Publication Date: 2012

Permanent Link: https://doi.org/10.3929/ethz-a-007600536

Rights / License: In Copyright - Non-Commercial Use Permitted

This page was generated automatically upon download from the ETH Zurich Research Collection. For moreinformation please consult the Terms of use.

ETH Library

DISS. ETH NO. 20719

NEGOTIATING CLIMATE CHANGE: POSITIONING BEHAVIOR,COOPERATION AND BARGAINING SUCCESS

A dissertation submitted to

ETH ZURICH

for the degree of

Doctor of Sciences

presented by

FLORIAN WEILER

Mag. phil., University of Innsbruck

Master of Arts, Johns Hopkins University

born 9th August 1977

Innsbruck, Austria

accepted on recommendation of

Prof. Stefanie BailerProf. Katharina MichaelowaProf. David Armstrong III

2012

Table of contents

Curiculum Vitae Florian Weiler ..................................................... 4

Summary (English) ........................................................................ 7

Einleitung (Deutsch) ...................................................................... 8

Introduction ................................................................................... 9

Chapter 1 – Climate change negotiations, negotiation positions,and domestic structures ................................................................. 23

Chapter 2 – Determinants of bargaining success in theclimate change negotiations ............................................................ 52

Chapter 3 – Cooperation in the climate change negotiations:A network approach ......................................................................... 84

Chapter 4 – AOSIS in the UNFCCC negotiations:From unity to fragmentation? .......................................................... 117

Conclusion ....................................................................................... 146

Curriculum VitaeFlorian Weiler

Education

10.2009 – 08.2012 Federal Institute of Technology, Zurich, CHPh.D. in Political Science, expected October 2012

Title: ’Negotiating Climate Change: PositioningBehavior, Cooperation, and Bargaining Success’

The dissertation focuses on the UNFCCC negotiationsand investigates which determinants influence aparty’s positioning behavior. In addition, cooperativebehavior is examined through network analysis, andfinally I explore how these different componentsimpact a country’s bargaining success.

Selected coursework: Program Evaluation andCausal Inference, Econometrics for Ph.D.-Students,Quantitative Methods for Ph.D.-Students, Panel DataMethods, Time Series, Programming in R, MultilevelModels, Non-parametric Methods, etc.

09.2007 – 05.2009 Johns Hopkins University, Washington DC,USMA in Economics and International Relations

Concentrations: Quantitative economics and environ-mental studies

Selected coursework: Impact Evaluation,Econometrics, Applied Econometrics, Cost-Benefit-Analysis, Topics in Growth and Development,Global Climate Change Policy, Economics of NaturalResources, Corporate Finance, etc.

09.2000 – 05.2005 University of Innsbruck, ATBA and MA in Political Science

Final thesis: ’Are the UN Millennium Goals effectivetools to fight poverty?’

Spent one year as guest student at the University ofPavia (Italy)

Gschwaderstraße 11 • CH-8610 Uster • Phone: 0041-44-632 58 73Mobile: 0041-78-822 08 46 • Email: weilerf@ethz.ch

Employment

05.2009 – 09.2009 Consultant at the World Bank, WashingtonDC, US

Worked on the determinants of democratization (eco-nomic, institutional, and connectivity factors), collec-ted cross country data from various sources, and pre-pared a draft paper summarizing the findings.

01.2007 – 08.2007 Researcher at the Representation Office Tyrol,Brussels, BE

Gathered information about ongoing EU legislation,attended meetings, and informed relevant decisionmakers

06.2005 – 12.2006 Freelance Journalist

Summer Schools

2011 ICPSR Summer School, Ann Arbor, US

Courses: Advanced Regression Analysis, SurvivalAnalysis, Statistical Graphics for Visualizing Data

2011 ECPR Summer School, Essex, UK

Courses: Bayesian Inference for the Social Sciences,Introduction to Multilevel Modeling, Mathematics forSocial Scientists

Grants

2010 & 2011 Grant for attendance of Summer School, SwissNational Science Foundation

2010 & 2011 Grant for attendance of conference (APSA andECPR), Swiss Academy of Humanities and SocialSciences

2007 – 2008 Two-year scholarship for post-graduate studies inthe US, Federal Ministry of Science and Research(Austria)

Gschwaderstraße 11 • CH-8610 Uster • Phone: 0041-44-632 58 73Mobile: 0041-78-822 08 46 • Email: weilerf@ethz.ch

Skills

Level of knowledge:+ basic, ++ intermediate , +++ advanced

Field Section Knowledge in years

Statistical Software STATA +++ 5

R and S-Plus +++ 4

MLwiN ++ 3

SPSS ++ 3

JAGS and OpenBUGS ++ 2

SQL ++ 1

Brutus Computing ++ 1

EViews + 2

Other Software MS Office +++ 15

LATEX ++ 3

Languages German native speaker -

English proficient (C2) -

Italian proficient (C1) -

French intermediate (B1) -

Gschwaderstraße 11 • CH-8610 Uster • Phone: 0041-44-632 58 73Mobile: 0041-78-822 08 46 • Email: weilerf@ethz.ch

Summary

The recurring climate change negotiations under the auspices of the United Nations Frame-

work Convention on Climate Change (UNFCCC) are the most important diplomatic platform

to solve the global climate crisis. The decisions adopted during these meetings are thus very

important for the future of our species, which makes the negotiations a highly interesting

topic to study. This dissertations sets out to investigate various elements of the bargaining

process, more specifically countries’ positioning behavior, coalition formation and position

coordination, as well as the determinants of successful bargaining tactics. The synopsis of

the dissertations draws a picture of how a better understanding of these different components

can help to overcome the deadlock of the negotiations.

This thesis is part of the researcher project ‘Negotiating Climate Change’, funded by the

Swiss Network for International Studies (SNIS). All chapters rely on a new dataset collected

in the framework of this project. Over a roughly one year period the research team visited

various international conferences in Bangkok, Bonn, Barcelona, Copenhagen, and Cancun

to interview delegates and to gain a better understanding of the negotiation process. These

data will be made available to other researchers in the future.

The first chapter on positioning behavior shows that there exists a stark contrast between

Annex 1 and non-Annex 1 countries, but also that the gap between countries of these two

groups can be bridged under certain conditions. Most importantly, democratic countries

and those exhibiting high vulnerability levels to climate change impacts show a significantly

higher propensity to adopt more moderate positions. This might give countries willing to

cooperate a hint whom to cooperate with. The next chapter exploring bargaining success

finds that again vulnerability helps countries to attain their goals, as does a certain level of

power. In contrast, countries emitting high amounts of greenhouse gases and those adopting

relatively extreme positions on average lose in the negotiations.

In the third chapter position coordination is investigated. The Exponential Random

Graph Models employed in this part of the dissertation show that countries with similar

characteristics exhibit an increased propensity to cooperate in the negotiations. In particular,

similar democracy and emission levels as well as membership in the same international

organizations and similar cultural backgrounds induce countries to bring their positions in

line. This finding is corroborated in the final chapter of the dissertation, which studies the

Alliance of Small Island States (AOSIS) coalition group. It is found that these highly cohesive

group of countries was for a relatively long time able to act in unity in the negotiations, and

thus to achieve success previously inconceivable for those otherwise insignificant developing

states. These countries are therefore role models for other small countries for how to achieve

negotiations goals, but also for constructive negotiation behavior.

7

Zusammenfassung

Die Klimaverhandlungen unter der Schirmherrschaft der United Nation Framework Conven-

tion on Climate Change (UNFCCC) sind die wichtigste diplomatische Buhne zur Losung

der Klimakrise. Die Entscheidungen, die wahrend dieser Meetings getroffen werden, sind

daher von ausserster Wichtigkeit fur die Zukunft der Menschheit, und aus diesem Grund

sind die Verhandlungen auch ein hochinteressantes Studienfeld. Diese Dissertation unter-

sucht verschiedene Elemente des Verhandlungsprozesses, etwa die Verhandlugnspositionen

der Lander, die Entstehung von Koalitionen, oder die bestimmenden Faktoren von Verhand-

lungserfolg. Durch die Zusammenschau der einzelnen Teile dieser Dissertation entsteht ein

Bild des Verhandlungsprozesses, welches Einblicke gewahrt wie der Stillstand der Klimaver-

handlugen uberwunden werden konnte.

Diese Dissertation ist Teil des Projekts ‘Negotiating Climate Change’. Alle Kapitel

machen Gebrauch von einem neuen Datensatz, der fur dieses Projekt erstellt wurde. Uber

den Zeitraum von etwa einem Jahr nahm das Projektteam an internationalen Konferenzen

in Bonn, Bangkok, Barcelona und Kopenhagen teil, und interviewte dabei zahlreiche Ver-

handlungsdelegierte. Kunfigte werden die so gewonnen Daten auch anderen Forschern zur

Verfugung stehen.

Das erste Kapitel uber Verhandlungspositionen zeigt den starken Kontrast zwischen

Annex 1 und nicht-Annex 1 Staaten, jedoch auch das der Graben zwischen den beiden

Landergruppen uberwunden werden kann. Insbesondere demokratische und klimatisch sen-

sible Staaten neigen in signifikantem Ausmass dazu, moderatere Verhandlungspositionen

einzunehmen. Dies kann Landern Hinweise darauf geben, mit wem es sich zu kooperieren

lohnt. Das nachsten Kapitel zeigt, dass Klimasensibilitat und Macht wichtige Faktoren fur

den Verhandlungserfolg eines Landes sind, wahrend hohe Treibhausgasemissionen und die

Einnahme extremer Verhandlugnspositionen abtraglich fur den Erfolg sind.

Das dritte Kapitel handelt von Koordination und Koalitionsbildung. Die Modelle in

diesem Teil der Dissertation zeigen, dass Lander mit ahnlichen Eigenschaften in den Kli-

maverhandlungen verstarkt miteinander kooperieren. Ahnliche Demokratie- und Emissions-

niveaus, sowie Mitgliedschaft in denselben Internationalen Organisationen und ahnliches

Kulturgut veranlassen Staaten dazu, ihre Positionen abzustimmen. Diese Erkenntnis findet

Bestatigung im letzten Kapitel der Dissertation, das einen genauen Blick auf die Koalitions-

gruppe der kleinen Inselstaaten (AOSIS) wirft. Dabei zeigt sich, dass diese sehr kohasive

Gruppe von Landern uber einen geraumen Zeitraum die Einheit wahren konnte, und dadurch

weit erfolgreicher in den Verhandlungen war, als fur eine Gruppe relativ unbedeutender

Kleinstaaten angenommen werden konnte. Diese Staaten sind daher Musterbeispiele fur

andere Kleinstaaten, aber auch Vorbilder fur konstruktives Verhandlungsverhalten.

8

Introduction

Despite the now widely recognized need to act swiftly to prevent dangerous levels of global

warming, real progress in the ongoing climate change negotiations is slow and cumbersome

at best. Particularly the Conference of the Parties (COP) 15 in Copenhagen, where a major

breakthrough had been expected, ended after two weeks of gridlock with a paltry three-

page document, the ‘Copenhagen Accord’ (see UNFCCC, 2009). Since then, the follow-up

conferences in Cancun and Durban have shown more promising signs. However, the process

is still painstakingly slow. At the core of the problem is a conflict between the developing

countries, also known as the non-Annex 1 countries in the climate change negotiations, and

the developed countries listed in the Annex 1 of the Kyoto Protocol (and hence called Annex

1 countries). The question is whether a solution to overcome this conflict is possible, and a

compromise to avoid global warming can be found.

To contribute to answering this question, this dissertation seeks to dissipate the negoti-

ation process into various parts, and investigates how countries formulate their negotiation

positions, the formation of cooperation and the role of coalition groups within the United Na-

tions Framework Convention on Climate Change (UNFCCC) negotiations, and ultimately

the bargaining success different players are able to attain separately. Yet first a general

framework for the analysis of the negotiation process needs to be established. As a starting

point I propose the likewise neorealist and institutionalist perspective that states are unitary

and rational actors (see e.g. Keohane, 1984; Waltz, 1979).

If states are indeed unitary and rational actors, the question arises how, according to

theory, they would/should most likely behave in the climate change negotiations. Given

that the climate system and greenhouse gas emission are global public goods (Stern, 2007),

the costs of polluting are not born by the emitter alone, while the benefits of a country’s

abatement efforts are commonly shared. If we consider rational choice models to reflect

reality, under such circumstances countries are expected not to limit their own emissions

sufficiently to combat global warming as the costs outweigh the benefits from an individual

perspective. Thus, there is no incentive for any rational individual, firm, or state to reduce

emissions enough such that the aggregate outcome is sufficient to prevent (or at least limit)

global warming. As a consequence, Stern (2008) concludes that these greenhouse gas emis-

sions are (negative) externalities, i.e. the emitter of the greenhouse gas only pays parts of its

price and transfers the rest of the costs to society. As in the climate change case the costs

are occurring on a global level, the consequential market failure is also global in its nature,

hence needs to be addressed there. As in such global settings there is no central authority to

impose rules to correct the market failure, the expected outcome is free-riding of the agents,

9

and too little action to combat the adverse effects of the externality.

This understanding of emissions as externalities is not limited to individuals. If this

rational choice approach is applied to countries and governments, we expect relatively little

action in the form of policies forcing their citizens (and enterprises) to reduce emissions

taken by individual states. No state alone can secure the global action needed to prevent

global warming. Hence, every state is expected to be reluctant to reduce its own emissions

as the costs would fall on its own citizens who would only gain a relatively small proportion

of the benefits they would have to pay for (Brennan, 2009). This, however, does not mean

that we should expect no action at all, yet instead of climate change mitigation, the more

likely course taken by countries might be unilateral self-protection, e.g. adequate measures

to adapt to future impacts triggered by global warming. Starting out overly simplistic, such

a tragedy of the commons situation (Hardin, 1968) can be formalized using a game theory

approach for only two players in a one shot game (Hopmann, 1996; Morrow, 1994; Ostrom,

1990). The matrix for this game, a classical prisoner’s dilemma, looks as shown in Table 1.

Table 1: Pay-off matrix of the Prisoner’s Dilemma

Party ACooperate Defect

Party BCooperate 1,1 -3,3

Defect 3,-3 -2,-2

The best outcome for both parties would be to cooperate. However, no matter whether

the other party opts to cooperate or not, each county has a dominant strategy to free ride and

thus not to implement strong mitigation efforts itself, or in other words to defect. (Dixit and

Nalebuff, 1991, p.13) note that “the jointly preferred outcome arises only when each [party]

chooses its individually worse strategy.” The most likely outcome, and a Nash-equilibrium, is

therefore the lower right box of the above diagram. From a common perspective, this outcome

is the least favorable one. As already mentioned, this game is of course oversimplified, as in

reality there are more than two players and the game is repeated in numerous negotiation

rounds. Furthermore, actions to prevent climate change should be seen as a cooperative

game, i.e. players are not making their decisions (entirely) independent from each other,

while the prisoner’s dilemma is a non-cooperative game. Regarding these aspects, Brennan

(2009, p.311) notes that in n-prisoner’s dilemma experiments with repeated rounds and

the chance to discuss and to punish, a proportion of the participants usually adopts the

cooperative strategy. However, the defectors are in almost all experiments in the majority,

and cooperation further declines in repeated rounds.

For climate change this implies that, although co-operation is possible and a large number

10

of players are involved, the dominant strategy of defecting is still expected to prevail. This

is in line with game theory’s prediction that existing dominant strategies will be played

even in co-operative game settings as well as with the repeatedly and increasingly voiced

claim that finding solutions to prevent global warming through global negotiation rounds

is difficult. Although this view is widely shared in the literature (see e.g. Cline, 1992;

Nordhaus and Boyer, 2000), others claim that at least some level of emission abatement

might still be expected under certain circumstances (Nordhaus, 2008, p.116-122). Ward

(1996), on the other hand, believes that in such repeated Prisoner’s Dilemma games some

form of cooperation is possible if countries play conditionally cooperative strategies (i.e. they

cooperate as long as other follow suit) and if the future is not discounted too heavily. Much

earlier Olson (1965:43-52), writing about collective action in general, argued that groups of

limited size in which defection will be noticed might, or equally well might not, be able to

provide a public good collectively, yet some form of group coordination is required to achieve

cooperation. And Carraro and Siniscalco (1993) believe, that in settings where free-riding

is possible, as in the climate change case, the formation of stable and profitable cooperative

agreements of subgroups (coalition) is possible (and necessary) to overcome the Prisoner’s

Dilemma.

The question arises whether there are other means of dealing with the global warming

crisis outside of the traditional negotiation setting. Barrett (2008, p.9-20) notes that there

are three distinct possibilities to overcome global free-rider problems: a) strong leadership of

one or more (preferably influential) parties in the hope that others follow suit, b) making a

country’s climate policy contingent on those of others, and c) a legally binding international

agreement. Yet for the climate change case Barrett (2008, p.12) believes that the first two

options are inadequate, and indeed no serious efforts to solve the problem in this fashion

have so far been undertaken, hence the reliance on international negotiation rounds aiming

to establish widely accepted, legally binding international treaties despite of the described

shortcomings. Evidently the incentives to agree to legally binding agreements are closely

related to the above shown payoff matrix and the Prisoner’s Dilemma.

To better understand the motivations of states to defect, it is necessary to introduce the

concept of Best Alternative to a Negotiated Agreement (BATNA). Hopmann (1996, p.56)

defines the BATNA as the “value of no agreement”, i.e. the payoff a country obtains if

the negotiations fail. Only if the value of an international climate treaty is higher than the

value of the BATNA are rational parties expected to accept the deal. Hence the value of

the BATNA is also known as the resistance point, as any negotiated outcome yielding a

lower payoff to a party than no agreement is expected to be opposed. One of the major

problems of the climate change negotiations is that, as shown above, defecting generally

11

yields higher payoffs than cooperation. In other words, the BATNA value is very hard to

overcome through negotiations. The free-riding problem is directly passed from the under-

lying commons problem to the climate change negotiations, and the dominant strategy not

to implement meaningful emission reduction policies to solve a global public good problem

translates to uncooperative behavior in the negotiations aiming at breaking the deadlock.

This short theoretical outline shows that reaching any meaningful agreement to tackle

global warming is difficult, and according to some authors names above very unlikely. How-

ever, as the Kyoto Protocol demonstrated, reaching an agreement can and should not be

ruled out. So what is it then that has changed since Kyoto? The Rio Declaration of 1992

(UN, 1992) states that

in view of the different contributions to global environmental degradation, states

have common but differentiated responsibilities. The developed countries ac-

knowledge the responsibility that they bear in the international pursuit to sus-

tainable development in view of the pressures their societies place on the global

environment and of the technologies and financial resources they command.

This ‘common but differentiated responsibilities’ principle is also stated several times in

the UNFCCC founding document and is dogmatically repeated again and again during the

climate change negotiations. It is also one of the major reasons why parties were able to

reach agreement in Kyoto. In 1997 the developed countries accepted their historic role as

CO2 emitters as well as the notion that developing countries could not be held liable for

global warming, nor would they be willing or able to pay. As no action was required from

developing countries to be in compliance with the Kyoto Protocol, they had no incentive to

reject the treaty; their resistance points could be overcome (it should be noted that there

exists a general commitment to adopt mitigation policies and to report on them, but no

targets; see Grubb and Depledge, 2001). Of course the story is not that simple. At the time

when Kyoto was signed, much less was known about global warming, and as it was only a

first step, individual country targets for the developed world were mostly not particularly

ambitious. Additionally, the Clean Development Mechanism (CDM) was introduced, offering

developing countries the prospect of attracting foreign investment, and developed countries

the opportunity to buy cheap credits abroad to offset domestic emissions.1 Common but

1 The CDM is one of three market mechanisms introduced in Kyoto, the others being emissions trad-ing within developed countries and Joint Implementation projects in economies in transition to marketeconomies. The main idea of the CDM is that projects to reduce CO2 emissions are implemented with fundsstemming from the developed world, the certified emission reductions (CERs) can then be sold on the car-bon market. With the thus generated funds people in developing countries can be compensated for eventuallosses due to the projects. The major critique of the CDM is that Western developers mainly focus on the“low hanging fruits”, i.e. they implement projects where emission reductions are cheapest. Economically,

12

differentiated responsibilities had helped lowering the BATNA value for both developed and

developing countries in 1997 and made a positive negotiation outcome possible. Hence, the

protocol was a compromise which offered both sides enough to accept the deal. Non-Annex

1 countries got the promise that Annex 1 countries would start cutting emissions, and could

even hope to generate some revenue in the process. Annex 1 members, on the other hand,

had ensured that the costs of compliance would be bearable. All sides were thus able to

accept the deal.

Yet the 1997 Kyoto agreement did not eradicate the commons problem. The United

States infamously refused to sign up, and others such as Canada, Japan, or New Zealand,

are failing to reach their targets. Even within the European block there are a host of

countries not meeting their obligations (most notably Spain and Italy), and the EU (the

old 15) as a whole will only be able to reach its goals as a group, because some countries

are over compliant (e.g. Germany and the UK), a distinct advantage for the EU (see e.g.

Olivier et al., 2011). Consequently, the Kyoto Protocol is seen by many, including many

developing countries, as a failure. In the negotiations, the EU had to repeatedly fend off

accusations that the bloc would be unable to reach its targets. These failures to comply

with the Kyoto Protocol, and the painfully slow progress of the negotiations led to a climate

of mistrust between the North and the South, which reached a climax in Copenhagen and

reached its particular peak after the so-called ‘Danish text’ had leaked to the press (Dubash,

2009). This lead to accusations that the West wanted to kill the Kyoto Protocol and to

sideline the developing world (see Vidal, 2009). But also powerful developing countries were

accused for having made coordinated efforts to stall the negotiations, most prominently

China and India (Adam and Randerson, 2010). Benedick (1998), the ambassador for the

US for the negotiations dealing with the ozone hole which let to the Montreal Protocol,

demonstrated in his book how important mutual trust and a positive attitude were for

reaching an agreement in Montreal. As demonstrated, in the climate change negotiations

instead of trust mutual accusations were the rule. In addition, the delegates were unable

to abate the stark differences between the negotiation positions of the two sides. This

combination ruined the conference in Copenhagen.

To better understand the difference between the outcomes of Kyoto and Copenhagen it

must be noted that the world has substantially changed since 1997. Developing countries,

now major and relevant players on the international level, are playing a self-confident role

in expressing their national self-interest. Developed countries, on the other hand, see the

this makes sense, as to obtain efficiency emissions should be reduced where it is cheapest (indeed this is thebasic idea behind CDM). Yet it also means that should developing countries ever take up emission reduc-tion commitments themselves, the lowest cost opportunities will already be gone, exploited by developedcountries. For a more detailed analysis of the CDM and other market mechanisms see Hepburn (2007).

13

time fit to shift some of the burden to reduce emissions to richer developing countries, when

in Rio and Kyoto they were still prepared to shoulder the burden alone. Kyoto was able to

overcome the prisoner’s dilemma due to a clear division between developed and developing

countries. As some of these countries such as China, India, and Brazil have since grown

to challenge the West economically and politically, the prisoner’s dilemma and thus the

dominant strategy to defect is again setting the rules of the game. This makes finding a

negotiated agreement much more challenging than back in 1997.

However, despite all the difficulties and shortcomings, what the Kyoto Protocol never-

theless demonstrates is that some form of cooperation is possible, as theorized by Olson

(1965), Ward (1996), or (Nordhaus, 2008), but also that such a treaty must seem beneficial

to all involved parties. Rational choice theory provides a useful framework to understand the

major issues and strategic choices countries face when engaging in the international climate

change negotiations. Each country’s preference is that other countries cut their emissions by

as much as possible, ideally without having to accept binding abatement levels itself. While

in Kyoto negotiators were able to overcome this problem, in the Copenhagen the negotiation

positions of the parties were too far apart and hardened to find a compromise. Hence, one

important part of this dissertation is to take a closer look at bargaining positions of countries

and what drives them.

More generally, in this dissertation I seek to explore the behavior of countries in the

negotiations. A more detailed analysis of various elements of the bargaining process might

help to better understand the problems of finding an agreement in the climate change case,

but also contribute to a better understanding of bargaining situations between states in

general. Such essential ingredients for understanding negotiations are the afore mentioned

position formation, the determinants of success in the negotiations, and the analysis of

existing coalitions and coalition building. These are the topics of the various chapters of this

dissertation.

First, I look into the formation of national negotiation positions of parties. As Moravcsik

(1997) notes, state preferences matter in determining the outcome of international negotia-

tions. Not only are they an important factor in explaining the positioning behavior of nations

during negotiations, understanding preferences is also a necessary condition for defining the

bargaining space, and parties’ BATNAs. For example, some scholars believe the reason for

the failure of the Copenhagen summit can be attributed to fundamentally opposing interests

of the involved parties (see e.g. Dubash, 2009; Guerin and Wemaere, 2009).

Preferences cannot be easily determined. As Frieden (1999) points out, preferences are

intermingled with strategic behavior and the resulting (observed) positions thus represent

a mixture of these two sets of determinants. Hence, the purpose of the first part of this

14

dissertation is to explain the positioning behavior of countries, more specifically whether

they are mainly determined by sincere preferences, or whether other factors, such as strate-

gic considerations, also play a role when countries present their negotiation positions in the

climate change negotiations. In the paper which constitutes the first part of this disserta-

tion, written jointly with Stefanie Bailer (currently under review), an array of hypotheses

regarding positioning behavior is proposed and then tested empirically.

Possessing knowledge about negotiations positions and the bargaining space is a neces-

sary condition for the second part of the dissertation, which is interested in determining

the bargaining success or failure of parties and the respective reasons. In order to measure

success, an actual negotiated agreement is required. Unfortunately, at this point the negoti-

ations are far from over, and many more negotiation round might likely follow before either

a global, legally binding treaty is signed, or the negotiations break down. Therefore, for the

purpose of measuring success I employ the Cancun Agreements, against which I compare

the negotiation positions. A relatively wide range of issues are covered in climate change

negotiations. Following Hinich and Munger (1997), I construct the bargaining space for eight

issues areas and measure the success of countries a) across these eight issue dimensions of

the negotiations, and b) in a single combined measure of success which accounts for salience

values of the eight proposed issues. This second part of the dissertation, already published

in the journal Climate Policy (see Weiler, 2012), thus seeks to shed light on the question of

who is winning in the UNFCCC, which are the sources of power necessary to achieve the

negotiations goals, and ultimately what this means in a setting where countries are expected

to play a dominant strategy to defect.

As mentioned above, in order to reach an agreement against which individual positions

can be evaluated, some form of coordination among actors is needed in multi-party settings.

This is the topic of the third part of the dissertation, which employs a network of cooperative

ties between parties in the climate change negotiations to answer the question why countries

coordinate their positions during the bargaining process. More specifically, the network is

constructed from statements countries made jointly by two or more parties during a two-year

period prior to the COP 15 in Copenhagen. The thus generated network then serves as the

dependent variable for a number of exponential random graph models (ERGMs). These

models on the one hand allow the researcher to test both main and homophily effects on

the network, e.g. if countries with certain characteristics tend to be more active than others

(main effect) or whether they have an increased likelihood to cooperate with each other

(homophily effect). ERGMs allow scholars to define dyadic dependence models through

the inclusion of additional parameters. Without such dependence specifications, the models

imply the underlying assumption that cooperation between two nodes (in my case countries)

15

is independent from already existing ties in the network. This for social settings rather

unrealistic assumption can be overcome for example through the inclusion of triad closure

parameters, i.e. the premise that two countries with ties to the same country are more likely

to cooperate as well (a friend of a friend is a friend). Thus, the third part of the dissertation

sheds light on the inner workings of the negotiations and how countries aim at increasing

their influence through reciprocal assistance and support. This coordinating activity also

reduces the complexity of the bargaining process, as the number of opposing positions (and

one could argue of players) is thereby reduced. In addition, this part of the dissertation

employs a hitherto new approach to model international relations and is therefore a twofold

contribution to the field.

In the last part of the dissertation, a paper already published in Climate Policy (see

Betzold et al., 2012), some of the findings from the previous chapters are critically assed in

a case study on the negotiation behavior and success of the Alliance of Small Island States

(AOSIS). Starting from the premise that AOSIS was remarkably successful for a group of

otherwise relatively powerless states during the early stages of the negotiation process, this

part of the dissertation sets out to explore why the success of the alliance has been decreasing

over recent years. Success is therefore tracked over three distinct negotiation periods, and

it is found that group activity has declined as time progressed, and consensus on a host of

issues was increasingly hard to find with the coalition group. Still, overall AOSIS is still one

of the most uniform coalition groups of the UNFCCC and the level of coordination is high,

which is the reason why small island states are still able to play a key role in the negotiations

and to achieve at least some success.

In all four parts of this disseration a newly created dataset is employed. Within the scope

of the project ‘Negotiating Climate Change’, a comprehensive dataset regarding countries’

positions, negotiations strategies, stakeholder influence, and issue saliency was collected. To

achieve this, our team relied on three different sources of information. First, we interviewed

delegation members during international conferences, e.g. in Bonn, Bangkok, and Copen-

hagen. Through the interviews we collected the first set of data on party positions, but

also on strategies and stakeholders. Second, in order to check the validity of the interview

data, but also to obtain positioning data for a bigger number of countries, we hand-coded all

country submissions to the UNFCCC made between the meetings in Bali (December 2007)

and Copenhagen.2 A pretested coding-scheme guided us through this process, and led to a

second dataset including a higher number of countries, but fewer issues. The correlation of

issues included in both datasets is generelly fairly high. Third, we hand-coded the Earth

2 All submissions from Bali onward can be downloaded from the UNFCCC website athttp://unfccc.int/documentation/submissions from parties/items/5900.php/

16

Negotiation Bulletins (ENBs) over the same time horizon as the submissions.3 The ENBs

report almost every statement made during these negotiations, the topic of the statement,

and who supports or cooperates (joint statements) with whom. From this last dataset we

generated countries’ salience values for the negotiation issues (for which we already had col-

lected position data). The data further served to create a network of cooperative ties, which

is used as the dependent variable in one of the papers in this dissertation. Table 2 provides

an overview of the obtained data.

Scholars have discussed whether negotiation analysis should better be carried out in a

qualitative or quantitative way. In a paper on that topic, the renowned negotiation expert

(Hopmann, 2002) favors the former, yet he concedes that quantitative studies also have their

advantages. For example, such studies tend to be less prone to suffer from the problems of

subjectivity and bias. In addition, quantitative studies require a precise operationalization

of the variables in question, which forces the researcher to be much more explicit about

the theoretical implications and assumptions than with a quantitative research designs. In

addition to these advantages, various researchers have already demonstrated that collect-

ing quantitative data via interviews (see e.g. Bailer, 2004; Thomson and Stokman, 2006) or

content analysis of appropriate negotiation documents (see e.g. Bueno de Mesquita and Stok-

man, 1994; Hug and Konig, 2002; Konig, 1997) are valuable methods for gaining information

about negotiations.

Interviewing country delegates has the advantage of being able to obtain information

which cannot be easily extracted from documents. For example, negotiation strategies such

as vocalizing demands, making promises or threats, or proposing mutual concessions are

usually not captured in written negotiation documents. Also the ENBs only capture such

strategic behavior imperfectly. For this reason, we asked delegates a total of 10 questions

regarding their party’s strategies in the climate change negotiations. Also for negotiation

positions it is easier and more reliable to directly ask delegates who possess the relevant

knowledge, as the information the documents might not be detailed enough for the re-

searcher to make an informed judgment of where the party stands on an issue. The obvious

disadvantage of such an interview approach, on the other hand, is that a number of delega-

tions will not answer the interview request, and that the delegate selected is not necessarily

representative of her delegation. Thus, there is the problem of possible self-selection bias

and a dataset in which a host of crucial negotiation parties are missing.4

To overcome these shortcoming, we also hand-coded the above mentioned country sub-

3 All ENBs can be found at http://www.iisd.ca/process/climate atm.htm4 For example, the island state of Tuvalu, which received a lot of attention during he negotiation, refused

to be interviewed on the grounds that such projects as ours possibly could further derail the negotiationprocess. Other countries such as Isreal or Italy did not even give a reason, but simply refused to participate.

17

Table 2: Summary of data obtained for the project ‘Negotiating Climate Change’

Source Obtained information DetailsInterview data Negotiation positions Country positions on stringency tar-

gets for Annex 1, actions for non-Annex1 parties, differentiation of commit-ments, use of market mechanisms, mit-igation finance and allocation, adapta-tion finance and allocation, and report-ing mechanisms.

Negotiation strategies Three soft strategies (express under-standing for others, propose new so-lutions, exchange of concessions) andseven hard strategies (declaration notto change positions, directly criticizeothers, ignore others demands, hidereal objective, demand concessions, is-sue threats, issue promises).

Stakeholder influence Influence on negotiation positions ofvarious ministries, the national parlia-ment, the domestic public opinion, var-ious domestic and international stake-holder group from NGOs to industrylobbies, the media, and negotiationpartners.

Success ratings Success measured in two ways, a) thedistance of stated negotiation positionsfrom outcomes in Cancun, and b) eachinterviewee was asked to assess successrating of 5 other countries.

Submissions data Negotiation positions Country positions on stringency tar-gets for Annex 1, actions for non-Annex1 parties,market mechanisms, adapta-tion finance, mitigation finance, and re-porting mechanisms.

ENB data Salience ratings Count of times topics were mentionedby a negotiating party used to con-struct saliency scores.

Cooperation Joint statements and mutual supportregistered. These data were used toconstruct a network of coordinated ties.

18

mission with a pretested coding scheme. Thus, we were able to obtain negotiation positions

for a much larger number of counties, yet only for a smaller number of issues (6 instead of

12). And only 3 could be measured at the exactly same scale as in the interviews. In a

recent paper employing computer assisted content analysis, Genovese (2012) takes a critical

view of our data collection effort. For example, she correctly points out that submissions

are voluntary contribution to the negotiations, and proposes instead to rely on the National

Communications (NCs) which countries are obliged to prepare. However, although I agree

with the premise that such obligatory documents are preferable over voluntary texts as the

submissions are, I reject the proposed NCs on the grounds that these documents have an

entirely different purpose and are not intended to present negotiation positions. In contrast,

the three issues for which we obtained interview and hand-coded data, emission reduction

targets for Annex 1 countries, adaptation finance, and mitigation finance, show relatively

high correlation coefficients of 0.92, 0.70, and 0.69, respectively. This in my view justifies our

choice of documents for the content analysis, and is an indication for the reliability of both

the interview but also the hand-coded submission data. Regarding validity, I believe that

the interview data have the higher degree of internal validity, as the information is obtained

directly from involved negotiators and less assumptions have to be made. However, external

validity is most likely higher for the submission data, as this dataset contains all crucial

players and not only those self-selecting to be interviewed.

In the coming pages I make extensive use of these data. I hope the proposed models

and the results of the four papers are interesting for the reader, and that my research is

able to make a small contribution to the already substantial knowledge we possess about

negotiations.

Acknowledgements: I thank Kirsty and my parents for their continuous support. I

especially thank Stefanie Bailer for her guidance, advice, and intellectual support over the

past three years. Additional thanks goes to Katja and Axel Michaelowa, David Armstrong,

Joyeeta Gupta, John Odell, Paula Castro, Simone Gunther, Tamaki Ohmura, Lena Hornlein

and David Willumsen. Financial support from the Swiss Network for International Studies

for funding my PhD, and the Swiss National Science Foundation for two summer school

grants is gratefully acknowledged.

19

References

Adam, D. and J. Randerson (May 2010). Secret Copenhagen record-ing reveals resistance form China and India. The Guardian.http://www.guardian.co.uk/environment/2010/may/07/secret-copenhagen-talks-climate-recording.

Bailer, S. (2004). Bargaining success in the European Union: The impact of exogenous andendogenous power resources. European Union Politics 5 (1), 99–123.

Barrett, S. (2008). Negotiation strategies for a post-Kyoto regime. BC Journal SpecialVolume, Fall 2008, 9–20.

Benedick, R. (1998). Ozone Diplomacy. New Directions in Safeguarding the Planet. Cam-bridge, MA: Harvard University Press.

Betzold, C., P. Castro, and F. Weiler (2012). AOSIS in the UNFCCC negotiations: Fromunity to fragmentation? Climate Policy 12 (5), 591–613.

Brennan, G. (2009). Climate change: A rational choice politics view. The Australian Journalof Agriculture and Resource Economics 53, 309–326.

Bueno de Mesquita, B. and F. N. Stokman (1994). European Community Decision Making:Models, Applications, and Comparisons. New Haven and London: Yale University Press.

Carraro, C. and D. Siniscalco (1993). Strategies for the international protection of theenvironment. Journal of Public Economics 52, 309–328.

Cline, W. (1992). The Economics of Global Warming. Washington, D.C.: Institute forInternational Economics.

Dixit, A. K. and B. J. Nalebuff (1991). Thinking Strategically. The Competititve Edge inBusiness, Politics, and Everyday Life. New York, London: W. W. Norton & Company.

Dubash, N. (2009). Copenhagen: Climate of mistrust. Economic and PoliticalWeekly 44 (52), 8–11.

Genovese, F. (2012). States’ interest and bargaining positions at the UN climate changenegotiations. Paper presented at the EPSA 2nd General Conference, Berlin, June 2012.

Grubb, M. and J. Depledge (2001). The seven myths of Kyoto. Climate Policy 1 (2), 269–272.

Guerin, E. and M. Wemaere (2009). The Copenhagen Accord: What hap-pened? Is it a good deal? Who wins and who loses? What isnext? http://www.iddri.org/Publications/The-Copenhagen-Accord-What-happened-Is-it-a-good-deal-Who-wins-and-who-loses-What-is-next.

Hardin, G. (1968). The tragedy of the commons. Science 162 (3859), 1243–1248.

20

Hepburn, C. (2007). Carbon trading: a review of the kyoto mechanisms. Annual Review ofEnvironment and Resources 32, 375–393.

Hinich, M. and M. Munger (1997). Analytical Politics. Cambridge, New York, Melbourne:Cambridge University Press.

Hopmann, P. T. (1996). The Negotiation Process and the Resolution of International Con-flicts. Columbia: University of South Carolina Press.

Hopmann, P. T. (2002). Negotiating data: Reflections on the qualitative and quantitativeanalysis of negotiation processes. International Negotiation 7, 67–85.

Hug, S. and T. Konig (2002). In view of ratification: Governmental preferences and domes-tic constraints at the Amsterdam Intergovernmental Conference. International Organiza-tion 56 (02), 447–476.

Keohane, R. O. (1984). After Hegemony: Cooperation and Discord in the World PoliticalEconomy. Princeton: Princeton Universtiy Press.

Konig, T. (1997). Europa auf dem Weg zum Mehrheitssystem. Grunde und KonsequenzenNationaler und Parlamentarischer Integration. Opladen: Westdeutscher Verlag.

Moravcsik, A. (1997). Taking preferences seriously: A liberal theory of international politics.International Organization 51 (4), 513–553.

Morrow, J. (1994). Game Theory for Political Scientists. Princeton, NJ: Princeton UniversityPress.

Nordhaus, W. (2008). A Question of Balance. New Haven, CT: Yale Univesity Press.

Nordhaus, W. and J. Boyer (2000). Warming the world: economic models of global warming.Cambridge, MA: MIT Press.

Olivier, J., G. Jansen, J. Peters, and J. Wilson (2011). Long-term trend in CO2 emissions.2011 report. http://www.pbl.nl/sites/default/files/cms/publicaties/C02

Olson, M. (1965). The Logic of Collective Action. Public Goods and the Theory of Groups.Cambridge, London: Harvard University Press.

Ostrom, E. (1990). Governing the commons. The evoluton of institutions for collective action.Cambridge: Cambridge University Press.

Stern, N. (2007). The Economics of Climate Change: The Stern Review. Cambridge: Cam-bridge University Press.

Stern, N. (2008). The economics of climate change. American Economic Review 98 (2), 1–37.

Thomson, R. and F. N. Stokman (2006). Research design: Measuring actors positions,salience and capabilities. In R. Thomson, F. N. Stokman, C. H. Achen, and T. Konig(Eds.), The European Union Decides, pp. 54–85. Cambridge, UK: Cambridge UniversityPres.

21

UN (1992). Rio Declaration on Environment and Development. Rio de Jenairo, Brazil.

UNFCCC (2009). Copenhagen Accord. http://unfccc.int/resource/docs/2009/cop15/eng/l07.pdf.

Vidal, J. (December 2009). Copenhagen climate summit in disarray after ‘Danish text’leak. The Guardian. http://www.guardian.co.uk/environment/2009/dec/08/copenhagen-climate-summit-disarray-danish-text.

Waltz, K. N. (1979). Theory of International Politics. New York: Random House.

Ward, H. (1996). Game theory and the politics of global warming: The state of play andbeyond. Political Studies 44 (5), 850–871.

Weiler, F. (2012). Preference attainment: Determinants of bargaining success in the climatechange negotiations. Climate Policy 12 (5), 552–574.

22

Climate Change Negotiations, NegotiationPositions and Domestic Structures

Florian Weiler and Stefanie Bailer

Center for Comparative and International StudiesFederal Institute of Technology, Zurich

Paper submitted to International Studies Quarterly

Abstract

International negotiations analyses such as the UN debates on climate change usuallyconsider negotiation positions of governments as given although the determinants ofthese are not yet sufficiently investigated. The positions reflect not only the actual -often economic - interests of states, but also the strategic considerations of a govern-ment regarding the domestic interests it wants to respect. The choice of negotiationpositions determines whether a government opts for a more cooperative bargaining be-havior or for a more confrontational approach that risks a negotiation breakdown. Thischoice will be influenced by the question whether countries profit or suffer from climatechange. We investigate this question with a novel dataset on the current UNFCCC ne-gotiations, in which the positions of all participating governments were collected withhand-coded protocols and expert interviews with negotiators. Our multivariate dataanalysis shows that negotiation positions are influenced by climate change vulnerabilityand by domestic stakeholders.

Keywords: bargaining positions, positioning behavior, climate change, negotiations,UNFCCC

1 Introduction

When the Copenhagen Summit of December 2009 was concluded with a disappointing and

meager negotiation outcome that stood in strong contrast with the high hopes which had

preceded this important meeting, policy analysts and journalists alike offered various ex-

planations for this. Simplistic accounts blamed China for being too selfish, whereas more

complex analyses pointed out that the interests of major players were basically too far apart

for being reconciled in a common treaty. This clash of interests combined with irreconcil-

able negotiation positions, chiefly between the US and China, are probably the prevalent

explanations for the failure of Copenhagen. The explanations of these negotiation positions

is our motivation to investigate whether economic, domestic and strategic considerations

are responsible for the choice of a certain negotiation position by states in climate change

negotiations.

According to arguments pertaining to the free-rider problem that climate change is

plagued with (Barrett and Stavins, 2002; Brennan, 2009), all countries prefer action on the

part of others to reduce emissions over similar policies adopted at the domestic level. If all

countries agreed to reduce their emissions, the result might be that the prisoner’s dilemma

inherent to reducing CO2-emissions (Hopmann, 1996; Ostrom, 1990) would be overcome.

However, due to the particular structure of the climate change negotiations and the division

of states into Annex I and non-Annex I countries, this scenario is unlikely. Since the adop-

tion of the Kyoto Protocol (KP) in 1997, the community of states is divided into two groups

when negotiating climate change: On the one side are the so-called Annex 1 countries listed

in the appendix of the KP, on the other the non-Annex 1 countries. This division, some-

times referred to as the “Kyoto Firewall” (Bodansky, 2010, p.233-234), allows the developing

countries (non-Annex 1) to play their dominant strategy to free-ride (Barrett and Stavins,

2002; Brennan, 2009), without having to fear retaliating measures from Annex 1 countries.

The rationality behind the decision not to ask developing countries to take up binding

emission reduction commitments was (and still is) the historic fact that the developed world

is overwhelmingly responsible for the increased CO2 content in the atmosphere (den Elzen

et al., 2005; Srinivasan et al., 2008). But the world has changed dramatically since 1997.

For example, Chinese emissions have been growing enormously ever since, making China

the biggest emitter of greenhouse gas emissions in the world today. Some Annex 1 coun-

tries, among others the US, have therefore called for an abolition of the Kyoto Protocol and

for a different treaty design, partially because these countries fear that domestic firms sub-

ject to strong emission regulations might shift production, and therefore jobs, to developing

countries where they face no emission reduction obligations. This emission leakage, so the

24

argument goes, would not only harm the developed economies, but also render an interna-

tional climate change agreement inefficient (Babiker, 2005; Blanford et al., 2008), and in

turn lower developed countries’ eagerness to promote an ambitious international agreement.

Non-Annex 1 countries, on the other hand, have an incentive to overstate their demands

during the climate change negotiations, as under Kyoto protocol rules, the costs have to be

borne by Annex 1 parties. This difference between Annex 1 and non-Annex 1 countries,

induced by the treaty design, is crucial to understanding positioning behavior in the climate

change negotiations. Starting from this central premise, this paper sets out to analyze and

understand countries’ negotiation positions in the UNFCCC negotiations.

So far, several studies have investigated the influencing factors on environmental policy or

environmental output (Bernauer and Koubi, 2009; Fredriksson and Gaston, 2000; Holzinger

and Sommerer, 2011; Neumayer, 2002) and came to different conclusions concerning the

impact of both exogenous, often economic, and endogenous, domestic factors. Whereas these

studies mostly have environmental output variables as their subject of analysis (e.g. carbon

dioxide emissions or the number of environmental treaties signed), we focus on the choice

of negotiation position in environmental negotiations in this paper. Analyzing negotiation

positions requires an examination of governments’ strategic and diplomatic considerations,

which is a so far somewhat neglected aspect in the study of environmental negotiations.

With this paper we therefore want to contribute to the filling of this research gap.

Based on insights of game theory (e.g. Morrow, 1994; Cline, 1992; Nordhaus and Boyer,

2000; Ward, 1996; Barrett, 2003) and the already mentioned free-rider problem, we expect

states to choose a negotiation position by which the costs of reducing emissions are shifted

to others. However, we do not believe that the desire to shift costs towards others is the

only factor influencing their choice. Instead, we expect additional factors to play a crucial

role also. We propose that governments have, on the one hand, a certain leeway regarding

the choice of negotiation positions, depending on the preferences of the parties in power

and of the electorate. On the other hand, such positions are influenced by domestic stake-

holders, whose leverage in turn depends on a country’s domestic structure and institutions

(Rogowski, 1999). The notion that domestic preferences matter in international negotiations

is not new. Already fifty years ago, Schelling (1960) noted that governments could increase

negotiation success if their win-sets were limited by domestic constraints, which he called

the “paradox of weakness”. If government A could credibly show that deviations from its

preferred position were limited by the home audience, this alleged weakness might force

government B to accept an outcome closer to the one preferred by party A (Fearon, 1997;

Putnam, 1988; Schultz, 2001). According to this theory, when choosing positions in inter-

national negotiations, governments thus play two games the international and the domestic

25

Figure 1: Difference between sincere and strategic preferences on a theoretical issue dimen-sion

one - with a single move, hence the associated term ‘two-level games’. (Kroll and Shogren,

2008). Although McLean and Stone (2012) find evidence that the domestic level was not

crucial for determining the ratification process of the Kyoto Protocol, we claim that domestic

interests in the climate change negotiations influence the negotiation behavior particularly of

democratic governments during the climate negotiations. Thus we investigate various factors

that influence the negotiation position a country eventually adopts: government preferences,

domestic institutions, the preferences of domestic interest groups, and diplomatic/strategic

considerations. The negotiation positions which we investigate do not only encompass the

sincere interests of a state but also the strategic considerations of a government which can

decide which domestic actor it wants to listen to (see Figure 1 for a graphical representation).

In the first part of the paper we discuss theoretical explanations of a country’s positioning

behavior in the UNFCCC negotiations and derive corresponding hypotheses. Next we outline

our methodological approach and describe the data. We then present tests of the proposed

hypotheses using multivariate analyses and conclude by suggesting further paths for research.

2 Determining factors of negotiation positions

The dependent variable of this study, as mentioned above, refers to the negotiation posi-

tions chosen by governments in the UNFCCC climate negotiations on various issues. These

positions are determined by national economic interests, the power and treaty constellation

at the international level, domestic stakeholders, and the institutions which mediate their

influence. We distinguish negotiation positions according to how “cooperative” they are,

a position being deemed increasingly cooperative the closer it is to the opinion center of a

negotiation issue. Proximity to the negotiation counterparts’ position facilitates agreement

26

and fosters cooperation.

Crucial to our understanding of a cooperative negotiation position is the distinction

between Annex 1 and non-Annex 1 countries mentioned above. In general we expect countries

of the non-Annex 1 group to make significantly higher demands regarding the issues analyzed

in this paper (emission reduction targets and mitigation finance), than the former group is

willing to offer. That this disparity is indeed present in the data is indicated by Figure 2a

and 2b depicting the two issues analyzed in this paper (described in more detail below).

As can be seen, Annex 1 countries tend to adopt less generous positions regarding both

emission reduction targets and mitigation finance than demanded by non-Annex 1 countries.

Cooperative behavior thus takes on the opposite meaning for each of the two groups of

countries. While for Annex 1 countries increased cooperation means offering more and

adopting positions further to the right on our issue scale, non-Annex 1 countries act more

cooperatively when they make less dramatic demands and thus position themselves lower on

the suggested scale. This contrast between Annex 1 and non-Annex 1 countries induced by

the treaty design has to be considered in all hypotheses presented below.

Figure 2: Graphical representation of the two dependent variables

25 30 35 40 45 50

Figure 2a Negotiaton positions for Annex 1 reduction targets

●●

●●

●

●●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

● ●

●

●

●

●

●●

●

●

●●

●

●

●

●●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●●

●

●●

●

●

●

●●

●●

●

●

●

●

●

●

●

●

0 20 40 60 80 100

Figure 2b Negotiation positions for mitigation finance

●

●

●

●●

●

●

●

●

●

●

●

●

●

●

●●

●●

●

●

●

●●

●●

●

●●●

●

●

●● ●

●

●●

●

●●

●●

●●

●

●●

●●

● ●

●

●

●

●

●

●

●

●

●

●●●

●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●● ●

●

●

●

●

●

●

●

●

● ●●

●

●

●

●

●

●●

●

●

●

●●

●

●

●

●

●

●

●●●●

●

●

●●

●

●●

●

●

●

●

●

●

●

●

●

●

●●●

●

●

●

●

●

Annex1non−Annex 1

Reduction targetsin percent ofAnnex 1 total

0 − Voluntary contributions50 − Taxation & market mechanisms100 − mandatory contributions

Starting out from this contrast, we derive hypotheses regarding cooperation, i.e. we

expect certain characteristics of countries to induce states of one group to play a less egoistic

game by moving closer to the (mean) positions of states in the other group, which moves the

win-sets of negotiation parties closer to each other and facilitates agreement. The Annex

1/non-Annex 1 difference is therefore crucial to this paper and deserves to be reiterated:

27

since the former group tends to adopt positions on the lower end of the scale of the issue

dimensions presented in Figure 2, more cooperative bargaining behavior is congruent with

positions located on the higher end of the spectrum. For non-Annex 1 countries the expected

effect is just the opposite, i.e. increased cooperation leads to a shift of positions to the lower

end of the issue dimensions scale. If a proposed factor is hypothesized to increase cooperation,

this means that we expect the coefficient to be positive for Annex 1 countries and negative

for the non-Annex 1 group. To make this distinction clear, we propose separate hypotheses

for the two groups and in each case clearly state the direction of the expected effect.

2.1 Economic and political variables

First we consider vulnerability to climate change impacts. When analyzing international

negotiations on acid rain and stratospheric ozone, (Sprinz and Vaahtoranta, 1994) found

that countries consider inter alia their ecological vulnerability when choosing policies and

positions on global environmental issues. Countries strongly affected by climate change are

more dependent on an international agreement that tackles/addresses this particular issue.

Thus, we expect them to be more willing to cooperate and more interested in finding a

compromise during the negotiations than their less vulnerable counterparts.

H1a: Non-Annex 1 countries that are more vulnerable to climate change impacts

are expected to choose more cooperative negotiation positions (negative expected

effect).

H1b: Annex 1 countries that are more vulnerable to climate change impacts

are expected to choose more cooperative negotiation positions (negative expected

effect).

Next we turn to a country’s power position in the international system. Power allows a

country to influence the outcome of international negotiations (Morgenthau, 1948). Grundig

(2006) shows that power-based explanations must indeed be considered to explain interna-

tional cooperation, particularly in the climate change case. Defining power in climate change

negotiations as the ability to affect global emissions, both economic power and overall emis-

sions determine how important and thus powerful a state is.

In the case of Annex 1 countries, big powers in terms of overall emissions lose more,

relative to others, when agreement on rigorous restrictions on CO2 emissions is reached.

Imagine two otherwise similar countries in terms of development, vulnerability to climate

change, etc., with one being much more reliant on emission intensive coal for energy produc-

tion than the other. This state will, on the one hand, have a higher impact on the climate

28

change talks as its policy is more relevant for tackling climate change. A more ambitious

climate change agreement, on the other hand, means a big challenge for such a country’s

economy when it decreases its dependence on fossil fuels, as this is costly and fraught with

uncertainties. Similarly, the bigger the economic power of a state, the more it is under

current UNFCCC rules expected to reduce emissions domestically as well as to contribute

to international funds for the support of poorer countries. For non-Annex 1 countries the

story is similar. Although powerful developing nations are not (yet) obliged to adopt binding

emission reduction targets, they are increasingly under pressure to show responsibility and

to define reduction targets as well. Thus, these countries cannot simply demand higher tar-

gets of developed countries without being affected themselves. Power, particularly economic

power, means more responsibility and more burden sharing, on the basis of which we expect

the effect of economic power whether in Annex I or not to motivate countries to be less

cooperative.

H2: The more powerful countries are, the less cooperative their behavior in in-

ternational climate change negotiations is expected to be (negative expected effect

for both groups).

2.2 Domestic variables

The two-level game theory states that domestic interest groups do have an impact on coun-

tries’ positioning decision in international bargaining situations such as global climate change

negotiations (Putnam, 1988). However, societies are not uniform structures.

With regard to representation it can be argued that more democratic forms of government

are more likely to take positions that reflect an aim of maximizing general welfare, while

less democratic states are more prone to favor particular interests. (Moravcsik, 1997), for

example, believes, that if

most individuals and groups in society, while acquisitive, tend to be risk-averse,

the more unbiased [and thus democratic] the range of domestic groups repre-

sented, the less likely they will support policies that impose high net costs or

risks on a broad range of social actors. (p. 531)

Thus, the median voter’s position is expected to prevail in democracies (Hinich and Munger,

1997). As an example for this democratic mechanism, Garrett and Lange (1986) show

that more democratic states are less likely to protect industries from foreign competition, if

competition is to the advantage of the general public.

29

Previous research has also indicated that democracies show stronger environmental com-

mitment though not necessarily better environmental outcomes than non-democracies (Neu-

mayer, 2002). Since voters in democracies are better informed and have the opportunity to

express their concerns more freely, their chances of exerting pressure on politicians are much

higher. In authoritarian states, on the other hand, such interests can be silenced more easily

by the authority. This linkage was also identified by Fredriksson and Gaston (2000), who

confirm the positive effect of civil liberties on the probability of states signing environmental

treaties.

Assuming that median voters’ positions are not congruent with special interest groups’

preferences, we expect higher democratic ratings to be associated with a higher propensity

on the part of governments to favor international cooperation as a means of dealing with

climate change. To use the language of Moravcsik (1997), risk-averse voters want to avoid

the high risk and net costs of a warming climate, and thus favor negotiation positions which

increase the chances of the negotiations ending in agreement. Thus, as in the vulnerability

case above, Annex 1 countries with higher democracy ratings should offer more, while their

non-Annex 1 counterparts are expected to demand less.

H3a: Non-Annex 1 countries with higher democracy ratings are expected to as-

sume more cooperative negotiation positions (negative expected effect).

H3b: Annex 1 countries with higher democracy ratings are expected to assume

more cooperative negotiation positions (positive expected effect).

The argument of the median voter preferring more cooperation (and thus environmental

protection) might seem counterintuitive, as the dominant strategy to free-ride on the effort

of others described in relation to states can equally well be applied to the individual level. In

other words, when given the choice of paying for emitting greenhouse gases or not, rational

choice theory suggests that a majority of people would opt for the latter option. However,

people usually do not have the choice whether to pay a tax or not, indeed a single person’s

influence on the implementation of new laws is usually quite limited. Brennan (2009) shows

that in such circumstances “doing the right thing” (i.e. supporting a policy that would be

rejected if one would decide alone) can generate a higher payoff than outright opposition

to a certain policy. As an example, think of voluntary carbon offsets in aviation. Everyone

can personally decide whether or not to pay this extra charge, which implies a very high

incentive for a majority of passengers, even for otherwise environmentally conscious people,

to free-ride. However, when legislation on the obligatory taxation of air travel is proposed,

the possibility of influencing the outcome for a single person is usually close to zero. In such

a case, environmentally conscious people might support the policy, earning a small payoff

30

from having followed their conscience. If there is a majority against the tax, which our single

individual cannot influence, he/she will get a higher payoff from not paying the tax, plus the

small value for having followed his/her conscience. If there is a majority in favor of the policy,

taxes will have to be paid. Yet this would be the case whether our individual has supported

the policy or not, although being in favor of the policy still yields a higher payoff for him/her.

This is why it can be in the interest of the median voter to desire environmentally friendly

policies from an economic perspective (Brennan, 2009).

Irrespective of a country’s democratic status, certain groups are able to influence domes-

tic decisions more than others because they are better organized and have a higher spending

capacity. Such lobby groups are able to influence governmental positions in international ne-

gotiations to different degrees, particularly if the constituency of the lobby group in question

is expected to be strongly affected by legislation following the climate change negotiations.

While there are a number of studies concluding that domestic interest groups do im-

pact national negotiation behavior and positions on climate change, most authors focus on

studying one particular pressure group, many being interested particularly in the role of

the business lobby (see e.g. Bryner, 2008; Newell and Patterson, 1998). Giving his study a

broader aim, Newell (2000) looks into the behavior of four different non-state actors (the

mass media, environmental pressure groups, the fossil fuel lobby, and Working Group 1 of

the IPCC) and shows that it is easier for these groups to forward their interest by influenc-

ing states’ positions through lobbying at the national level than by intervening directly at

international negotiations. Grundig (2009) notes that in providing (useful) information to

governments, lobby groups have an incentive to overstate the losses to their industries or the

environment; also their influence depends on how much they spend on lobbying efforts.

Therefore, we expect that a bigger fossil fuel lobby biases a country toward less coopera-

tive (and hence less environmentally friendly) positions, while nuclear and renewable energy

lobbies, as well as environmental NGOs ,have the opposite effect (Grundig, 2009). Dolsak

(2001) further states that bigger carbon intensive sectors bias countries towards less auda-

cious environmental commitments, but is unable to find empirical evidence for this. Again,

the difference between Annex 1 and non-Annex 1 countries is relevant here and therefore

has to be considered in the formulation of the following hypotheses.

H4a: Non-Annex 1 countries with bigger fossil fuel industries choose less coop-

erative negotiation positions (positive expected effect), while bigger nuclear, re-

newable energy, and agricultural industries lead to more cooperative positions

(negative expected effect).

31

H4b: Annex 1 countries with bigger fossil fuel industries assume less cooperative

negotiation positions (negative expected effect), while bigger nuclear, renewable

energy, and agricultural industries lead to cooperative positions (positive expected

effect).

It is especially in the case of democratic countries where we expect that certain domestic

actors play a substantial role in influencing a government’s position. Moreover, democratic

governments, and in particular coalition governments, face the problem that, in some in-

stances, they have to reconcile differing opinions of ministries so that an ordering of prefer-

ences is necessary. There is still little research as to how the intra-coalition dynamics and

cabinet decision making function (Kaarbo, 2008). Within the European Union, (Schneider

and Baltz, 2003) found that lead ministries are decisive in forming national negotiation posi-

tions in EU Council negotiations. It is, however, not yet evident which ministry is the most

decisive. We assume that for the Copenhagen negotiations it is the economics, finance and

environment ministries that are particularly relevant; we further assume that the environ-

ment ministry is the most decisive player in the cabinet, since it is likely to have the relevant

experience for the Copenhagen negotiations. However, the finance and economics ministries

are certainly additional important players given that commitments to address climate change

bear costs for the involved states.

Next to the dominant ministry in a country, the national public opinion might also

play an influential role. It is especially the climate change negotiations, which are highly

publicized, that are bound to be particularly influenced by the public’s attitude. Voters know

and care about climate policy; it is, however, unclear whether they are prepared to pay for

the costs of climate change policies. Public opinion is not always in favor of environment

protection, in particular if it is costly for some. It matters most whether winners and

losers are organized, and whether they have access to decision makers and can thus mean

political benefit for politicians (Dolsak, 2001). Furthermore, the attention voters pay to

climate change varies: (Harrison and McIntosh Sundstrom, 2007), for example, showed that

American and Canadian voters cared for the ratification of the Kyoto protocol, but were

less concerned about environmental politics in general. If voters do not sufficiently care

about a topic, interest groups stand a far higher chance of influencing governments since

they also provide them with information. However, it could be argued that the value change

from materialist to post-materialist values (Inglehart, 1977; Inglehart and Welzel, 2005) in

industrialized countries could motivate citizens to favor climate-friendly policies.

32

H5a: For non-Annex 1 countries, more environmentally friendly public opinion

and environment ministry values are expected to lead to more cooperative negoti-

ation positions (negative expected effect), while a stronger influence on the part

of the economics ministry leads to less cooperative negotiation positions (positive

expected effect).

H5b: For Annex 1 countries, higher public opinion and environment ministry

values are expected to lead to more cooperative positions (positive expected ef-

fect), while a stronger influence on the part of the economics ministry leads to

less cooperative positions (negative expected effect).

Apart from the economic and domestic interests of a state, the partisan preferences of

governments might also be relevant to the explanation of a particular negotiation position.

Several authors maintain that the partisan orientation of governments, in particular their

position on the left-right scale, matters for explaining their negotiation positions (Hosli

et al., 2011) as well as for explaining their policies, e.g. on financial globalization (Quinn

and Toyoda, 2007), capital controls (Kastner and Rector, 2003) or foreign trade and aid

(Milner and Tingley, 2011). As for climate policy, (Neumayer, 2003) discerns a positive

effect of left-wing governments on the reduction of air pollution, and Jensen and Spoon

(2011) can show that for EU countries, green party representation in government predicts

progress towards Kyoto Protocol targets. Since the ideological position of parties is a crucial

factor for distinguishing them from their competitors in national elections, we might expect

an influence of the partisan orientation of a negotiating government. We assume that left-

wing governments care more about environmental issues and climate change than the more

business-friendly/orientated right-wing parties.

H6a: Non-Annex 1 countries with more left-wing oriented governments that favor

the environment versus economic growth are expected to adopt more cooperative

positions (negative expected effect).

H6b: Annex 1 countries with more left-wing oriented governments that favor

the environment versus economic growth are expected to adopt more cooperative

positions (positive expected effect).

3 Data on state preferences in negotiations

Although international institutions have received increasing attention in the scholarly de-

bate on global governance, knowledge about positioning behavior within these organiza-

tional frameworks remains very limited. Some highly important negotiation rounds such as

33

the General Assembly in the United Nations (Dreher et al., 2008) and the Council of the

EU (Tallberg, 2008; Thomson and Stokman, 2006) are relatively well investigated, but the

respective research mostly attempts to explain negotiation outcomes rather than negotiation

positions. Negotiations that are more focused on single issues have so far not been system-

atically investigated at all, yet they are an interesting case, as their focus on one supposedly

single issue (which is of course broken up into a multitude of issues during the negotiation

process) makes issue linkages for the sake of coalition building impossible.

So far, negotiation studies have suffered from a lack of data due to the extreme secrecy

that shrouds international negotiations (Gabel et al., 2002). A few studies have made use

of the final voting records (Hayes-Renshaw et al., 2006; Mattila and Lane, 2001; Mattila,

2004, 2006) to analyze voting patterns at the end of international negotiations. To gather

information on negotiation positions at the beginning of Council deliberations, the analysis

of negotiation protocols (Bueno de Mesquita and Stokman, 1994; Hopmann, 2002; Hug and

Konig, 2002; Konig, 1997) or the direct interviewing of negotiation participants have been

used (Thomson et al., 2006). Whereas text analysis bears the advantage that it is easily

traceable by other researchers and that it is cheaper (Sullivan and Selck, 2007), interviews

allow for the identification of crucial negotiation issues and the saliencies attached to them.

These methods of identifying negotiation positions - documents and interviews - suffer

from the fact that it is unclear just how strongly the identified positions reflect the actual

preferences of actors, which as a result, most researchers tend to just assume to be the case

(e.g. for qualitative studies see Dinan, 1999; Hosli, 2000; Moravcsik, 1998, and for quantita-

tive studies see Bueno de Mesquita and Stokman 1994; Konig 1997; Thomson and Stokman

2006). While Achen (2006) considers it possible to measure sincere positions, Bueno de

Mesquita (2004) believes that it is practically impossible to ascertain the real opinion of a

negotiator. He argues that it is especially in situations with incomplete information, where

there is a low probability of finding out the real preference of negotiators, that it is far too

advantageous for a diplomat to assume a strategic position. However, in our case, this prob-

lem is not serious since we are investigating negotiation positions that encompass sincere

considerations and strategic interests.

3.1 Dependent variable

Data for the dependent variable, i.e. country positions on emission reduction targets and

mitigation finance, were obtained by hand-coding all submissions made by negotiating par-

ties to the UNFCCC over the two years prior to the Copenhagen negotiation round 2009.

Submissions present the view of negotiating parties in written form, are compiled into official

34

UNFCCC documents and can be downloaded from the organization’s website. In total, the

hand-coding effort comprised of 43 official UNFCCC documents and a total of over 1,600

pages of proposed legal text. To ensure inter-coder reliability, a codebook was designed and

separately tested on 25 pages by three coders. Then the codebook was adjusted to straighten

out discrepancies found in the first coding round and an additional 25 pages were coded.

After checking the coding scheme for consistency again, we found that the inter-coder re-

liability was very high when the adjusted codebook was used and that differences among

the three coders were reduced substantially. The main aim of this coding process was to

generate a dataset on the negotiation positions for all countries on the issues of emissions

reduction targets, the use of market mechanisms, mitigation and adaptation finance, as well

as Measurement, Reporting and Verification (MRV) of greenhouse gas mitigation.

The issues were coded on a scale ranging from 0 to 100 (apart from emission reduction

targets for which the actual number was recorded). As both submissions of individual

countries as well as group submissions were coded, a decision was made regarding how best to

combine these different sources of information: individual submissions of a country regarding

a given issue were given preference over group submissions. In cases where more than one

individual country submission was made over the coding period on a given issue, the average

was taken. If no individual submission was made by a country regarding a given issue, the

group submission of the most important negotiation group of that country was taken as a

proxy. This decision was justified on the grounds that if the group position in question did

not accurately reflect a country’s own views, the delegation would have formulated its own

submission. As in the case of individual submissions, the average was taken when multiple

group submissions on the same issue were made.

For the models presented in this paper we use two issues as dependent variables. First, we

use the core issue of the climate change negotiations: Annex 1 emission reduction targets.

The variable “reduction targets” describes a country’s negotiation position on reduction

targets for all Annex 1 countries by 2020 as an aggregate (measured in per cent of greenhouse

gas reductions). As Annex 1 is congruent with the developed world, this allows countries

to specifically target industrialized countries with their demands. The mean demanded

reduction target over all observations in the dataset is 39.8, yet Annex 1 countries on average

want much lower targets (31.1) than non-Annex 1 parties (42.2). This already shows that

the division of negotiating parties into the Annex 1 and non-Annex 1 groups is one, if not

the major, fault line in the climate change negotiations. The country aiming for the lowest

Annex 1 reduction targets is Russia (25), while Bolivia demands the maximum observed

value of 49.

The second dependent variable is called “mitigation funds”. Positions were coded on a

35

dimension ranging from 0 to 100, where 0 means that mitigation funds flowing from Annex

1 countries to developing countries should consist entirely of voluntary contributions, while

100 indicates that mitigation funds should be a mandatory 1% of developed countries’ GDP

each year. Unsurprisingly, Annex 1 countries, with a mean of 61.8, are on average closer to

favoring voluntary funding than non-Annex 1 countries with a mean of 76.2 (the overall mean

is 72.8). However, as this issue dimension was not as hotly debated during the conferences

prior to and post Copenhagen, the battle lines for this issue area were not as clearly drawn

as in the case of emission reduction targets.

3.2 Independent variables

3.2.1 Ecological vulnerability

To measure vulnerability to climate change impacts we use the Environmental Vulnerability

Index (EVI) developed by the South-Pacific Applied Geoscience Commission (SOPAC). The

EVI measures 50 different indices, 13 of which are utilized to compile a sub-index capturing

vulnerability to climate change (Kaly et al., 2004). Although the EVI might be problematic

for a variety of reasons (Barnett et al., 2008), the main criticism that such a constructed

index is unable to capture complex socio-ecological processes equally applies to all alternative

indices.

3.2.2 Power

Reflecting the strategic choice of positions as assumed by neo-realism, we include a variable

measuring power in all models presented below. To measure power, we use a country’s total

GDP at purchasing power parity (the log thereof).

3.2.3 Democratic status

A higher democratic rating, according to the theory presented above, is expected to be

associated with higher commitment levels. We operationalize democratic status using the

Polity IV index, which has the advantage of being carefully crafted, this being among the

main reasons for the widespread use of this index in the literature. The index ranges from

0 (least democratic) to 10 (most democratic) and the average value for the countries in our

data set is 6.31.

36

3.2.4 Franchise

The franchise or the level of influence different interest groups can assert is measured using

the fraction of GDP stemming from the relevant industries or pressure groups. We use the

World Development Indicators (WDI) to summate GDP generated by agriculture, forestry,

and the oil, gas, and coal industry, and then divide the product by overall GDP to generate

proxies for polluting as well as environmental interest groups. As this generated a number in

percentage-form, we can use nominal values for the calculations and do not have to rely on

distorted purchasing power parity values. The WDIs do not include data for the percentage

of GDP generated by either the nuclear industry or through the production of renewable

energy. Whether our proxy for the green industry is able to capture the real influence of

industries with interests in higher abatement levels is therefore up for debate.

3.2.5 Influence of ministries and public opinion

To measure the possible influence of domestic actors, we use data stemming from interviews

with negotiating parties over a time period of about nine months prior to and after the

Conference of the parties (COP) 15 in Copenhagen. Interview partners were asked to rate

the influence of their environmental and finance ministries, as well as of the domestic public

opinion, on a scale from 1 (=very low) to 5 (=very high). As only 56 countries responded

to our interview requests, the number of cases in the models that include these variables is

drastically reduced.

3.2.6 Partisan orientation

To account for the partisan orientation of the negotiating governments, we use data from

the Database on Political Institutions provided by the World Bank. This partisan variable

simply captures whether the largest government party is considered to be on the left (=1),

center (=2) or the right (=3) of the political spectrum.

3.2.7 Free riders

As explained in detail above, one of the major problems of the suggested model is the

existence of a so-called “Firewall” between Annex I and non-Annex I parties (Bodansky,

2010, p.234). This firewall allows developing countries to participate in the Kyoto protocol

without having to accept binding emission reduction targets and without having to contribute

to the proposed mitigation funds. Such legally binding measures only apply to developed

nations, i.e. countries named in the Annex I to the Kyoto Protocol. Although there are

a number of reasons why such a division between the rich and the poor can no longer be

37

upheld (Castro, 2010), developing countries strongly resist the idea of adopting binding

targets themselves. This architecture of the Kyoto Protocol is somewhat problematic for

our purpose, as it allows developing countries to demand a very high level of environmental

ambition for any future treaty, as they can assume that any costs associated with such a

new treaty will be borne by Annex I countries. In other words, if a country knows that

it will not have to pay for its demands, this can lead to a shift in its preferences towards

more ambitious overall targets, which are in turn associated with higher total costs, which

will have to be shouldered by industrialized countries. Annex I countries on the other hand,

knowing that the burden of paying for an international treaty will fall predominantly on

them, will have preferences associated with lower costs, as they will have to carry the full

weight of the bill. We account for this by introducing a dummy variable indicating whether

a country belongs to the Annex 1 group or not and by including interaction terms with the

above described independent variables, thus allowing the effects of the proposed hypotheses

to differ for Annex 1 and non-Annex 1 countries. Table 1 provides summary statistics for

the independent variables described in this section.

Table 1: Descriptive statistics of the independent variables

Variable name Obs. Mean s.d. Min. Max.Ecological vulnerability 137 3.44 0.76 1.76 5.50Power (log of GDP) 140 24.04 2.48 18.67 30.53Democratic status 118 5.36 3.89 0 10Franchise I (oil, gas, coal) 108 17.87 24.55 0 89.13Franchise II (agriculture, forestry) 108 13.8 13.27 0.07 61.30Partisan orientation 98 2.14 0.93 1 3Influence of ministries I (Environment) 55 8.04 1.50 3 9Influence of ministries II (Finance) 38 5.32 2.17 1 9Influence of public opinion 54 5.39 1.86 1 9Index for political competitiveness 54 6.44 1.37 1 7World value survey 36 1.85 0.22 1.48 2.21

4 Analysis

In Tables 2 and 3 we list the results of our multivariate data analysis for both issue areas,

i.e. mitigation finance and emission reduction targets.

38

Table 2: Results for Mitigation Finance

Model 1 Model 2 Model 3 Model 4 Model 5 Model 6

(Intercept) 83.98∗∗ 91.60∗∗ 67.36∗∗ 10.96 266.97∗∗ −29.79(19.61) (17.29) (21.05) (59.70) (78.57) (31.01)

Power(GDP) 0.05 0.48 0.16 1.58 −1.95 2.77∗∗

(0.84) (0.65) (0.87) (1.48) (1.76) (0.90)Democracy −1.56∗∗

(0.52)Annex 1 −108.81∗∗ −60.76∗∗ −0.58 52.22 −140.08∗∗ 100.35∗∗

(40.27) (20.19) (7.84) (56.85) (64.65) (32.56)Democracy*Annex1 10.39∗∗

(4.22)Vulnerability −8.21∗∗

(1.98)Vulnerability*Annex1 12.96∗∗

(5.41)Green industry 0.22

(0.15)Emitter industry 0.12

(0.11)Green industry*Annex1 −2.99∗∗

(1.43)Emitter industry*Annex1 −0.29

(0.39)World Value Survey (WVS) 13.24 −70.37∗∗

(26.22) (27.20)WVS*Annex1 −35.83 65.10∗

(31.93) (34.22)Public opinion −16.44∗

(8.97)Public opinion*Annex1 16.43

(13.22)Finance ministry (FM) 1.76

(1.15)Environmental ministry (EM) 4.14∗∗

(1.83)FM*Annex1 −8.44∗∗

(2.36)EM*Annex1 −9.13∗∗

(3.33)N 119 137 107 67 29 31R2 0.25 0.24 0.21 0.12 0.54 0.55adj. R2 0.22 0.22 0.16 0.07 0.41 0.44F-value 9.48∗∗ 10.61∗∗ 4.41∗∗ 2.17∗ 4.30∗∗ 4.86∗∗

Robust standard errors in parentheses∗∗p < 0.05, ∗p < 0.10

39

Table 3: Results for Emission Reduction Targets

Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7

(Intercept) 56.41∗∗ 58.26∗∗ 59.50∗∗ 56.10∗∗ 18.10∗ 48.42∗∗ 64.22∗∗

(3.36) (4.11) (3.81) (4.32) (9.89) (12.49) (8.37)Power (GDP) −0.62∗∗ −0.63∗∗ −0.62∗∗ −0.68∗∗ −0.25 −0.95∗∗ −0.95∗∗

(0.15) (0.18) (0.14) (0.18) (0.28) (0.31) (0.25)Annex 1 −23.17∗∗ −12.15∗∗ −15.93∗∗ −6.61∗∗ 18.86∗∗ 1.86 −7.97

(10.45) (3.11) (4.32) (1.45) (8.82) (9.82) (11.98)Power*Annex1 0.53

(0.40)Democracy −0.26∗∗

(0.10)Democracy*Annex1 0.43

(0.34)Vulnerability −0.94∗∗

(0.42)Vulnerability*Annex1 1.95∗

(1.17)Green industry 0.07∗∗

(0.03)Emitter industry 0.03

(0.02)Green industry*Annex1 −0.22

(0.26)Emitter industry*Annex1 −0.06

(0.07)World Value Survey (WVS) 16.19∗∗ 10.57∗∗

(3.72) (3.93)WVS*Annex1 −15.84∗∗ −7.69

(4.89) (5.10)Public opinion −2.46∗

(1.38)Public opinion*Annex1 3.48∗

(2.02)Finance ministry (FM) −0.18

(0.31)Environmental ministry (EM) 0.11

(0.50)FM*Annex1 −0.16

(1.15)EM*Annex1 0.15

(1.16)N 140 118 137 108 67 29 30R2 0.70 0.71 0.70 0.72 0.70 0.88 0.80adj. R2 0.69 0.70 0.69 0.70 0.68 0.85 0.75F-value 103.6∗∗ 70.86∗∗ 77.39∗∗ 42.52∗∗ 35.74∗∗ 27.17∗∗ 15.53∗∗

Robust standard errors in parentheses∗∗p < 0.05, ∗p < 0.10

40

4.1 Vulnerability

In the case of mitigation finance, non-Annex 1 countries demand fewer funds as vulnerability

increases, as revealed by the significant negative coefficient of 8.21 in model 2 of Table 2. The

very high intercept and the highly significant negative effect of the Annex 1 dummy indicates

that there is a considerable clash of interest between donor (Annex 1) and recipient (non-

Annex 1) countries. Thus, the negative effect on vulnerability for the latter group might be

a signal of the more vulnerable countries that, although they depend on increased mitigation

levels and financial funds regarding both mitigation and adaptation for their survival, they

are willing to enter a compromise for the sake of reaching an international, binding agreement.

Conversely, more vulnerable Annex 1 countries are willing to offer more mitigation money, as

shown by the positive and significant effect of 4.75 for this group. Panel b of Figure 3 shows

this result graphically and reveals that up to a vulnerability level of 4 (more than 75% of

the countries in the dataset) there is a significant difference between the two groups. When

vulnerability levels are very high, the clash in interests vanishes and countries of both groups

are increasingly willing to cooperate in order to overcome the deadlock of the negotiations.

For emission reduction targets the picture is very similar. While Annex 1 countries tend to

offer significantly less than what non-Annex 1 countries demand, this antagonism diminishes

as vulnerability increases. Non-Annex 1 countries demand lower emission reduction targets

(-0.94 and significant) as vulnerability increases, while Annex 1 countries offer higher ones

(1.01 and significant). Thus, we find that higher vulnerability levels increase the inclination

of countries to cooperate, and both hypotheses 3a and 3b are supported.

4.2 Power

Different to our argument regarding vulnerability, we presume the effect of power to have

the same direction for both groups, i.e. increased power levels are expected to coincide

with a reduced interest in cooperation. Indeed, when the interaction between Annex 1 and

power (measured by total GDP) is included in the model with mitigation finance as the

dependent variable, we find that neither the slope nor the interaction is significant from

zero. We therefore include power as a homogeneous variable without the interaction effect in

the different models presented in Table 2, but fail to find consistent significant effects across

the models. We therefore conclude that in the case of mitigation finance, power does not

significantly affect the bargaining behavior of countries.

In the case of emission reduction targets the story is different: although the interaction

effect is again insignificant, the slope coefficient of power is negative and highly significant

across all model specifications tested in this paper. Rich and powerful countries, it seems,

41

Figure 3: Conditional effects of the key independent variables on mitigation finance

0 2 4 6 8 10

−15

0−

100

−50

0

Democracy

Con

ditio

nal E

ffect

Panel a: Democracy

2 3 4 5

−60

−40

−20

020

Vulnerability

Con

ditio

nal E

ffect

Panel b: Vulnerability

0 5 10 15 20 25

−10

0−

500

Green Industry

Con

ditio

nal E

ffect

Panel c: Green Industry

1.4 1.6 1.8 2.0 2.2

−80

−60

−40

−20

020

Global Values

Con

ditio

nal E

ffect

Panel d: Global Values

The solid lines in the panels indicate the expected difference between Annex 1 and non-Annex 1countries regarding mitigation finance at different levels of the independent variable in question.The dashed lines show 95% confidence intervals.

do not exhibit great enthusiasm for committing themselves to cut back on their emissions.

Thus we find some evidence in support of hypothesis 2.

4.3 Democracy

Regarding mitigation finance, the interaction term between Annex 1 and democracy indicates

a relatively large (10.39) and highly significant effect. This positive coefficient in combination

with the negative Annex 1 dummy effect indicates a convergence of positions between the

two groups as the level of democracy increases. The effect of an increased democracy level

for non-Annex1 countries is -1.56 and significant at the 1% level, while for Annex 1 countries

this effect is highly positive, with a coefficient value of 8.83, and remains significant. This

therefore suggests that, as the level of democracy increases, countries of both groups are

willing to make concessions and to agree to the demands of the opposite camp. In other

words, developing countries with higher democracy values are willing to reduce their demands

significantly, while developed countries with higher democracy values in turn considerably

increase their offers. Panel a of Figure 3 shows this difference in the form of a graph: at

42

low levels of democracy, the gap between the two groups is rather large and significant, but

grows smaller as democracy levels increase; it is only at very high levels of democratic status

that the significance vanishes, which indicates that, at those levels of democracy, countries

are very willing to accommodate the other side’s wishes in order to reach an agreement.

For emission reduction targets, we again find that non-Annex 1 countries are willing to

reduce their demands, while countries belonging to the Annex 1 group are ready to offer

higher targets for their group as a whole (although the effect for the latter is not significant).

Yet again, higher levels of democracy bring countries of the two camps closer together, a

result which lends further support to both hypotheses 3a and 3b.

4.4 Green and emitter industries

For mitigation finance, we see in model 3 of Table 2 that both the coefficient for the emitter

industry variable and the corresponding interaction term are relatively small and insignifi-

cant. However, the interaction term between the green industry variable and the Annex 1

dummy is highly significant, yet contrary to the hypothesized negative effect. More formally,

the results indicate that as the share of GDP generated by the green industry increases, the

propensity of Annex 1 countries to offer substantial and binding mitigation funds dimin-

ishes. For non-Annex 1 countries the effect has the opposite sign, i.e. they demand more

funds from developed countries, although this effect is not significant. As a consequence,

the higher the green industry’s share of GDP, the wider the gap between the two groups,

as depicted in panel c of Figure 3. Here, it can also be seen that for low values of the

green industry’s GDP share, the difference between the Annex 1 and non-Annex 1 groups

is insignificant, yet quickly turns significant as green industry output increases. The reason

for this unexpected result might be that the domestic green industries of developed coun-

tries are concerned about money flowing to foreign governments where it is utilized to fund

their competitors. Therefore, these industries might lobby for more spending on domestic

mitigation and against increased mitigation expenditures far away from home, accompanied

by the prospect of strengthening foreign rivals.

The picture for the emission reduction case is somewhat similar, although this time

the slope coefficient is significant (and positive) only for non-Annex 1 countries, while both

emitter industry coefficients again are insignificant. This implies that a bigger green industry

in developing countries induces negotiation positions that reflect higher reduction target

demands from the Annex 1 countries as a block. These Annex 1 countries, on the other hand,

offer less as their green industry becomes more influential. Thus, for reasons that should be

further investigated, a prominent green industry has a negative effect for cooperation in the

43

climate change negotiations, and hypotheses 4a and 4b are not only rejected, but reversed.

4.5 Public opinion and ministry influence

Increased public awareness of climate change induces non-Annex 1 countries to act more

cooperatively with regard to mitigation finance. Both the coefficients for the influence of

public opinion on a country’s negotiation position and the World Value Survey (WVS) are

negative and highly significant. This implies that, as awareness of climate change grows,

developing countries lower their demands in order to facilitate the finding of an agreement.

The same cannot be said for the Annex 1 countries: as the interaction terms in Model 5 of

Table 2 show, the influence of public opinion and the WVS is very close to zero. Here, it

is rather the influence of the finance and environmental ministries that is substantial. As

expected, an increased influence of the finance ministry considerably reduces the willingness

of developed countries to pay for the others’ mitigation efforts (the corresponding effect for

non-Annex 1 countries is not significant). Surprisingly, the same is true for an increased

influence of the environmental ministry, with a significant slope coefficient of -4.99 for the

Annex 1 and 4.14 for the non-Annex 1 countries. This suggests that, as environmental

ministries increase their sway over negotiation positions, countries grow more reluctant to

cooperate. Again, the reason for Annex 1 countries might be that environmental ministries

believe that mitigation finance should be spent domestically before money is funneled to

foreign governments. Conversely, non-Annex 1 countries with active environmental ministries

may shy away from negotiation positions that are too cooperative as this might undermine

the formation of an effective treaty that is capable of combating global climate change.

Regarding emission reduction targets no significant effects can be found for ministerial

influence. An increased influence of the public opinion on negotiation positions on the

other hand again seems to foster cooperation, as in the mitigation finance case. However,

the WVS index has the opposite effect than for mitigation finance, which indicates that

developing countries with a higher public awareness of climate change demand significantly

higher aggregate reduction targets from Annex 1 countries. Overall, hypotheses 5a and

5b are somewhat supported by the findings regarding public opinion for both dependent

variables and the finance ministries’ influence in the case of mitigation finance. Regarding

the environmental ministries’ influence, the coefficients have unexpected sings and the results

are in general too fuzzy to support the proposed hypotheses.

44

4.6 Partisan orientation

The partisan orientation of the government does not significantly change a country’s nego-

tiation position for either mitigation finance or emission reduction targets. Indeed, as none

of the coefficients in question comes close to conventional levels of significance, hypotheses

6a and 6b must be rejected. We can thus confirm findings concerning EU negotiations that

have also shown the orientation of the government on a left-right axis not to impact its inter-

national negotiation behavior on climate change issues. This finding stands in contrast to a

study by Jensen and Spoon (2011), who found a distinct impact of the partisan orientation

of EU governments. This discrepancy in results might be due to our rather crude measure of

left-right orientation since there is no better global measure at the moment. Better data and

more intensive research regarding this question are therefore planned for future research.

5 Conclusion

Positions are a fundamental requirement for analyzing the behavior of states in international

negotiations. The choice of these positions results from the conscious decisions made on

the part of governments to assume a certain position in a negotiation; they encompass the

interests of states as well as strategic considerations. In our data set “Negotiating Climate

Change” we have collected new and unique data on the positions and bargaining strategies

of the UNFCCC member states in order to further our understanding of the dynamics of

international negotiations. Building on and extending existing research on negotiations in the

EU or other international treaties such as the Law of the Seabed negotiations (Brauninger,

2001), we show for the first time for climate change negotiations that negotiation positions are

influenced by both economic and domestic, as well as by strategic, considerations. Thus, we

extend existing research on states’ ratification behavior of environmental treaties or different

environmental policy outputs with our initial analysis of the determinants of environmental

negotiation positions.

We have found that the division of countries into the Annex I and non-Annex I groups,

determined by the climate treaty architecture, is a very powerful predictor for the choice of

negotiation positions. However, this is not the only institutional influence in negotiations:

the effect of democratic structures is rather impressive: democracies be it in the context of

developed or developing countries are more cooperative in climate change negotiations. Due

to their increased accountability towards voters they might be more afraid of not being able

to deliver the common good of environmental protection. Thus we could hope that recent

democratization developments are also good news for the fight against climate change. Ad-

45

ditionally, our findings suggest that governments choose their negotiation positions carefully

according to their economic and structural interests, but that they also consider how depen-

dent they are on international cooperation in addressing climate change. Hence, increased

vulnerability to climate change leads to more cooperative positions that promote common

solutions as opposed to negotiation breakdown.

A government can, at least to a certain extent, choose which stakeholder it considers

relevant and wants to grant influence to. Our results from the multivariate data analysis

clearly indicated this: climate change seems to be a political issue which is so salient in

domestic discussions that politicians react to the attitudes of the public and to environmen-

tal consciousness, although this is only the case for developing countries (non- Annex I).

Somehow less effective seems to be the influence of business interests and lobby groups from

the emitter and green industries. So far, we could not find a strong influence on the part

of these particular political actors, although we did initially expect one. A rather surprising

result was that pertaining to the effect of environmental ministries and the green industry:

although one might assume that these actors have an interest in international climate pro-

tection, we found their influence to be detrimental to cooperation and rather “egoistic”: If

subsidies are paid for climate protection, these should preferably be allocated within the

home nation according to these stakeholders.

If these findings also hold for other environmental treaties, we would on the one hand

conclude that the institutionalization of democratic channels promotes environmental pro-

tection; it helps to give the public opinion a voice in particular in developing countries.

However, domestic national stakeholders such as the green industry can be detrimental to

climate protection since they pursue their individualistic strategies and therewith motivate

governments to be less cooperative. Finding a balance between the voters’ interests and

the domestic interest groups will thus be a major challenge for governments in current and

future international negotiations.

46

References

Achen, C. H. (2006). Evaluating political decisionmaking models. In R. Thomson, F. N.Stokman, C. H. Achen, and T. Konig (Eds.), The European Union Decides: PoliticalEconomy of Institutions and Decisions, pp. 254–298. Cambridge: Cambridge UniversityPress.

Babiker, M. (2005). Climate change policy, market structure, and carbon leakage. Journalof International Economics 65 (2), 421–445.

Barnett, J., S. Lambert, and I. Fry (2008). The hazards of indicators: Insights from the Envi-ronmental Vulnerability Index. Annals of the Association of American Geographers 98 (1),102–119.

Barrett, S. (2003). Environment and Statecraft: The Strategy of Environmental Treaty-Making. Oxford: Oxford University Press.

Barrett, S. and R. Stavins (2002). Increasing Participation and Compliance in InternationalClimate Change Agreements. Fondazione Eni Enrico Mattei, Nota di Lavoro 94.

Bernauer, T. and V. Koubi (2009). Effects of political institutions on air quality. Ecologicaleconomics 68 (5), 1355–1365.

Blanford, G., R. Richels, and T. Rutherford (2008). Revised emission growth projections forChina: Why post-Kyoto climate policy must look east. The Harvard Project on Interna-tional Climate Agreements, Discussion Paper 08-06.

Bodansky, D. (2010). The Copenhagen climate change conference - A postmortem. AmericanJournal of International Law 104 (2), 230–240.

Brauninger, T. (2001). Nationale Interessen und internationale Verhandlungen. Determinan-ten von Staatenpositionen in der internationalen Politik. Konstanzer Online-Publikatins-System (KOPS).

Brennan, G. (2009). Climate change: A rational choice politics view. The Australian Journalof Agriculture and Resource Economics 53, 309–326.

Bryner, G. (2008). Failure and opportunity: Environmental groups in US climate changepolicy. Environmental Politics 17 (2), 319–336.

Bueno de Mesquita, B. (2004). Decision-making models, rigor and new puzzles. EuropeanUnion Politics 5 (1), 125–138.

Bueno de Mesquita, B. and F. N. Stokman (1994). European Community Decision Making:Models, Applications, and Comparisons. New Haven and London: Yale University Press.

Castro, P. (2010). Climate change mitigation in advanced developing countries: Empiricalanalysis of the low-hanging fruit issue in the current CDM. CIS Working Paper 54 .

47

Cline, W. (1992). The Economics of Global Warming. Washington, D.C.: Institute forInternational Economics.

den Elzen, M., J. Fruglestvedt, N. Hohne, C. Trudinger, J. Lowe, B. Matthews, B. Romstad,C. P. de Cammpo, and N. Andronova (2005). Analysing countries’ contribution to climatechange: Scientific and policy-related choices. Environmental Science & Policy 8 (6), 614–636.

Dinan, D. (1999). Ever Closer Union: An Introduction to European Integration (2nd Editioned.). Bolder, CO: Lynne Rienner Publishers Inc.

Dolsak, N. (2001). Mitigating global climate change: Why are some countries more commit-ted than others? Policy Studies Journal 29 (3), 414–436.

Dreher, A., P. Nunnenkamp, and R. Thiele (2008). Does US aid buy UN general assemblyvotes? A disaggregated analysis. Public Choice 136 (1), 139–164.

Fearon, J. D. (1997). Signaling foreign policy interests: Tying hands versus sinking costs.The Journal of Conflict Resolution 41 (1), 68–90.

Fredriksson, P. G. and N. Gaston (2000). Ratification of the 1992 climate change convention:What determines legislative delay? Public Choice 104 (3), 345–368.

Gabel, M., S. Hix, and G. Schneider (2002). Who is afraid of cumulative research? Thescarcity of EU decision making data and what can be done about this. European UnionPolitics 3 (4), 481–500.

Garrett, G. and P. Lange (1986). Performance in a hostile world: Economic growth incapitalist democracies. World Politics 38 (4), 517–545.

Grundig, F. (2006). Patterns of international cooperation and the explanatory power ofrelative gains: An analysis of cooperation on global climate change, ozone depletion, andinternational trade. International Studies Quarterly 50 (4), 781–801.

Grundig, F. (2009). Political strategy and climate policy: A rational choice perspective.Environmental Politics 18 (5), 747–764.

Harrison, K. and L. McIntosh Sundstrom (2007). The comparative politics of climate change.Global Environmental Politics 7 (4), 1–18.

Hayes-Renshaw, F., W. van Aken, and H. Wallace (2006). When and why the Council ofMinisters of the EU votes explicitly. Journal of Common Market Studies 44 (1), 161–194.

Hinich, M. and M. Munger (1997). Analytical Politics. Cambridge, New York, Melbourne:Cambridge University Press.

Holzinger, K. and T. Sommerer (2011). race to the bottomor race to Brussels? environmentalcompetition in Europe. JCMS: Journal of Common Market Studies 49 (2), 315–339.

48

Hopmann, P. T. (1996). The Negotiation Process and the Resolution of International Con-flicts. Columbia: University of South Carolina Press.

Hopmann, P. T. (2002). Negotiating data: Reflections on the qualitative and quantitativeanalysis of negotiation processes. International Negotiation 7, 67–85.

Hosli, M. O. (2000). The creation of the European economic and monetary union (EMU):Intergovernmental negotiations and two-level games. Journal of European Public Pol-icy 7 (5), 744–766.

Hosli, M. O., M. Mattila, and M. Uriot (2011). Voting in the Council of the European Unionafter the 2004 enlargement: A comparison of old and new member states. JCMS: Journalof Common Market Studies 7, 1249–1270.

Hug, S. and T. Konig (2002). In view of ratification: Governmental preferences and domes-tic constraints at the Amsterdam Intergovernmental Conference. International Organiza-tion 56 (02), 447–476.

Inglehart, R. (1977). The Silent Revolution. Changing Values and Political Styles AmongWestern Publics. New Jersey: Princeton University Press.

Inglehart, R. and C. Welzel (2005). Modernization, Cultural Change, and Democracy: TheHuman Development Sequence. Cambridge: Cambridge University Press.

Jensen, C. B. and J.-J. Spoon (2011). Testing the party matters thesis: Explaining progresstowards Kyoto Protocol targets. Political Studies 59 (1), 99–115.

Kaarbo, J. (2008). Coalition cabinet decision making: Institutional and psychological factors.International Studies Review 10 (1), 57–86.

Kaly, U., C. Pratt, and J. Mitchell (2004). The demonstration environmental vulnerabilityindex (EVI) 2004. Technical report, SOPAC Technical Report 384.

Kastner, S. L. and C. Rector (2003). International regimes, domestic veto-players, andcapital controls policy stability. International Studies Quarterly 47 (1), 1–22.

Konig, T. (1997). Europa auf dem Weg zum Mehrheitssystem. Grunde und KonsequenzenNationaler und Parlamentarischer Integration. Opladen: Westdeutscher Verlag.

Kroll, S. and J. Shogren (2008). Domestic politics and climate change: international publicgoods in two-level games. Cambridge Review of International Affairs 21 (4), 563–583.

Mattila, M. (2004). Contested decisions: Empirical analysis of voting in the European UnionCouncil of Ministers. European Journal of Political Research 43 (1), 29–50.

Mattila, M. (2006). Voting and coalitions in the council. two years after enlargement. InD. Naurin and H. Wallace (Eds.), Workshop: Who Governs in the Council of Ministers?Florence: European University Institute.

49

Mattila, M. and J.-E. Lane (2001). Why unanimity in the Council? A roll-call analysis ofCouncil voting. European Union Politics 2 (1), 31–52.

Milner, H. V. and D. H. Tingley (2011). Who supports global economic engagement?The sources of preferences in American foreign economic policy. International Organi-zation 65 (01), 37–68.

Moravcsik, A. (1997). Taking preferences seriously: A liberal theory of international politics.International Organization 51 (4), 513–553.

Moravcsik, A. (1998). The Choice for Europe: Social Surpose and State Power from Messinato Maastricht. Ithaca: Cornell University Press.

Morgenthau, H. (1948). Politics Among Nations. The Struggle for Power and Peace. NewYork: Alfred A. Knopf.

Morrow, J. (1994). Game Theory for Political Scientists. Princeton, NJ: Princeton UniversityPress.

Neumayer, E. (2002). Do democracies exhibit stronger international environmental commit-ment? A cross-country analysis. Journal of Peace Research 39 (2), 139–164.

Neumayer, E. (2003). Are left-wing party strength and corporatism good for the envi-ronment? Evidence from panel analysis of air pollution in OECD countries. Ecologicaleconomics 45 (2), 203–220.

Newell, P. (2000). Climate for Change. Non-state Actors and the Global Politics of theGreenhouse. Cambridge, New York: Cambridge University Press.

Newell, P. and M. Patterson (1998). A climate for business: Globale warming, the state andcapital. Review of International Political Economy 5 (4), 679–703.

Nordhaus, W. and J. Boyer (2000). Warming the world: economic models of global warming.Cambridge, MA: MIT Press.

Ostrom, E. (1990). Governing the commons. The evoluton of institutions for collective action.Cambridge: Cambridge University Press.

Putnam, R. D. (1988). Diplomacy and domestic politics: The logic of two-level games.International Organization 42 (3), 427–460.

Quinn, D. P. and A. M. Toyoda (2007). Ideology and voter preferences as determinants offinancial globalization. American Journal of Political Science 51 (2), 344–363.

Rogowski, R. (1999). Institutions as constraints on strategic choice. In D. Lake and R. Powell(Eds.), Strategic Choice and International Relations, pp. 115–136. Princeton, New Jersey:Princeton University Press.

Schneider, G. and K. Baltz (2003). The power of specialization: How interest groups influenceEU legislation. Rivista di Politica Economica 93, 253–283.

50

Schultz, K. (2001). Looking for audience costs. The Journal of Conflict Resolution 45 (1),32–60.

Sprinz, D. and T. Vaahtoranta (1994). The interest-based explanation of international envi-ronmental policy. International Organization 48 (1), 77–105.

Srinivasan, T., S. Carey, E. Hallstein, P. Higgins, A. Kerr, L. Koteen, A. Smith, R. Watsonf,J. Harte, and R. B. Norgaard (2008). The debt of nations and the distribution of ecologicalimpacts form human nature. Proceedings of the National Academy of Sciences of the UnitedStates of America 105 (5), 1768–1773.

Sullivan, J. and T. J. Selck (2007). Political preferences, revealed positions and strategicvotes: Explaining decision-making in the EU Council. Journal of European Public Pol-icy 14 (7), 1150–1161.

Tallberg, J. (2008). Bargaining power in the European Council. JCMS: Journal of CommonMarket Studies 46 (3), 685–708.

Thomson, R. and F. N. Stokman (2006). Research design: Measuring actors positions,salience and capabilities. In R. Thomson, F. N. Stokman, C. H. Achen, and T. Konig(Eds.), The European Union Decides, pp. 54–85. Cambridge, UK: Cambridge UniversityPres.

Ward, H. (1996). Game theory and the politics of global warming: The state of play andbeyond. Political Studies 44 (5), 850–871.

51

Determinats of Bargaining Success in theClimate Change Negotiations

Florian Weiler

Center for Comparative and International StudiesFederal Institute of Technology, Zurich

Paper published in Climate Policy 12(5)

Abstract

A novel data set, combining interview data with negotiation delegates and hand-codeddata of delegate statements, was used to empirically test six hypotheses about thedeterminants of bargaining success in the United Nations Framework Convention onClimate Change (UNFCCC) negotiations. The success of a state’s bargaining strategywas evaluated by first measuring the distance from a state’s original position on eightpolicy issues (e.g. emissions reduction targets) to the current state of the negotiations.The results were then readjusted using salience weights to control for how importanteach negotiation issue has been for each delegation. It was found that the externalpower of a state and how vulnerable a state is to climate change positively influence itsbargaining success, while the extremity of a state’s position and its share of emissionsappears to negatively influence it. In addition, the use of soft bargaining strategies bya state, which mutually benefits all concerned actors, was found to positively influencesuccess when a negotiation issue was particularly salient to it. Thus, it appears thatthe influence of powerful nations, such as the US and China, in the climate changenegotiations may not be as strong as previously thought.

Keywords: bargaining success, climate change, negotiations, preference attainment,UNFCCC

1 Introduction

What are the factors that determine bargaining success, i.e. the level of influence a party

exerts over negotiated outcomes, in international negotiations? Although there is a substan-

tial theoretical literature on bargaining situations and negotiations, there are few studies by

social scientists that include a large number of cases. One exception is the Council of the

EU, as negotiations in this institution have been covered more extensively (see e.g. Mat-

tila and Lane, 2001; Mattila, 2004; Hayes-Renshaw et al., 2006). Furthermore, some of the

findings regarding the EU are inconsistent. Some researchers conclude that the negotiations

are balanced over a larger number of issues, producing neither winners nor losers (Bailer,

2004; Arregui and Thomson, 2009), while others disagree and claim that some countries

perform better than others (Selck and Kaeding, 2004; Selck and Kuipers, 2005). There is

even less agreement about the factors that actually determine bargaining success, and there

has hardly been any work on international bargaining situations that involve most (if not

all) of the countries of the world.

A novel data set on the climate change negotiations, gathered over the past 2 years,

is used to fill this research gap. Using the Cancun Agreements as a reference point, the

positions of a substantial number of the relevant actors were collected and compared with

the actual outcomes of the negotiations as they currently stand.

An introduction to the theoretical background of bargaining situations generally, and

in particular the climate change negotiations, is provided in Section 2. In Section 3, six

hypotheses are derived on the effect of power, salience, bargaining strategies, etc. on ne-

gotiation success. In Section 4, the hypotheses are operationalized and the independent

and dependent variables are described. Section 5 presents the results of the analysis. In

Section 6, it is concluded that external power (measured by total gross domestic product,

GDP), vulnerability to climate change impacts, extremity of negotiation position, and share

of greenhouse gas (GHG) emissions are the most important determinants of success in the

climate change negotiations.

2 Theoretical background

When asked to evaluate the success of various countries in the climate change negotiations,

particularly with respect to the Cancun Agreements, the Canadian delegate stated that:

In Canada’s view, the Cancun Agreement represents significant progress in the

negotiations and was successful in that it reflected a perfectly fine balance of

the views of all Parties. In that context, I would say, that all Parties should be

53

somewhere in the mid-point of your scale [measuring success] in terms of having

had to make some sacrifices on the one hand but seeing their views reflected

throughout the Agreement on the other. This shows the value of international

relations, where countries can give and take on national positions to come up

with compromises that can work for all.1

The quote suggests two ways of thinking about success in international negotiations. Success

might be assessed at the aggregate level, where a bargaining process can be considered

successful if it ends in an agreement, preferably framed by a legal text. However, success

in negotiations might also be thought of in terms of the influence a party exerts on the

outcome of the negotiations, i.e. bargaining success. It is this second notion of success that

is analysed here.

From the perspective of a single country delegation, success can be regarded as the

value contained by a treaty for the state it represents. This can be measured by the distance

between the country’s original position and the negotiated outcome. As Milner (1992, p.468)

notes, underlying this characterization of success as ‘proximity to a negotiated outcome’ is

an assumption that ‘cooperation provides actors with gains or rewards’, and that these

benefits are usually not shared equally among negotiating parties. Thus, bargaining success

is ameasure of how closely a negotiated agreement tracks a country’s preferences (assuming

that countries rank potential outcomes according to related payoffs from high to low). Some

researchers thus favour the term ‘preference attainment’ over that of ‘success’ (see e.g. Traber,

2010). The two terms are used interchangeably here.

As negotiations on climate - a global public good - constitute a Prisoner’s Dilemma, they

are plagued by a severe free-rider problem: while emissions commonly accrue domestically,

the damage caused by unabated emissions is shared with the rest of the world (Carraro

and Siniscalco, 1993, pp.309-311). A country can therefore fail to reduce its own domestic

emissions while at the same time benefit from emissions reductions achieved elsewhere in

the world. From the perspective of a single country, the payoff for free-riding and playing

the uncooperative strategy (e.g. not reducing emissions) can look greater than that of

cooperating with other countries. However, if the result is that all (or most) of the involved

parties do not cooperate, as the Prisoner’s Dilemma suggests, then the worst outcome with

the lowest payoffs collectively (e.g. no reduction in emissions) is reached (Hopmann, 1996,

pp.73-75).

1 This statement, received on 2 March 2011, was made via e-mail in response to a follow-up question askedin a face-to-face interview conducted during one of the many conferences visited by members of the researchteam. The question asked negotiators to assess the negotiation success of a number of other delegationsregarding the Cancun Agreements.

54

The consequences of this dilemma are reflected in the progress of climate negotiations

over previous years. As a result, some researchers portray the possibility of finding a solution

to tackle the global climate crisis in rather pessimistic terms (see Helm, 2008; Brennan, 2009;

Dimitrov, 2010b). However, it is assumed here that coordination between players is possible

and that they can either redefine the rules of the game to overcome the Prisoner’s Dilemma

through reciprocal expectations, or suffer the consequences of failure together (Schelling,

1960, p.107). Given these assumptions, an actor’s negotiation position may not necessarily

be the result of his actual preferences, but may instead reflect strategic choices made to

influence the outcomes of the negotiations (Frieden, 1999, pp.41-45; see also Putnam, 1988;

Moravcsik, 1997; Morrow, 1999). Unfortunately, because the underlying preferences of actors

cannot be measured (Lake and Powell, 1999, pp-18-19), the analysis here uses the concept

of success to investigate their strategic choices.

How successful have single parties been with respect to the Cancun Agreements and

what are the determinants of negotiation success? Providing answers to these questions is

potentially problematic for five reasons. First, the Cancun Agreements are not the final

outcome of the ongoing United Nations Framework Convention on Climate Change (UN-

FCCC) negotiations. A reevaluation of the same data on country positions when compared

to a future treaty might thus result in rather different conclusions than those reached here.

Second, the cross-sectional (rather than longitudinal) design of this study prevents following

the development of country positions and their impact on the various treaties produced by

the UNFCCC over time. Indeed, some of the factors that appear to explain successmight

turn out to be insignificant due to peculiarities of the negotiation process in the specific

period examined here. Third, the data collection effort, described in more detail below, was

conducted over an 11-month period between the UNFCCC meetings in Bangkok (Septem-

ber 2009) and Bonn (August 2010), a period that included the Fifteenth Conference of the

Parties (COP 15) held in Copenhagen. However, owing to the fact that neither the dynamic

nor the scope of the negotiations in Copenhagen changed, it seems reasonable to maintain

that the Copenhagen Accord was not particularly influential on party positions and that

the positions obtained before and after Copenhagen can therefore be assumed to capture

the same negotiation period. Nevertheless, this assumption of preference stability before

and after Copenhagen is crucial for the results presented here and should be kept in mind.

Fourth, the climate change negotiations are embedded in a wider framework of diplomatic

negotiations and international relations. Thus, apparent losers in the UNFCCC negotiation

process might be compensated in a different diplomatic arena.2 Such potential side payments

2 For example, Bolivia, regarded as one of the big losers of the past two COPs in Copenhagen and Cancun,was compensated by the General Assembly of the UN with the recognition of the country’s long-time goal

55

and compensations are not taken into account here. Finally, the proximity of a party’s po-

sition to the final negotiated outcome might to some extent be explained by luck (Barry,

1980a,b). However, as success is measured across several negotiation issues, the likelihood of

consistently achieving high success values through sheer luck in some or all of these is rather

low.

3 Hypotheses

A long-standing debate among scholars of international relations is how gains and rewards

are divided among negotiating parties (for a summary see Katzenstein et al., 1998).

3.1 Power

Prima facie, how powerful a nation is would seem to have a major role to play in negotiation

success. Realism assumes that states care particularly about relative gains, i.e. how well off

they will be compared to other countries after agreeing to a treaty. Neoliberal institutional-

ism maintains that states’ actions are best explained through absolute gains (Powell, 1991,

pp.1303-1306; Milner 1992, pp.470-473), i.e. maximizing their own utility, independently of

the payoffs of other countries.

Realism presumes that states are essentially looking for a balanced distribution of gains

in international negotiations. They

define balance and equity as a distribution of gains that roughly maintains pre-

cooperation balances of capabilities ... No nation will concede political advan-

tages to another nation without the expectation ... of receiving proportionate

advantages in return. (Grieco, 1990, p.47)

Realism thus implies that more powerful nations should be expected to prevail in interna-

tional negotiations.

From the perspective of neoliberal institutionalism, the question arises whether power

can still be used to determine negotiation success. Seeking absolute gains might help ne-

gotiating parties to reach the Pareto frontier, i.e. the set of possible negotiation outcomes

that maximize the gains for all countries combined (Krasner, 1991). However, coming to

an agreement on a specific point along this frontier still requires bargaining and the use

of power. Absolute gains, combined with the kind of reciprocity assumed by many liberal

of making access to clean water and sanitation a basic human right.

56

thinkers, might, after all, have the same result as relative gains - power might matter after

all (Milner, 1992).

Power - as a measure of influencing negotiation outcomes - has both an internal and ex-

ternal dimension. External power resources, such as a country’s economic strength (Drahos,

2003), are “determined by an actor’s environment and therefore difficult to change during

the course of negotiations” (Bailer, 2004, p.100).

H1: Countries with more external power resources are better able to realize their

goals in the climate change negotiations.

Internal power resources, such as delegation size and the negotiation skills of diplomatic

staff, are more subtle than their external cousins and can be changed during negotiations

(Snyder and Diesing, 1977; Antonides, 1991). Internal power resources of governments are

harder to observe and are predominantly linked to the diplomatic delegations of states. A

distinct internal power resource is the bargaining skill of a negotiating party, particularly of

the delegation leader (Snyder and Diesing, 1977; Hopmann, 1996). In the wider debate on

leadership in negotiations, leading-through skills have been called ‘entrepreneurial’ leader-

ship (Skodvin and Andresen, 2006) or ‘instrumental’ leadership (Underdal, 1998), as opposed

to ‘power-based’ and ‘directional’ leadership. Indeed, skilled negotiators behave in a differ-

ent manner during negotiations from their unskilled counterparts (Rackham, 1999). More

specifically, highly skilled representatives are inter alia generally better prepared, ask more

questions, explore more options, set clearer limits, and are more likely to consider long-term

goals than their less skilled counterparts.

Internal power, and its skilful use, can lead to negotiation dynamics that cannot be

explained by appeal to external power resources alone. Nevertheless, the role of internal

power resources is often not considered in power-oriented negotiation analyses (Odell, 2010).

However, this source of power might be a crucial factor in understanding bargaining processes

and explaining negotiated outcomes.

H2: Countries with more internal power resources are better able to realize their

goals in the climate change negotiations.

3.2 Salience

In general, salience indicates the importance of an issue for an actor (Laver, 2001). However,

it is important to recognize that salience has an actor- and an issue-specific

component ... A whole policy field might be deemed particularly important

57

or one could look at individual legislative proposals or even issues within that

proposal. (Warntjen, 2012, p.169)

In the present context, actor-specific salience indicates how important climate change is for a

country, which in turn depends heavily on the expected consequences of a changing climate

for a given country. The stakes are higher for those countries with greater vulnerability

to climate change impacts and they will therefore lobby for higher mandatory emissions

reduction goals.

A second dimension of actor-specific salience is the political vulnerability of countries

to increased global mitigation efforts. Politically vulnerable countries might try to slow

the negotiations down, or they might demand compensations for their expected losses. An

example of such obstructionist behaviour is that of Saudi Arabia (Depledge, 2008).

Negotiating parties with higher actor-specific salience might be able to assert more in-

fluence than others for whom the issue is less salient, simply because it makes it difficult

to ignore them (Fearon, 1994, 1997). In the context of climate change, such audience costs

imply that otherwise dominant negotiation parties may come under pressure to take the

concerns of less powerful but highly affected countries seriously. For example, the future

of the Small Island Developing States is at stake if the climate negotiations fail and the

sea level continues to rise unabated. Thus, having a high level of actor-specific salience can

lend a country a high moral authority, which may become another source of power during

negotiations (Jonsson, 1981).

However, it has been argued that countries with higher salience have tended to bear

a greater portion of the costs associated with a climate treaty, as impatience to solve the

problem has induced them to accept worse deals (Grundig et al., 2001, pp.162-165). In

the EU context, it has been shown that countries that attach higher salience to an issue

have tended to make larger concessions (Schneider, 2005) and that “the urgency that the

negotiating member attributes to an issue decreases rather than increases the bargaining

success” (Bailer, 2004, p.115).

H3a: The higher a country’s vulnerability to climate change impacts, the higher

(lower) the country’s likelihood of success in climate change negotiations.

H3b: The higher a country’s political vulnerability to increased global mitigation

efforts, the higher (lower) the country’s likelihood of success in climate change

negotiations.

Although the effects of climate change on a particular country might explain why some

countries are more concerned about the progress of climate change negotiations than others,

58

less vulnerable countries may still attach high salience to specific issues of particular impor-

tance (see Warntjen, 2012). Evidently, the salience levels highly vulnerable states attach to

different issues can also vary: a low-lying/small island country might hold that establishing

a given emissions reduction target, or that adaptation finance, is the most crucial issue.

H4: An increased level of salience a country attaches to a single issue (issue-

specific salience) increases/decreases that country’s chances to finish negotiations

on the issue successfully.

3.2.1 Extremity of negotiation positions

It has been shown in the context of the European Council that some countries have adopted

extreme positions relative to their negotiation partners to achieve certain goals (Schneider

and Cederman, 1994). Adopted positions not only reflect an actor’s preferences, but also

reflect strategic choices that have been considered necessary to achieve a desired negotiation

goal (see Frieden, 1999; Morrow, 1999). Given the positions of other players in the negotia-

tions, governments can choose their positions accordingly in order to influence negotiations.

Thus, an extreme position on a negotiation issue may be the result of a state making a strate-

gic choice and exaggerating its own sincere preference in order to encourage other players to

move in the preferred direction.

However, adopting an extreme position in multi-party negotiations - as occurred in the

EU negotiations - may increase the likelihood that the offending party will both be left out

by other parties and (as a consequence) end up relatively far away from the subsequent

negotiated outcome (Bailer, 2004). Prima facie, including the majority of the countries of

the world in UNFCCC negotiations will further decrease the probability that a country can

influence the outcome of negotiations by deliberately adopting an extreme position.

H5: Countries who adopt more extreme positions diminish their chances of suc-

cess in the climate change negotiations.

3.3 Hard versus soft bargaining strategies

As well as power resources, governments and their respective diplomatic delegations make

use of a variety of negotiation strategies. Researchers from distinct fields such as business,

psychology, and law have attempted to analyse how negotiation strategies are best deployed

in order to achieve success. In the field of international relations this has largely been

neglected to date (Odell, 2010). Several ways of classifying negotiation strategies have been

proposed (see Lax and Sebenius, 1986; Walton and McKersie, 1991; Carnevale and Pruitt,

59

1992; Hopmann, 1996; Dur and Mateo, 2008). In this article, the distinction of Dur and

Mateo (2008) between hard and soft bargaining strategies is used (see also Bailer, 2012).

The aim of using hard and soft strategies is to move negotiations closer to an outcome

that is preferred by the party employing them. Hard bargaining strategies, i.e. conflictive or

aggressive tactics, aim to benefit one country at the expense of another. Examples include

threats and demands, which strong parties may use to directly influence the negotiating

positions of supposedly weaker states (Matthews, 1989). Soft bargaining strategies, defined

as cooperative or friendly tactics (Dur and Mateo, 2008), aim to advance negotiations for

the mutual benefit of all the parties involved and include proposing solutions in the common

interest (e.g. to overcome a stalemate) and compromise (Jonsson, 1981; Odell, 2002).

Whether a strategic behaviour is successful is heavily dependent on the other attributes

(e.g. power) of the actor. For example, if a great power threatens a client state with economic

sanctions, it is more likely to succeed than if the client state were to threaten the great

power. Hence, although the use of any strategy is theoretically open to each player, they

must be used wisely in international negotiations. Countries with a large amount of external

power might resort to bullying less powerful states into cooperating with their agendas by

using hard bargaining strategies. Equally, countries with little or no external power might

compensate for this lack through the prudent use of soft bargaining strategies. Thus, the

presence of power resources may be responsible for a country’s choice of bargaining strategies.

However, in the UNFCCC negotiations, even small states with little external power can use

hard bargaining strategies due to the fact that every country has the power to veto a given

proposal. Equally, powerful states might wish to demonstrate responsible leadership in the

UNFCCC negotiations and choose to use soft bargaining strategies.3

It has been claimed that hard bargaining strategies can only be employed credibly by

the most powerful countries (Pruitt, 1983). However, others have claimed that the use of

hard bargaining strategies by weaker states might be successful if the issue of negotiation is

particularly salient for them (Habeeb, 1988). Thus, four hypotheses regarding the relations

between external power, hard and soft bargaining strategies, and issue-specific salience, and

how they affect bargaining success, are tested:

H6a: At high levels of external power, an increased use of hard strategies im-

proves bargaining success (positive interaction).

H6b: At low levels of external power, an increased use of soft strategies improves

bargaining success (negative interaction).

3 Some interviewees refused to answer one or more of the questions, some felt unable to answer some ofthem, and some had to cut the interview short due to a lack of time.

60

H6c: At high levels of salience, an increased use of hard strategies improves

bargaining success (positive interaction).

H6d: At high levels of salience, an increased use of soft strategies improves

bargaining success (positive interaction).

4 Operationalization of the model

Two newly created data sets were used to create the model: interview data gathered from

UNFCCC meetings over an 11-month period (from AWG-KP 9/AWG-LCA 7 in Bangkok,

September 2009, to AWG-KP 11/AWG-LCA 9 in Bonn, April 2010) and hand-coded delegate

statement data gleaned from the Earth Negotiations Bulletin (ENB) over a 24-month period

(from COP 13 in Bali, December 2007, to COP 15 in Copenhagen, December 2009). (See

Appendix 1 for further details of the data sets.) Bargaining success was measured over the

following eight climate policy issues:

• Annex I emission reduction targets

• Non-Annex I reduction targets and actions

• Use of market mechanisms

• Mitigation finance

• Mitigation allocation

• Adaptation finance

• Adaptation allocation

• Measuring, reporting, and verification (MRV)

4.1 Measuring success

Two measures of a country’s bargaining success in the climate negotiations (the dependent

variable) were used. The first success measure - the distance between a party’s position as

given in the interviews and the negotiated outcome at the Cancun Agreements in December

2010 - was computed at the issue level. There were 58 countries and eight issues in the

interview data set, and therefore 464 possible values of success. Owing to the fact that

some delegates did not answer all the questions posed in the interviews, only 382 values of

success were obtained. The second success measure used the results of the first measure to

61

generate one single, aggregate success value for each of the 58 countries in the data set. Both

quantities were used in the regressions as dependent variables.

Although it is a fairly common approach in negotiation research (see e.g. Bueno de

Mesquita and Organski, 1994; Bailer, 2004; Steunenberg and Selck, 2006; Thomson and

Stokman, 2006; Arregui and Thomson, 2009), the use of interview data to obtain success

measures carries certain risks, in particular measurement error.4 The obvious problem is

that the maximum distances from the original positions to the negotiated outcomes are dif-

ferent across issues, which hampers comparison across issues. Therefore, bargaining success

(preference attainment) for the eight issues was standardized using the following equation:

suc1ij =

[1− |posij − outj|

maxj

]∗ 100 (1)

where suc1ij is the success of country i on issue j, posij the position of country i on issue j,

outj the outcome of issue j, and maxj the maximal distance a country can have from the

outcome on issue j. The absolute value in the numerator was taken to treat equal distances

from the outcome alike, regardless of the direction in which a country’s position deviated

from the outcome.Without further adjustments, the first measure of success would range

from 0 (most successful negotiation outcome) to 1 (least successful negotiation outcome).

Thus, the resulting scores were subtracted from 1 and multiplied by 100 to yield success

scores ranging from 0 (least successful) to 100 (most successful).

However, using this success score has the potential to be quite misleading, as a country

might wrongly appear to be rather unsuccessful. For example, a delegation might be fairly

satisfied with the outcomes of negotiations if it only really cared about one or two particular

issues (and on which it might score relatively well). Yet such a country might appear to be a

loser in the negotiations if success were measured only using equation (1). As Golub (2010)

has observed, most studies of negotiation success use unweighted success measures as their

dependent variables and are thus flawed. Moreover, including salience as an independent

variable - as this and other studies do - is not sufficient to avoid potentially misleading

results.

The second approach to measure bargaining success thus used salience weights to readjust

the original success measures in order to account for how important each issue has been to

4 Owing to the fact that only one delegate per country was interviewed, the risk of measurement errorwas quite high. To attenuate this problem, the interviews were compared with party data submitted to theUNFCCC (see Appendix 1). Only three of the eight issues contained in the interview data set - emissionsreduction targets, mitigation finance, and adaptation finance - were coded in exactly the same way as in thesubmissions data set and were directly comparable. However, these issues showed relatively high correlationcoefficients of 0.92, 0.70, and 0.69, respectively. Although the overlap is far from perfect, this lends theinterview data a measure of credibility.

62

each country (see Dur, 2008; Golub, 2010). The salience weights were manually coded from

the negotiation protocols reported in the ENBs (IISD, 2009) (see Appendix 1). It was

assumed that the more a country has discussed an issue, the more important it is to it.

(This is congruent with a conception of salience as the ‘level of effort’ a country exerts in

negotiations; see Bueno de Mesquita, 2003, pp.589-590.) Thus, the fraction of statements a

country made on the eight issues during the two-year period, between COP 13 in Bali and

COP 15 in Copenhagen (11 negotiation rounds and 90 negotiation days), was used as the

salience weight.5 The following formula, suggested by Hinich and Munger (1997, p.80), was

used to calculate the second measure of bargaining success:6

suc2i =√

[suc1i]T Ai [suc1i] (2)

where suc2i is a country’s overall measure of success, suc1i is a vector of success for all issues

computed in equation (1), and Ai is a matrix containing the salience weights in the diag-

onal elements.7 The off-diagonal elements represent interaction terms, which measure how

much success on one issue depends on the outcome of another. For example, small island

states might value financial aid highly. However, if global emissions levels are not sufficiently

reduced and the states cease to exist because of rising sea levels, even very high amounts

of adaptation finance will not help them. Accordingly, reducing emissions and financial aid

5 Merely counting how often each issue has been mentioned by a party during negotiations and usingthis as the salience measure would not be appropriate without further adjustment due to the fact that thetotal number of interventions has varied widely across negotiating parties. This would result in incomparablesuccess scores when applied to equation (2). Hence, the salience weights were standardized by dividing thenumber of statements a country has made on an issue by the total number of interventions it has made onthe eight relevant issues. The resulting success measures thus sum up to 1, regardless of the total number ofinterventions a country has made, thus preventing distortion of the weights for each country and permittingcomparison across countries. The salience weights used thus consisted of the fraction of all statements acountry has made on the different issues.

6 Note that instead of the raw distances to measure success proposed by Hinich and Munger (1997), thestandardized distances obtained using equation (1) were used.

7 As mentioned, some delegates did not answer all the questions. In such cases, the calculated successscore was biased downwards, i.e. the country appeared to be less successful than it may in fact have been(unless the missing success score would in fact have been 0). It was assumed that success for the missingissues was similar to those for which success values were obtained, and the second measure of success thushad to be adjusted. As the salience weights were obtained for all issues, the author was able to computethe combined weight of the missing values and adjust the obtained success score accordingly (see examplebelow in this endnote). For countries with no missing scores (the majority of the countries in the data set),the weights of missing values were zero and thus the value for the second measure of success did not change.If, however, one issue was missing and the salience weight for this issue was, say, 0.2, the country scorefor success was too small (unless it did not achieve any success on the issue). If the country score for theremaining issues was, say, 45, then this was adjusted accordingly (e.g. 45/(1 – 0.2) = 56.25). This method- which worked particularly well for those countries for which information on only one or two positions wasmissing - yielded the average success score of the issues and the bias vanished. For countries missing muchor most of the data, however, this method would be very misleading. Fortunately, there were only a fewcountries for which there were many missing data points.

63

theoretically demonstrates positive complementarity, but, unfortunately, there is no proce-

dure that can be used to calculate the size of the interactions (and in many cases it is even

theoretically difficult to assess whether complementarity exists at all between two issues). It

was therefore assumed that the issues are separable, and the interaction terms were set to

equal zero. Using matrix algebra, the resulting second measure of success is thus one single

value for each country (for a list of all countries from most to least successful, see Appendix

3).8

4.2 Determinants of success

The four putative determinants of bargaining success - the power of a state, the salience of

an issue to a state, the extremity of a state’s negotiation position, and the use of hard/soft

bargaining strategies - were operationalized for the analysis as detailed below (see Table 1

for descriptive statistics for each independent variable).

Table 1: Descriptive statistics of the dependent and independent variables

Variable Obs. Mean s.d. Min. Max. SourceSuccess 1 382 52.50 32.01 0 100 Interview dataSuccess 2 58 46.80 22.59 0 95 Interv. & ENBLog of GDP 58 26.05 2.65 18.66 30.42 World BankCO2 emissions 58 1.33 4.03 0 23.55 UN(2011)Vulnerability 58 3.36 0.73 1.67 4.90 Kaly et al. (2004)Salience 348 16.66 13.56 0 100 ENB dataDelegation size 57 43.42 35.81 5 173 UNFCCC(2010)Pos. extremity (issue-levle) 367 18.98 14.08 0.17 85.17 Interview dataPos. extremity (mean) 57 19.34 9.74 1.97 62.08 Interview dataHard strategies 58 4.06 1.66 1 9 Interview dataSoft strategies 58 4.99 1.64 1 9 Interview data

4.2.1 Power

External power: One of the main sources of a country’s external power is economic

power, which can be regarded as a coercive force in international negotiations (Waltz, 1979;

8 For calculating the second measure of success, the issues of mitigation allocation and adaptation allo-cation were omitted from the analysis, as salience weights were not obtainable for them. Hence, only 288 ofthe 382 success values originally obtained using equation (1) were used to construct the values reported inAppendix 3.

64

Keohane, 1984). Large economies such as those of China and the US can make use of

their size in order to force other countries with close economic ties to lean their way in the

negotiations. Thus, the logarithm of total GDP (World Bank, 2011) was used as the first

measure of power.

Internal power: This was operationalized through the use of a country’s delegation size,

information that was obtained from the official UNFCCC participants list at the Cancun

Climate Change Conference in December 2010 (UNFCCC, 2010). It might be argued that

richer (more economically powerful) countries can send more people to international confer-

ences and that delegation size should therefore be used to represent the external power of

a state. However, governments are able to choose and control the size of their delegation.

A relatively high positive correlation is found for delegation size with both bargaining skills

and total GDP of over 0.5, indicating that the size of a diplomatic delegation is indeed a

function of power, but also an indicator of a party’s bargaining skills. As control for power

was undertaken in all models, delegation size represents internal power in this article.

4.2.2 Salience

Actor-specific salience: Given the two dimensions of actor-specific salience, two opera-

tionalizations were required. First, the extent to which a country is vulnerable to climate

change impacts (hypothesis H4a) was measured using the Environmental Vulnerability In-

dex (EVI) developed by the South-Pacific Applied Geoscience Commission (SOPAC) and the

United Nations Environmental Programme (UNEP). In total, the EVI measures 50 indices,

13 of which are used to construct a sub-index for climate change vulnerability (see Kaly

et al., 2004). Although the EVI has been criticized for a number of reasons (see Barnett

et al., 2008) - for example, on the grounds that it is impossible to quantify complex social-

ecological processes - this criticism is not particular to the EVI and applies to all indices

that measure vulnerability.

Second, political vulnerability wasmeasured by a country’s share of global GHG-emissions.

This variable was constructed using the CO2 emissions of all countries as reported by the

(UN, 2011) in the Millennium Development Goals Indicators (MDGIs).

Issue-specific salience: Issue-specific salience was measured by calculating the fraction

of statements a country has made on the eight issues, listed above, from Bali in 2007 to

Copenhagen in 2009, as reported in the ENB (IISD, 2009). Some parties, particularly smaller

countries, only made a limited number of interventions, even over the full 2-year period.

Therefore, these individual statements made up only 50the final salience scores, while the

65

remainder was derived from group statements of a country’s most important negotiation

coalition. This issue-specific salience was only included as an independent variable in the

models that used the first measure for success. It was also used to construct the Ai matrix

of equation (2), used to calculate the values of the second measure of success.

4.2.3 Negotiation positions

The extremity of a country’s position on a negotiation issue was measured by the distance to

the mean position in the data set. Thus, extremity values for all eight issues of interest were

computed for, and tested with, the first success measure. The mean extremity values for

each country, over the six variables, were then used to construct the second success measure

and used to test hypothesis H5.

4.2.4 Hard and soft bargaining strategies

During the first round of interviews, negotiators were asked to assess, on a scale from 1

(never) to 9 (very often), how often they used 10 (three soft, seven hard) kinds of negotia-

tion strategies. The three soft bargaining strategies were proposals in the common interest,

exchanges of concessions, and expressions of understanding for other country’s positions.

The seven hard bargaining strategies were threats, promises, direct criticisms, open decla-

rations not to change a position, demands for concessions from others, ignoring demands of

others, and hiding one’s real negotiation objectives.9 As with measuring a country’s negoti-

ation positions, quantifying their strategic behaviour using the interview data risks similar

measurement error.10 The indicators for the bargaining strategies used in the calculations

below were derived by taking the mean over all hard and then all soft bargaining strate-

gies. To operationalize hypothesis H3, interaction terms between the independent variables

of hard/soft bargaining strategies and external power, and between hard/soft strategies and

salience, were constructed, yielding four interaction terms.

5 Results

Table 2 lists the main findings. All the models used ordinary least squares (OLS) with

clustered standard errors to account for the particular structure of the climate change nego-

9 See (Bailer, 2012) for a more general discussion of hard and soft bargaining strategies.10 Unfortunately, the use of hard and soft strategies by a state cannot be checked as easily as the negoti-

ation positions using other data sources. This is because coding the strategies on the basis of the ENB andUNFCCC submission data is very difficult and highly unreliable.

66

tiations, i.e. the collaboration of negotiating parties in coalition groups.11 Models 1-4 used

the first success measure as the dependent variable, and Models 5-7 used the second success

measure.

5.1 External power

The external power variable, total GDP (in logarithmic form), was highly significant in

almost all the model specifications and suggests, unsurprisingly, that an increase in power

improves the probability of success in climate change negotiations.12 The size of the effect

was considerable across the models. Using the coefficient shown in Model 1 of Table 2,

ceteris paribus, a success score more than 30 points higher at the maximum of total GDP

than that for the lowest levels of GDP would be expected. The results provide relatively

strong evidence for the validity of hypothesis H1.

5.2 Internal power

In all the models tested, the influence of the internal power of a country, measured by

delegation size, fell far short of conventional significance levels. In addition, the size of the

coefficient across all models was very small. Only when total GDP was omitted from the

regressions did the coefficient on delegation size become significant, indicating - contrary to

the assumption (see Section 4.2.1) - that it was acting as a proxy for external power rather

than internal power. However, when external power was controlled for by other means, the

effect of internal power on bargaining success was negligible. The results therefore cast doubt

on the validity of hypothesis H2.

5.3 Actor-specific salience

The first actor-specific salience factor, vulnerability to climate change impacts, persistently

showed significant positive coefficients across the models (with the exception of Model 7),

11 The coalition groups for which cluster-corrected standard errors were computed were the EU, theEnvironmental Integrity Group (EIG), the Umbrella Group, G77 and China, the African Group, the LDCs,and the Alliance of Small Island States (AOSIS). Note that each country was only assigned to one of theseclusters (that indicated as the most important negotiation group during the interviews). Note, however, thatsome countries are members of more than one coalition. For example, although Comoros is a member ofAOSIS, the LDCs, the African Group, and G77 and China, the interview data suggested that it has regardedAOSIS as the most important coalition group and the one most likely to achieve its goals.

12 When the effect of total GDP on choice of negotiation strategy was modelled, the significance of totalGDP diminished and in some cases even vanished, despite the fact that the interaction terms themselveswere not significant. In the case of Model 3, the significance remained over the whole range of both hard andsoft strategies, although at lower P values than without interactions. Model 7, however, was the exception,as it exhibited no significance for total GDP over the entire range of both hard and soft strategies.

67

Table 2: Regression results using success measure as dependent variable

Dependent variable: Success 1 Dependent variable: Success 2Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7

Log of GDP 3.79∗∗∗ 3.40∗∗∗ 4.43∗∗ 3.97∗∗∗ 5.91∗∗∗ 5.67∗∗∗ 5.41(1.01) (0.70) (1.96) (1.01) (1.44) (1.03) (4.09)

Del. size 0.01 0.02 0.02 −0.03(0.08) (0.08) (0.09) (0.09)

Vulnerability 4.99∗∗ 3.18∗ 5.42∗∗ 4.93∗∗ 5.11∗∗ 5.01∗ 4.64(1.97) (1.69) (2.29) (1.90) (2.39) (2.61) (3.06)

Emissions −0.76∗∗ −0.89∗ −0.58 −0.66∗ −1.38∗∗ −1.43∗∗ −1.71∗∗

(0.34) (0.49) (0.33) (0.34) (0.43) (0.43) (0.67)Salience −0.14 −0.14 −0.60

(0.24) (0.24) (0.32)Pos. extremity −0.91∗∗∗ −0.58∗∗∗ −0.98∗∗∗ −1.00∗∗∗ −0.74∗∗∗ −0.74∗∗∗ −0.69∗∗∗

(0.15) (0.10) (0.13) (0.13) (0.16) (0.16) (0.12)Soft strategies 4.11 −3.64∗∗∗ −9.38

(12.51) (0.88) (19.81)Hard strategies −1.55 −0.16 5.00

(9.24) (1.92) (22.63)GDP*soft strat. −0.19 0.42

(0.49) (0.80)GDP*hard strat. 0.03 −0.21

(0.35) (0.91)Salience*soft 0.18∗∗

(0.08)Salience*hard −0.03

(0.08)(Intercept) −41.25 −31.54 −50.93 −30.13 −100.48∗∗∗ −95.42∗∗∗ −90.53

(29.97) (20.53) (53.46) (27.91) (35.58) (26.63) (96.56)N 278 368 278 278 56 56 56R2 0.14 0.09 0.15 0.16 0.56 0.56 0.56adj. R2 0.12 0.08 0.11 0.12 0.50 0.52 0.49Resid. sd 31.71 30.85 31.89 31.68 15.79 15.50 16.02Clustered standard errors are given in parentheses. Models include fossil fuel rents as additional controls∗ indicates significance at p < 0.10, ∗∗ at p < 0.05, and ∗∗∗ at p < 0.01

indicating that higher vulnerability has tended to increase bargaining success, thus providing

some support for hypothesis H3a. The size of the coefficient was fairly stable at around 5 for

both success measures, yet the significance level dropped somewhat when using the second

success measure. Employing the coefficient of 4.99 found in Model 1 for computational

purposes, it was found that, between the minimum (1.67) and the maximum (4.9) observed

vulnerability levels, the expected difference of success was 16.12 (i.e. slightly less than one-

68

sixth of the maximum possible success difference in the data set).

A negative significant effect of political vulnerability was found, indicating that a higher

share of global GHG emissions has been detrimental to a country’s bargaining success, thus

undermining hypothesis H3b. Although the significance of the coefficient was somewhat

weaker than in the GDP case, the persistence across the models accords some credibility to

the results (even the exception, Model 3, was close to significance).

It is interesting to note that those countries that are highly vulnerable to climate change

impacts appear to have some negotiation leverage over those countries that are large GHG

emitters. However, the view that the vulnerability of a state to climate change may assist

it in pressurizing large GHG-emitting states to cooperate would be too simplistic for two

reasons. First, large GHG emitters also tend to be highly powerful. Hence, the negative effect

of higher CO2 emissions is likely to be offset by the strong positive effect that external power

has on negotiation success. Equally, it is likely that the positive effect of vulnerability on

bargaining success will only partly make up for the lack of power of small states. Second, and

following on from this, some powerful countries are also relatively vulnerable (as measured

by the EVI) to climate change impacts (e.g. Germany at 4.15, China at 3.85). Accordingly,

there is an even more direct, salience-related source for these countries - vulnerability to

climate change impacts - that offsets the negative effect of emitting GHGs. Consequently,

the result that climate change has had positive effects on UNFCCC negotiation success

should not be interpreted as inconsistent with the dominant view that the most powerful

countries are the most influential in the climate negotiations.

5.4 Issue-specific salience

Using the first success measure, issue-specific salience was not significant across any of the

model specifications tested, thus casting doubt on hypothesis H4. However, in one case,

the issue-specific salience coefficient did come close to significance when it was tested for

interactions with hard/soft bargaining strategies. To further investigate this coefficient, the

appropriate formula presented by (see web appendix, Table 1, case 3 in Brambor et al.,

2006) was used. At very low levels of using soft bargaining strategies, issue-specific salience

displays negative marginal effects (see Figure 1). The marginal effect of issuespecific salience

was insignificant for all other combinations of bargaining strategies. The left-hand panel of

Figure 1 shows that with soft bargaining strategies fixed at 1, salience negatively impacted

success at all levels of hard strategies. At higher levels of soft strategy use, shifting the

whole curve upward, the marginal effects of salience became insignificant. Changing the

level of hard strategies in this setting had little influence on this salience effect due to the

69

rather small effect of the interaction coefficient. The right-hand panel of Figure 1 confirms

this. Levels of soft strategic use below the value of 3 cause the marginal effects of salience

to be negative. Changing the level of hard strategies (fixed in the graph at 5) makes little

difference, as the curve barely shifts. There is therefore some evidence that at higher levels of

issue-specific salience, negotiators should employ soft bargaining strategies to emphasize the

importance of a negotiation issue, lest their concerns be ignored. Thus, the results support,

to some extent, hypothesis H6d. On the other hand, the use of hard bargaining strategies

does not appear to significantly change the salience effect on bargaining success, thus casting

doubt on hypothesis H6c.

Figure 1: Marginal effects of salience

1 2 3 4 5 6 7 8 9

−2

−1

0

1

2

Hard Strategies

Mar

gina

l Effe

ct

Soft Strategies fixed at 1

1 2 3 4 5 6 7 8 9

−1

0

1

2

Soft Strategies

Mar

gina

l Effe

ct

Hard Strategies fixed at 5

5.5 Extremity of positions

Taking extreme positions during climate change negotiations has greatly diminished bargain-

ing success. A move away from the average position in the sample reduced the success score

by between 0.58 (Model 2 of Table 2) and 1.00 (Model 4). This holds true for both measures

of success and the corresponding extremity variable. Furthermore, the coefficient on extreme

positions was significant at the 99% confidence level across all tested models, supporting the

validity of hypothesis H5. This is probably the strongest finding of this study.

For example, the US had the second highest average extremity score across the eight issues

70

(52.8), which is one of the reasons why the most powerful country in the world was relatively

unsuccessful at the negotiations, with one of the lowest overall success scores in the data set

(see Appendix 3).13 However, it should be kept in mind that the Cancun Agreements were

assumed as the reference point. Owing to the perceived failure of the Copenhagen Climate

Change Conference, an increase in domestic scepticism regarding the negotiations, and the

sustained economic crisis, the Obama administration consequently paid little attention to

the issue of climate change (see Brewer, 2011, pp.7-10). Moreover, the Cancun Agreements

must be seen as a provisional step in the continuing climate negotiations. Hence, there is

still time for powerful actors to achieve their goals. The US could still be an influential and

indeed crucial player if it so chose.

5.6 Hard and soft bargaining strategies

The coefficients on both hard and soft strategies, when interacted with economic power,

were insignificant and had comparably large standard errors. The same was true for the

corresponding interaction effects. Hence, the use of these strategies by a country has not

significantly increased bargaining success if external power is taken into account. Thus,

the results cast doubt on hypotheses H6a and H6b. As already mentioned in Section 5.4,

a similar result was obtained for the use of hard bargaining strategies when issue-specific

salience is taken into account. However, when issue-specific salience is taken into account,

soft bargaining strategies were both highly significant (Model 4, Table 2), as can be seen in

Figure 2. Although an increased use of soft strategies exhibits negative marginal effects at

very low levels of issue-specific salience, it exhibits positive marginal effects in combination

with salience levels above 70. Thus the data lend some support to hypothesis H6d.

6 Conclusion

The presented analysis suggests that bargaining success has been positively affected by a

country’s external power and vulnerability to climate change impacts and negatively af-

fected by the extremity of a country’s negotiation position and its share of GHG emissions.

Additionally, there is some support for the view that the use of soft bargaining strategies

in the climate change negotiations by a country has increased the likelihood of bargaining

success for issues that have been highly salient to it. The results obtained for power, vul-

nerability to climate change impacts, the share of GHGs, and the extremity of positions,

13 Other powerful negotiating parties appear to have taken comparatively moderate positions. For exam-ple, the average extremity value for China was 20.6, India 16.7, the EU 12.9, and the Russian Federation34.7.

71

Figure 2: Marginal effects of salience

0 20 40 60 80 100

−10

0

10

20

Salience

Mar

gina

l Effe

ct

in particular, are rather stable over various model specifications for both of the dependent

variables proposed above.

Some of the large countries, in particular the US and China, were not particularly suc-

cessful in the negotiations regarding any of the eight issues examined (see Appendix 3 for

detailed statistics). Although both are highly powerful, their extremely large share of GHG

emissions, and the extremity of their negotiation positions (especially for the US), were detri-

mental to their bargaining success. This is rather surprising given that most accounts of the

climate change negotiations claim that powerful countries, such as the US, China, India, and

Brazil, have the most influence (see e.g. Dimitrov, 2010a; Cozier, 2011). So how can the

results of this study be aligned with the usual interpretation of the negotiation process?

The effects of the independent variables on the bargaining success of countries involved

in the UNFCCC negotiations over the eight issues specified in Section 4 were assessed at the

aggregate level (e.g. Annex I emissions reduction targets). If the UNFCCC specifies these

issues at the aggregate level, then Member States can accept an official document - in this

case the Cancun Agreements - without agreeing to any binding commitments individually.

For example, as a part of the Cancun Agreements, developed countries committed (starting

in 2020) to jointly mobilize US$100 billion a year for both adaptation and mitigation. This

bargaining solution makes the US - which according to the interview and submissions data

prefers paying for adaptation and mitigation using a mix of voluntary donations and con-

tributions through market mechanisms - an apparent loser on the issues of mitigation and

72

adaptation finance as the (aggregate) commitment of developing countries to donate $100

billion annually is far away from the US bargaining position. However, it is likely that the

US will actually only provide a small proportion of the $100 billion each year. Similarly,

although it is recognized in the Cancun Agreements that a 25-40% reduction of emissions

(below 1990 levels) by Annex I Parties is needed by 2020, there is little inclination by these

Parties to make any such binding commitments.

For an environmental treaty to be effective, it must ensure both participation and com-

pliance, while setting meaningful targets (Barrett, 2008, pp.240-241). If the UNFCCC is

unable to deliver such results, the whole negotiation process is in danger of losing both legit-

imacy and credibility. It has been shown that powerful countries such as the US, Russia, and

China have not been particularly successful (with respect to the eight issues analysed) in the

climate change negotiations. Thus, this could be an indication that these countries have lost

faith (if there was any to begin with) in the legitimacy and credibility of the negotiations. At

the least, the analysis might indicate that these countries have not made as much of an effort

to cooperate and succeed in the negotiations as they might have if they still considered the

negotiation process fully functional. A less pessimistic interpretation would be that, because

the Cancun Agreements represents an intermediate stage of the international climate change

negotiations, the results of the analysis might not accurately reflect the countries’ attitudes

or preferences towards the UNFCCC negotiations. Although a future climate treaty may

of course more accurately reflect the preferences of the major players in the climate change

negotiations, it is clear from the analysis that the negotiation process hangs in the balance.

73

References

Antonides, G. (1991). Psychological variables in negotiation. Kyklos 44 (3), 347–362.

Arregui, J. and R. Thomson (2009). States’ bargaining success in the European Union.Journal of European Public Policy 16 (5), 655–676.

Bailer, S. (2004). Bargaining success in the European Union: The impact of exogenous andendogenous power resources. European Union Politics 5 (1), 99–123.

Bailer, S. (2012). Bargaining strategies in climate change negotiations. Climate Policy 12 (5),534–551.

Barnett, J., S. Lambert, and I. Fry (2008). The hazards of indicators: Insights from the Envi-ronmental Vulnerability Index. Annals of the Association of American Geographers 98 (1),102–119.

Barrett, S. (2008). Negotiation strategies for a post-Kyoto regime. BC Journal SpecialVolume, Fall 2008, 9–20.

Barry, B. (1980a). Is it better to be powerful or lucky? Part I. Political Studies 28 (2),183–194.

Barry, B. (1980b). Is it better to be powerful or lucky? Part II. Political Studies 28 (3),338–352.

Brambor, T., W. R. Clark, and M. Golder (2006). Understanding interaction models: Im-proving empirical analyses. Political Analysis 14 (1), 63–82.

Brennan, G. (2009). Climate change: A rational choice politics view. The Australian Journalof Agriculture and Resource Economics 53, 309–326.

Brewer, P. R. (2011). Polarisation in the USA: Climate change, party politics, and publicopinion in the Obama era. European Political Science 11 (1), 7–17.

Bueno de Mesquita, B. (2003). Principles of International Politics. People’s Power, Prefer-ences, and Perceptions. Washington, D.C.: CQ Press.

Bueno de Mesquita, B. and A. F. Organski (1994). Policy outcomes and policy interventions:An expected utility analysis. In B. Bueno de Mesquita and F. N. Stokman (Eds.), EuropeanCommunity Decision Making: Models, Applications, and Comparisons. New Haven, CT:Yale University Press.

Carnevale, P. J. and D. J. Pruitt (1992). Negotiation and mediation. Annual Review ofPsychology 43 (1), 531–82.

Carraro, C. and D. Siniscalco (1993). Strategies for the international protection of theenvironment. Journal of Public Economics 52, 309–328.

74

Cozier, M. (2011). Restoring confidence at the Cancun Climate Change Conference. Green-house Gases: Science and Technology 1 (1), 8–10.

Depledge, J. (2008). Striving for No: Saudi Arabia in the climate change regime. GlobalEnvironmental Politics 8 (4), 9–35.

Dimitrov, R. (2010a). Inside Copenhagen: The state of climate governance. Global Environ-mental Politics 10 (2), 18–24.

Dimitrov, R. (2010b). Inside UN climate change negotiations: The Copenhagen conference.Review of Policy Research 27 (6), 795–821.

Drahos, P. (2003). When the weak bargain with the strong: Negotiations in the world tradeorganization. International Organization 8 (1), 79–109.

Dur, A. (2008). Measuring interest group influence in the EU. European Union Politics 9 (4),559–576.

Dur, A. and G. Mateo (2008). Bargaining power and negotiation tactics: The negotiationson the EU’s financial perspectives, 2007-2013. UCD Dublin European Institute WorkingPaper 08-2 .

Fearon, J. D. (1994). Domstic political audiences and the escalation of international disputes.American Political Science Review 90 (7), 15–35.

Fearon, J. D. (1997). Signaling foreign policy interests: Tying hands versus sinking costs.The Journal of Conflict Resolution 41 (1), 68–90.

Frieden, J. (1999). Actors and preferences in international relations. In D. Lake and R. Powell(Eds.), Strategic Choice and International Relations, pp. 39–76. Princeton, New Jersey:Princeton University Press.

Golub, J. (2010). Relative gains in the European Union: The effect of preferences, rules,and norms on legislative decisionmaking. Paper presented at the Center for Comparativeand International Studies, 29 April 2010.

Grieco, J. (1990). Cooperation Among Nations: Europe, America, and Non-Tariff Barriersto Trade. Cornell University Press.

Grundig, F., H. Ward, and E. P. Zorick (2001). Modelling Global Climate Negotiations,Volume 153-182 of International Relations and Global Climate Change. Cambridge, MA:MIT Press.

Habeeb, W. M. (1988). Power and Tactics in International Negotiations: How Weak NationsBargain with Strong Nations. Baltimore: Johns Hopkins University Press.

Hayes-Renshaw, F., W. van Aken, and H. Wallace (2006). When and why the Council ofMinisters of the EU votes explicitly. Journal of Common Market Studies 44 (1), 161–194.

75

Helm, D. (2008). Climate-change policy: Why has so little been achieved? Oxford Reviewof Economic Policy 24 (2), 211–238.

Hinich, M. and M. Munger (1997). Analytical Politics. Cambridge, New York, Melbourne:Cambridge University Press.

Hopmann, P. T. (1996). The Negotiation Process and the Resolution of International Con-flicts. Columbia: University of South Carolina Press.

IISD (2007-2009). Earth negotiation bulletin. http://www.iisd.ca/vol12.

Jonsson, C. (1981). Bargaining power: Notes on an elusive concept. Cooperation and Con-flict 16 (4), 249–57.

Kaly, U., C. Pratt, and J. Mitchell (2004). The demonstration environmental vulnerabilityindex (EVI) 2004. Technical report, SOPAC Technical Report 384.

Keohane, R. O. (1984). After Hegemony: Cooperation and Discord in the World PoliticalEconomy. Princeton: Princeton Universtiy Press.

Krasner, S. D. (1991). Global communications and national power: Life on the paretofrontier. World Politics 43 (3), 336–366.

Lake, D. and R. Powell (1999). International relations: A strategic-choice approach. InD. Lake and R. Powell (Eds.), Strategic Choice and International Relations, pp. 3–38.Princeton, New Jersey: Princeton University Press.

Laver, M. (2001). Positions and salience in the policies of political actors. In M. Laver (Ed.),Estimating the Policy Positions of Political Actors. London: Routledge.

Lax, D. A. and J. K. Sebenius (1986). The Manager as Negotiator. Bargainig for Cooperationand Competitive Gain. New York, London: The Free Press.

Matthews, S. A. (1989). Veto threats: Rhetoric in a bargaining game. The Quarterly Jounalof Economics 104 (2), 347–69.

Mattila, M. (2004). Contested decisions: Empirical analysis of voting in the European UnionCouncil of Ministers. European Journal of Political Research 43 (1), 29–50.

Mattila, M. and J.-E. Lane (2001). Why unanimity in the Council? A roll-call analysis ofCouncil voting. European Union Politics 2 (1), 31–52.

Milner, H. (1992). International theories of cooperation among nations: Strengths andweaknesses. World Politics 44 (3), 466–496.

Moravcsik, A. (1997). Taking preferences seriously: A liberal theory of international politics.International Organization 51 (4), 513–553.

Morrow, J. (1999). The strategic setting of choices: Signaling, commitment, and negoti-ation in international politics. In D. Lake and R. Powell (Eds.), Strategic Choice andInternational Relations, pp. 77–114. Princeton, New Jersey: Princeton University Press.

76

Odell, J. S. (2002). Creating data on international negotiation strategies, alternatives andoutcomes. International Negotiation 7 (1), 39–52.

Odell, J. S. (2010). Three islands of knowledgs about negotiations in international organi-zations. Journal of European Public Policy 17 (5), 619–32.

Powell, R. (1991). Absolute and relative gains in international relations theory. The AmericanPolitical Science Review 85 (4), 1303–1320.

Pruitt, D. G. (1983). Strategic choice in negotiations. American Behavioral Scientist 27 (2),167–194.

Putnam, R. D. (1988). Diplomacy and domestic politics: The logic of two-level games.International Organization 42 (3), 427–460.

Rackham, N. (1999). The behavior of successful negotiators. In R. Lewicki, D. Saunders, andJ. Minton (Eds.), Negotiation: Readings, Exercises, and Cases. Boston, MA: Irwin/TheMcGraw-Hill Companies.

Schelling, T. C. (1960). The Stategy of Conflict. Cambridge, London: Harvard UniversityPress.

Schneider, G. and L.-E. Cederman (1994). The change of tide in political cooperation: Alimited information model of European integration. International Organization 48 (04),633–662.

Selck, T. and M. Kaeding (2004). Divergent interests and different success rates:France, Ger-many, Italy, and the United Kingdom in EU legislative negotiations. French Politics 2 (1),81–95.

Selck, T. and S. Kuipers (2005). Shared hesitance, joint success: Denmark, Finland, andSweden in the European Union policy process. Journal of European Public Policy 12 (1),157–176.

Skodvin, T. and S. Andresen (2006). Leadership revisited. Global Environmental Poli-tics 6 (3), 13–27.

Snyder, G. H. and P. Diesing (1977). Conflict among Nations. Bargaining, Decision Makingand System Structure in International Crises. Princeton: Princeton University Press.

Steunenberg, B. and T. J. Selck (2006). Testing procedural models of EU legislative decision-making. In R. Thomson, F. N. Stokman, C. H. Achen, and T. Konig (Eds.), The EuropeanUnion Decides, pp. 54–85. Cambridge, UK: Cambridge University Pres.

Thomson, R. and F. N. Stokman (2006). Research design: Measuring actors positions,salience and capabilities. In R. Thomson, F. N. Stokman, C. H. Achen, and T. Konig(Eds.), The European Union Decides, pp. 54–85. Cambridge, UK: Cambridge UniversityPres.

77

Traber, D. (2010). Preference attainment in Swiss legislative decisioin-making. Draft pre-pared for the annual meeting of the Swiss Political Science Association, University ofGeneva, 30.3.2010.

UN (2011). Millennium Delelopment Goals Indicators. Carbon dioxide emissions (CO2).http://mdgs.un.org/unsd/mdg/data.aspx.

Underdal, A. (1998). Leadership in international environmental negotiations. In A. Underdal(Ed.), The Politics of International Environmental Management, pp. 101–128. Dordrecht,Boston, London: Kluwer Academic Publishers.

UNFCCC (2010). List of participants. http://unfccc.int/resource/docs/2010/cop16/eng/inf01p01.pdf(Part 1) and http://unfccc.int/resource/docs/2010/cop16/eng/inf01p02.pdf (Part 2).

Walton, R. E. and R. B. McKersie (1991). A Behavioral Theory of Labor Negotiations. AnAnalysis of a Social Interaction System (2nd ed.). Ithaca, NY: ILR Press.

Waltz, K. N. (1979). Theory of International Politics. New York: Random House.

Warntjen, A. (2012). Measuring salience in EU legislative politics. European Union Poli-tics 13 (1), 168–182.

World Bank (2011). World Development Indicators. http://data.worldbank.org/data-catalog/worlddevelopment-indicators.

78

Appendix 1: Description of data sets

The three data sets collected over the past two years by the research team are described.

Note that full descriptions of the single variables used in this article are not given.

Interview data

A total of 60 interviews with 56 different country delegations, plus a delegate from the EU

and an expert and close adviser of the Least Developed Countries (LDCs), were conducted

(note that Indonesia and Bangladesh were interviewed twice). The interviews covered dele-

gates from countries in all five continents and world geographical regions, and all UNFCCC

coalition groups, and were therefore considered to be representative of all the possible posi-

tions in the UNFCCC negotiations.

The interviews were mostly conducted face to face, and took place during the three

UNFCCC negotiation meetings in 2009 (Bangkok, Barcelona, and Copenhagen) and the

three meetings in Bonn in 2010. Some interviews were conducted by phone during the

same period of time. Interviews were divided into three different blocks of questions (plus

a short introductory block on delegation size and composition) regarding country position,

negotiation strategy, and influence of institutions and stakeholders. Delegates were first asked

to indicate their own country’s position on eight negotiation issues during the negotiations

(see Appendix 2). The issues and the definition of their extreme points were previously

identified by the project team, which included Axel Michaelowa, a close UNFCCC negotiation

observer and climate expert. All issues, apart from mitigation targets, were measured on a

scale from 0 to 100. Delegates were requested to provide the value on this scale that best

reflected their country’s position. The scale was designed in each case to cover all possible

positions on the relevant issue. For example, for the question “Who should primarily finance

the action on adaptation?”, a value of 0 corresponded to “Voluntary financing by the private

sector”, and a value of 100 corresponded to “Mandatory financing of around $100bn per year

by industrialized countries”. In addition, respondents were asked to describe their position

in words in order to aid the identification of the country position on the scale. Second,

negotiators were asked to rate, on a scale from 0 (never) to 9 (very often), how often their

delegations had used ten kinds of bargaining strategies (see Bailer, 2012, for a more detailed

description). Finally, negotiators were asked to rate, on a scale from 0 (very low) to 9 (very

high), the influence of 15 institutions/stakeholders (e.g. the economics and environment

ministries, the national parliament, different industries, NGOs, national and international

media, the public) on their country’s negotiation positions.

79

ENB data

The Earth Negotiations Bulletin (ENB), published by the International Institute for Sus-

tainable Development (IISD), has been reporting on a host of international environmental

meetings and negotiations since 1992. An issue of the ENB is published for every day of

the UNFCCC negotiations, and includes a summary (which is usually one sentence long) of

every statement made by the negotiation parties in the publicly accessible meetings.

All the ENBs on the climate change negotiations during the 2-year period from COP

13 in Bali (December 2007) to COP 15 in Copenhagen (December 2009) were hand-coded.

This period contained 11 negotiation rounds and a total of 90 negotiation days. For every

statement reported, four properties were recorded: (i) who made the statement; (ii) which

segment of the negotiations it was made in (e.g. COP, AWG-KP, COP/MOP, AWG-LCA,

SBI, SBSTA); (iii) themain topic of the statement (e.g.mitigation, adaptation, finance, mea-

suring, reporting, and verification), any subcategories, and (if applicable) the kind of bar-

gaining strategy used; (iv) whether the statement was issued by a single country or jointly

(and who they were) and whether it was later supported or opposed (and if so, by whom).

After the statements of each negotiation day were coded, they were aggregated for every

negotiation round, and finally combined to obtain estimates for the whole 2-year period.

Thus, in addition to providing a general overview of the salience of each issue (i.e. which

topic was debated and how often) and how much each country cooperated (i.e. how many

joint statements there were and by whom they were supported), the data set provides a

summary of how these important quantities have evolved over time.

Although the ENB does not always record every single statement made during the ne-

gotiations, most are reported. However, there is a possibility that missing statements might

cause the variables derived from the ENBs to be somewhat distorted, a possibility that

deserves to be noted here.

Submission data

Submissions present the views and positions of negotiation parties in written form. Coun-

tries have the option to submit written statements on various issues to the UNFCCC prior

to the negotiation meetings; these are then compiled into official negotiation documents. All

submissions sent to the two working groups operating during the 2-year period - the Ad

HocWorking Group on Further Commitments for Annex I Parties under the Kyoto Protocol

(AWG-KP) and the AdHocWorking Group on Longterm Cooperative Action under the Con-

vention (AWG-LCA) - were hand-coded (covering a total of 43 official UNFCCC documents,

which summarized these submissions and more than 1600 pages of proposed legal text). A

80

codebook was designed, which was tested on about 50 pages by three different people, to

confirm inter-coder reliability. The main aim of this whole coding process was to generate a

data set with the negotiation positions of all countries on the issues of emissions reduction

targets, the use of market mechanisms, mitigation and adaptation finance, and MRV.

The issues were coded on a scale ranging from 0 to 100, while attempting to emulate

as closely as possible the issues of the interview data set (note that there was a bigger

sample of countries for the submissions). As submissions of individual countries as well as

group submissions were coded, a decision was made regarding how best to combine these

different sources of information. It was decided that individual submissions for a country

regarding a given issue were to be given preference over group submissions. If more than

one individual country submission was coded regarding a given issue, then the average was

taken. If there were no individual submissions for a country regarding a given issue, then

the group submissions of the most important negotiation group of that country were taken

as proxies. This assumption was justified on the grounds that if the group position of the

negotiation group that the country belonged to did not accurately reflect its own views, then

the delegation would have formulated its own submission. When there were multiple group

submissions on a given issue, the average was taken.

81

Appendix 2: The eight policy issues

The eight policy issues for which bargaining success was measured are listed and some of the

main features of the results are given.

• Annex I emissions reduction targets: The average success rating of this issue was 33.8.

Russia had the highest success rating, and Bolivia the lowest.

• Non-Annex I emissions reduction targets and actions: The average success score was

61.9. A number of countries achieved the highest possible value of success (Argentina,

Bolivia, China, Colombia, Egypt, Georgia, Ireland, Mexico, Namibia, Nigeria, Philip-

pines, South Africa, Sri Lanka, and Vietnam). The countries with the lowest success

value were Belize, Comoros, Tajikistan, Togo, and the US.

• Use of market mechanisms: This dimension measured how much a country desired

market mechanisms to play an important role in financing adaptation and mitigation.

The average success score was 75.1. The maximum success rating was achieved by

Bangladesh, Belgium, Belize, Egypt, Ethiopia, Hungary, Japan, the Maldives, Mexico,

the Netherlands, New Zealand, Nigeria, Norway, Panama, Russia, Slovenia, Sweden,

Tajikistan, the UK, and Vietnam. Micronesia and Namibia did not succeed at all in

bargaining over this issue.

• Mitigation finance: The average success value was 65.9. The Philippines achieved the

highest possible score, and the US achieved the lowest score of 0.

• Mitigation allocation: The average success value was 54. The highest scores were

obtained by Russia, Sweden, the US, and the UK. China and Vietnam had the lowest

scores.

• Adaptation finance: The average success rating was 45.7. Only the Netherlands ob-

tained the maximum score, and the US again obtained the lowest success value of

0.

• Adaptation allocation: The average success score was 64.2. No country obtained the

exact winning score, but the US and Vietnam obtained the lowest success value.

• Measurement, reporting, and verification (MRV): The average success score was 22.8.

This suggests that many of the countries scored did not score particularly well on this

issue. Indeed 35 of the 47 countries that answered questions regarding MRV were very

far away from their desired outcome. However, Argentina, Belgium, China, the EU,

Germany, Hungary, and the UK were all successful in their negotiations over this issue.

82

Appendix 3: Overall success of countries including salience

weights

Rank Country Success Rank Country Success1 New Zealand 95 31 Tajikistan 42.02 Hungary 82.8 32 Maldives 41.23 EUa 81.8 33 Bangladesh 39.54 Belgium 79.6 34 Algeria 37.35 Japan 78.7 35 Peru 36.46 United Kingdom 78.2 36 United Arab Emirates 36.27 Russian Federation 77.9 37 Canada 35.48 Norway 76.1 38 India 35.19 Austria 75.5 39 Uganda 34.110 Germany 74.7 40 Mauritania 32.611 Slovenia 73.1 41 Ghana 30.812 Mexico 72.1 42 LDCsb 30.713 Netherlands 71.7 43 Panama 30.614 Ireland 68.2 44 Samoa 30.215 Sweden 65.4 45 Botswana 29.816 Switzerland 64.0 46 Namibia 29.717 Ethiopia 62.7 47 Costa Rica 29.518 Nigeria 62.5 48 Zambia 28.919 Belarus 57.2 49 Tanzania 27.120 Colombia 56.2 50 Kiribati 26.621 Vietnam 56.1 51 Nepal 25.922 Argentina 55.7 52 Sri Lanka 24.623 Egypt 52.0 53 Micronesia 15.824 Papua New Guinea 50.0 54 United States 15.425 Indonesia 49.9 55 Togo 12.326 China 48.9 56 Bolivia 11.727 South Africa 46.8 57 Comoros 028 Mali 46.0 58 Georgia 029 Philippines 43.830 Belize 42.2

a Although the EU participated in the negotiations, and had its own delegation separate from

those of its Member States, it was not included in the the analysis.b The answers for least developed countries (LDCs) as an aggregate group were provided by

an expert working closely with the LDCs as one of the official coalition groups. As in the case

of the EU, the aggregate success scores of the LDCs were not included in the analysis.

83

Cooperation in the Climate ChangeNegotiations: A Network Approach

Florian Weiler

Center for Comparative and International StudiesFederal Institute of Technology, Zurich

Abstract

In international bargaining situations such as the climate change negotiations, statesform ties to support and reinforce each other’s views when they share common nego-tiation positions, and thus generate cooperative networks. So far, traditional networkanalysis has mainly focused on various centrality measures and how the actors in thenetwork are linked to each other. In this paper I employ exponential random graphmodels (ERGMs), a new approach to model networks allowing the researcher to for-mulate hypotheses derived from theory and to test them on the network serving as thedependent variable. At the same time, these models are able to account for dependencystructures in the network, i.e. the (realistic) assumption that tie formation is relatedto the formation of other ties can explicitly be modeled. The network investigatedin this study was constructed from observed cooperative behavior in the UNFCCCnegotiations between December 2007 and December 2009, and hypotheses regardingpower, democratic status, vulnerability to climate change impacts, integration into theinternational community, and culture are proposed and tested on this network.

Keywords: network, ergm, climate change, negotiations, UNFCCC

1 Introduction

In settings were states’ interests diverge, achieving jointly beneficial cooperative outcomes

is difficult in the international system, according to both neorealists and liberals, due to

the lack of a global government able to enforce rules (Keohane, 1984; Waltz, 1979). Ac-

cepting anarchy as the ordering principle of world politics1, one way to realize cooperative

outcomes, according to Oye (1986), is through international negotiations. This article an-

alyzes bargaining behavior of countries in the multilateral negotiations under the auspices

of the United Nations Framework Convention on Climate Change (UNFCCC), which aim

(amongst others) at reducing the global greenhouse gas emissions in order to limit global

warming. According to Olson (1965, p.43-52), some form of coordination which emerges

among the various participants in the climate change negotiations is necessary to achieve

cooperative outcomes.

Looking at the climate change negotiations we notice that states indeed coordinate their

positions on issues on which they share closely related interest, but also that some countries

are more willing to coordinate their positions than others.2 Countries declare their common

negotiation positions by issuing joint statements during official negotiation meetings. In this

paper I analyze the thus formed network in which each of the joint statements serves as a tie

(i.e. a link between two nodes, or in this case countries). Statements were collected over a

two year period prior to the Conference of the Parties (COP) 15 in Copenhagen (December

2009), and the data are part of a newly collected dataset covering various issues of the

climate change negotiations such as positions and bargaining behavior. Various hypotheses

regarding why countries might decide to issue joint statements are proposed and then tested

using a novel network approach employing exponential random graph models (ERGMs). In

what follows the terms joint statements and (network) ties will be used interchangeably.

Why do states form ties with some states, but not with others? I propose that certain

characteristics induce states to coordinate their positions and to make joint statements. For

example, small, relatively powerless states with only relatively limited domestic greenhouse

gas emissions might want to increase the pressure on big, powerful emitters by showing

them that they act in unity. This is clearly the case when Tuvalu bonds up with other small

island states such as Micronesia or Barbados. On the other hand powerful states are very

attractive partners, hence when interests on certain issues overlap smaller countries have

an incentive to issue joint statements with these players to amplify their opinion. For this

1 For a summary and a critique of this view see Wendt (1992) and also Lumsdaine (1993, p.3-29).2 Yet we also observe that countries with similar positions in many cases do not coordinate and issue

joint statements, hence similarity regarding positioning behavior alone is not enough to form ties in thenetwork.

85

reason, countries like the US, China, or India tend to form a greater number of ties than

less important states. Similar interests, or an increased level of prominence, can stem from

various country characteristics such as democratic status, vulnerability to climate change

impacts, integration into the international community, or cultural similarity. In this paper,

hypotheses regard these attributes are proposed and tested on the network formed through

joint statements.

More specifically, making use of ERGMs I explore coordination of bargaining positions

within the UNFCCC context and examine on the one hand which characteristics increase a

single country’s likelihood to form ties, and on the other hand whether shared characteristics

enhance the probability of position coordination among dyads. Conventional approaches to

analyze international relations and cooperation, such as neorealism and liberalism mentioned

above, focus on actors attributes to analyze cooperative behavior, most prominently power.

Traditional network analysis, on the other hand, is concerned with relational data, i.e. ties,

connections, and structures formed among players within a policy network (see e.g. Hafner-

Burton et al., 2009, p.559-560; Jonsson et al. 1998, p.324-326). ERGMs (described in more

detail below) combine these two ways of studying international relations by allowing the

researcher to test hypotheses regarding actors’ characteristics on a policy network, in the

case of this paper the network generated through the issuance of joint statements during the

various negotiation rounds of the climate change negotiations. Thus, the network serves as

the dependent variable of the analysis. This novel approach to study international relations,

although proposed in the literature (see Hafner-Burton et al., 2009, p.568), has to the best

of my knowledge not been used before.3

The aim of the ERGMs presented below is to explain the structure of the network formed

among states during UNFCCC negotiations. The paper does not investigate the chances of

the negotiations to lead to an agreement able to deal with the global climate crisis, but

instead sheds light on the inner workings of the bargaining process, which is crucial to gain

a better understanding of how international regimes are formed. It is therefore important

to distinguish between cooperative outcomes and coordination. The former, according to

Dillenbourg et al. (1995), “is accomplished by the division of labor among participants, as

an activity where each [party] is responsible for a portion of the problem solving”. Thus,

a cooperative outcome in the climate change case is congruent with finding an agreement

which allocates to each party its portion to solve the global climate crisis. Coordination, on

the other hand, is an effort by all or a subset of member countries to act in unity in the

3 Although there exists a working paper by Maliniak and Plouffe (2011) which applies this approach toexplain diplomatic ties between countries, creating the network based on the existence of embassies betweencountries.

86

pursuit of a common goal. The common goal of two or more countries coordinating their

positions in the climate change negotiations is to achieve an agreement as close to their initial

position, and thus as favorable to them, as possible (see Weiler, 2012). Coordination with

others is thus a crucial part of a country’s diplomatic behavior. Yet not all countries with

similar negotiation positions do form coordinated ties, a clear indication that there are also

strategic considerations at play. This coordination activity between more or less likeminded

players is the central topic of this paper.

However, coordination and potential cooperative outcomes are of course closely related.

The basic problem in the climate change case, i.e. why countries have difficulties to tackle

global warming, stems from the fact that the climate system is a global public good (Stern,

2007, p.37-38). Polluters shift costs caused by their own emissions onto others, and at

the same time benefit from abatement efforts undertaken in other countries (Barrett, 2001,

p.1836-1838). From a game theoretical point of view every country has a dominant strat-

egy to continue polluting, and a disincentive to implement meaningful abatement measures

domestically. Thus, climate change constitutes a classical Prisoner’s dilemma (Hopmann,

1996, p.37-52; Ostrom 1990, p.3-5). Olson (1965) argues that coordination is required to

achieve cooperative outcomes and to overcome the Prisoner’s dilemma. Countries with simi-

lar interests should find it easiest and in their advantage to coordinate negotiation positions

in order to achieve common goals. Conversely, as coordination reduces the complexity of

the negotiations and creates reciprocal expectations, the prospects of finding a negotiated

agreement to limit global warming should increase (see Axelrod and Hamilton, 1981; Carraro

and Siniscalco, 1993; Dupont, 1994, 1996; Schelling, 1960, 2002). Studying coordination of

bargaining positions within the context of the UNFCCC negotiation is therefore important

in its own right, but also because understanding the structure of the negotiations may hint

at which challenges parties have to overcome to tackle the underlying Prisoner’s dilemma.

Hence, I strictly distinguished between cooperative outcomes and coordination in this article.

2 Hypotheses

IR scholars have long established that in two player games mutuality of interests plays a

crucial role to achieve cooperative outcomes (see e.g. Axelrod, 1967; Jervis, 1978). Agree-

ment among actors is facilitated if preferences are relatively similar, because in such settings

the bargaining space is comparatively narrow and solutions are not too far away of the pre-

ferred outcome of either party (Hinich and Munger, 1997; Hopmann, 1996). Yet this insight

is difficult to transfer to situations with multiple players and widely diverging interests. In

complex systems, as international negotiations with multiple players, countries agreeing with

87

each other generally still face the problem that numerous other parties oppose them. Zart-

man (1994) describes this problem of complexity of multiparty negotiations in detail and

highlights possible approaches to analyze such multifaceted bargaining situations. One way

to reduce complexity and to facilitate finding an agreement is to form coalitions among par-

ties with similar interests (Dupont, 1994, 1996). In other words, two (or more) parties with

shared interests and attitudes decide to coordinate their positions, in general or on selected

issues, and thus substantiate their views vis-a-vis the remaining negotiating parties. Coordi-

nation among likeminded negotiators is what the Advocacy Coalition Framework (ACF, see

e.g. Sabatier and Jenkins-Smith, 1993) proposes. The ACF postulates coordination through

“biased assimilation”, which assumes that actors with similar characteristics “tend to in-

terpret evidence in a way that supports their prior beliefs and values. According to the

ACF, biased assimilation is the most basic engine that drives collaborative networking and

coalition formation around shared believe systems” (Henry, 2011, p.365). Network analysts

call such a “tendency for nodes [i.e. countries] to form ties based on common attributes

... to share strength and minimize weaknesses” (Hafner-Burton et al., 2009, p.567-568) ho-

mophily. The Resource Dependency Theory (RDT), on the other hand, postulates that

actors perceived as more influential tend to form more ties and are better connected than

less powerful players (Henry, 2011; Weible, 2005). Influence, or perceived influence, accord-

ing to this theory is thus correlated with the numbers of ties formed by an actor. While the

ACF aims at explaining why ties are formed among players with similar characteristic and

interests (homophily effect), the RDT sheds light on the total number of ties or how well an

actor is connected in the network (main effect). The RDT is complementary to the ACF in

explaining the way policy networks are formed.

Power, for example, is an important asset to assert influence over others in international

relations (see e.g. Morgenthau, 1948; Waltz, 1979) as well as in international negotiations

(Bailer, 2004, or more specific to the climate change case Weiler 2012). However, in the

climate change negotiations no single party carries enough weight to determine the outcomes

of the international conferences. To the contrary, due to the principle of consensus, every

small state theoretically has the potential to stall the negotiations if a proposed negotiation

text is conflicting with its interests.4 The big players jointly might still be able to pressure

smaller countries into submission, as they dispose of various ways of retaliation, e.g. restrict

development aid, trade restrictions, etc. Power in the climate change negotiations is derived

from economic performance and the closely related greenhouse gas emissions. Hence, the

term power in this paper is congruent with economic clout. Of course the interests of different

4 Although in Copenhagen and Cancun the protest and resistance of Bolivia and some Small IslandDeveloping States (SIDS) was largely ignored, hence the unanimity principle was somewhat undermined.

88

powerful players might still be rather divers. Under current UNFCCC rules Annex 1 countries

are expected to contribute to the solution of the climate crisis both in terms of CO2 reductions

and financial transfers.5 On the other hand, powerful states of the non-Annex 1 group, such

as China or Brazil, are still exempt from these expectations. But they too come under

increasing pressure to accept binding commitments of their own, as a long-term solution to

limit global warming without their cooperation is infeasible. Thus, possessing power is a

force which unites the interests of countries. Big powers have a joint interest of preventing

too costly climate change agreements, as the financial burden of such an agreement will

mainly have to be carried by them. Following the ACF, a certain level of coordination

among powerful countries to prevent too ambitious, and by implication expensive, outcomes

should therefore be expected. The same is true for smaller, less powerful states, whose only

hope to counter a coalition of big powers is to coordinate as well. Similar levels of power are

thus translated into similar interests, as the ACF suggests.

Moreover, power is an important asset which makes countries attractive to all potential

partners, powerful or not, due to their political clout. Forming alliances with such powerful

players, for example, allows smaller countries to gain access to important information, or even

to “exploit” their partner by free-riding on the big country’s effort (Jones, 2007). According

to the RDT therefore, more powerful actors in networks are expected to form more ties and to

be more active than their less influential peers (Henry, 2011; Weible, 2005). Apart from this

theoretical reasoning, there are other explanations why to expect more powerful countries to

form a greater number of ties. Most prominently, big powers send on average much bigger

delegations to international meetings, which increases the capacity to work on more detailed

issues, attend more meetings, reach out to other parties, etc. As a consequence, finding

overlapping interests with potential partners is facilitated. In addition, delegation members

of more powerful countries are often better trained, have more experience, and as a result

know the structure of the climate change negotiations better than delegation members of

less powerful states (think e.g. of small developing countries). This further adds to their

influence over the negotiations and their attractiveness for other parties, hence the odds of

forming ties increase. Such effects based on nodal attributes of one player only are simply

called main effects by network analysts.

H1a: The likelihood to form ties increases as the power difference between two

countries decreases (homophily effect).

H1b: More powerful parties form more ties during climate change negotiations

(main effect).

5 The Annex 1 of the Kyoto Protocol names 39 developed countries, regarded as mainly responsible forthe CO2 emission which occurred prior to 1990, and allocates greenhouse gas reduction targets to them.

89

Sprinz and Vaahtoranta (1994) show that ecological vulnerability is an important factor

which shapes a country’s position in environmental negotiations such as those on climate

change due to common interests. Consequently, highly vulnerable countries constitute natu-

ral allies during the negotiations. Conversely, countries less vulnerable to climate change also

share common interest, e.g. their main goal might be to limit the costs of a potential treaty.

Relying on the ACF again, shared interests due to climate change vulnerability should lead

countries to form ties. This is what Buys et al. (2009) imply when they calculate vulnera-

bility values for most countries of the world and then conclude that based on these differing

vulnerabilities “countries can have very different orientations towards a global protocol”

(p.303).

Negotiating parties also gain some sort of influence from higher vulnerability levels. Ac-

cording to theory, it might be difficult for other parties for whom the issue is less salient

to ignore highly vulnerable states’ concerns, because their domestic audiences put pressure

on their governments to take concerns of weak and vulnerable countries seriously (Fearon,

1994, 1997). These so called audience costs force players for whom the issue is less salient

to consider more vulnerable party’s apprehensions. In other words, salience can help less

powerful states to be taken seriously by more powerful countries. Therefore, even small and

supposedly weak parties can increase their impact on the negotiations substantially.6 Vul-

nerability to climate change impacts thus may serve as a substitute for (or be additive to)

pure economic power during negotiations (Jonsson, 1981), although some authors are more

skeptical and state that vulnerability is a weakness as countries are more dependent on find-

ing a negotiated agreement and therefore more willing to make concessions (Grundig et al.,

2001). Yet, accepting vulnerability as a source of power, the RDT again suggests a higher

likelihood of more vulnerable countries to form ties in the network. This makes sense from

a purely logical point of view. Countries highly sensitive to changes in the climate system

have a high motivation to get involved in the negotiations, to lobby others by pressing their

case, to seek out potential allies, and in general to be more active, which ultimately causes

them to form more ties.

H2a: The likelihood to form ties increases as the vulnerability levels of two coun-

tries become more similar (homophily effect).

H2b: More vulnerable parties are more active and form more ties during climate

change negotiations (main effect).

Next, I turn the focus to how well integrated countries are in the international community

of states. Bernauer et al. (2009) for example postulate that “more extensive membership in

6 An example of small countries achieving a lot more than expected is the AOSIS negotiation group, inparticular Tuvalu (see also Betzold et al., 2012).

90

international organizations (IOs) motivates states to behave more cooperatively also when

it comes to forms of international cooperation that lie outside the scope of [that] specific

international organization they have joined at some prior time” (p.514) and also find evi-

dence for that claim. The major reason for coordinative behavior due to IO membership is

reciprocity. Countries working together in many international forums know each other well

and may even have established some form of mutual trust. This makes it more likely that

they will cooperate in other fields, such as climate change, as well. In addition, if it is true

that IO membership signals a broader willingness to cooperate to overcome collective action

problems, as claimed by Bernauer et al. (2009), not only should a higher propensity to form

ties among countries with high membership status be expected, but a generally increased

inclination of forming ties also with countries less engaged in the international system.

H3a: The likelihood to form ties increases as the intensity of IO membership of

two countries becomes more similar (homophily effect).7

H3b: Countries with more extensive IO membership form more ties during ne-

gotiations (main effect).

Democracy is another factor in the international domain inducing dyadic coordination,

most prominently expressed in the theory of democratic peace (see e.g. Maoz and Russett,

1993). Research has shown that democracies also tend to work more closely together than

non-democratic countries in other areas, e.g. in the field of international trade (Morrow et al.,

1998), the establishment of international organizations (Russett et al., 1998), or the formation

of alliances (Bennett, 1997; Thompson and Tucker, 1997). One reason to expect increased

coordinative behavior among democracies in negotiation settings is that political leaders of

democratic countries are more accustomed to the process of negotiating compromises than

their peers from less democratic states (see e.g. Dixon, 1994). More specific to the climate

change negotiations, a more “ambitious”8 treaty to tackle climate change and limit global

warming is in the interest of the general public, and particularly the poorer parts of a society

who will disproportionate feel the burden of a warmer climate. Suffering of the poor can more

easily be ignored in non-democratic countries. Thus, relying on the ACF again, countries

with the same democratic status are expected to exhibit an increased likelihood to coordinate

their positions.

7 This of course also means that more isolated countries are also expected to form ties with each other.This might seem counter intuitive, however, a recent example outside the climate change negotiations forexactly such behavior is the agreement signed by Iran and North Korea to cooperate on issues such as scienceand technology, student exchange, etc.

8 The term ambitious is used repeatedly during the negotiations and implies a treaty aiming at limitingglobal warming to a maximum of 2 degree Celsius.

91

For that last reason, the audience costs, I expect democracies not only to coordinate

positions and make joint statements among them, but also to be more active in forming ties

generally, as they have to signal to the home audience that they take domestic preferences

seriously. Neumayer (2002) has shown that democracies tend to exhibit more concern for the

environment than non-democratic countries, since the electorate tends to be better informed

about environmental issues than people in authoritarian states. Furthermore, citizens are

also able to express their views and to put pressure on their government to act in envi-

ronmentally friendly ways. This is also confirmed by Fredriksson and Gaston (2000), who

observe that the presence of civil liberties and democratic freedom increase the probability

of signing and ratifying environmental agreements.

H4a: The likelihood to form ties increases for countries at similar democracy

levels (homophily effect).

H4b: More democratic countries are more active and form more ties during

negotiations (main effect).

Another characteristic affecting the coordination of bargaining positions is culture. Adair

and Brett (2004) suggest that culture shapes three important factors determining negotiation

behavior: believes, goals, and norms. In addition, the authors further assume that there

exists a stark contrast between the individualist sense of self in Western cultures and the

more collectivist self-construal in Eastern cultures. Cultural differences thus lead to diverging

behavior regarding issues such as framing of issues, contextualization of the communication,

or even the way cooperation and conflict are perceived. Regarding the last point, in the

Eastern context trust is the major driver of cooperation, while the possibility of joint gains

induces Western negotiators to form cooperative ties. Gelfand et al. (2001) find empirical

evidence for this difference in perception of conflict and cooperation in Japan and the United

States. With respect to network analysis, (Goodwin and Emirbayer, 1994) state that cultural

factors are necessary for an adequate explanation of network formation. In the context of

the EU (Naurin, 2008) shows that factors such as language proximity, shared religion, or

shared popular cultural affinity influence the intensity with which countries cooperate in the

European Council.

H5: Countries with similar cultural background are more likely to form cooperate

ties (homophily effect).

In addition, coalition groups play a crucial role in the climate change negotiations. Coali-

tion formation is formalized within the UNFCCC context, and official negotiation groups

based on economic and/or regional characteristics appeared over time. Examples are the

92

African Group, the group of Least Developed Countries (LDCs), or the Alliance of Small

Island States (AOSIS). Official meeting are held at the negotiation group level during the

international negotiation rounds to coordinate negotiation positions of the coalition mem-

bers for the general meetings. It is therefore rather straight forward to expect countries

of the same coalition group to coordinate more often with each other than with countries

from other negotiation groups, first because they tend to share similar interests, and second

because regular interactions foster personal relations between negotiators over time, which

facilitates coordination (Rubin and Swap, 1994). As increased coordination among allies is

not particularly surprising or interesting, I abstain from proposing yet another hypothesis

and instead include coalition membership as a control variable in all the models proposed in

the remainder of this paper.

3 Research design

In order to test the proposed hypotheses I employ exponential random graph models (ERGMS).

In such models, the observed network is regarded as a self-organizing structure. It is further

assumed that social processes, which can be modeled, generate the dyadic relations (Robins

et al., 2007, p.175-177). Thus, in ERGMs the dependent variable is the observed network.

When modeling such a network purely on nodal attributes, the assumption underlying

the model is of dyadic independence, i.e. the formation of ties does not depend on other ties

(Goodreau et al., 2009, p.109). However, in social settings such as negotiations, it is rather

unrealistic to assume dyadic independence. Therefore the specification of dyadic dependence

models (also called dependence graphs) is proposed in the literature (see Jensen and Winzen,

2012, and in particular Snijders et al. 2006 for various possible model specifications). Fol-

lowing Goodreau et al. (2009) I propose models including one parameter for triad closure

(edge-wise shared partner parameter models) and another parameter for mean nodal degree9

in order to control for the dependency structures in the network. The former measures the

magnitude of triangular relationships formed in the network, or more specifically whether

two nodes forming ties with the same partner have an increased probability of forming a tie

as well (a friend of a friend is a friend, see also Robins et al., 2007). In practice, triad closure

is investigated making use of the geometrically weighted edge-wise shared partner (GWESP)

statistic provided by the ergm-package, part of the wider network-package written for the R

9 I employ the term ‘edges’ of the ergm-package for R for mean degree, which is measuring the same asthe ‘meandeg’-term of that same package, but is easier to interpret (see Morris et al., 2008). In addition,I also included the geometrically weighted dyadwise shared partner statistic (GWDSP, see Robins et al.,2007), yet this measure was not significant in any model tested and worsened the model fit, hence I droppedit.

93

environment (Butts, 2008; Goodreau et al., 2008; Hunter et al., 2008). The weighting pa-

rameters included in the GWESP statistic is fixed at a set value (again following Goodreau

et al., 2009, p.111-112) to avoid the need for curved exponential family models, additional

parameter estimation, and model degeneracy issues (see Hunter, 2007).10

ERGMs allow the estimation of homophily and main effects, both proposed in the hy-

potheses above, simultaneously. For the benefit of the reader unfamiliar ERGMs, I provide a

reading example how to interpret these effects here. If, for example, the estimated homophily

effect of a categorical variable is 0.3 (as in the democracy case below), these log-odds must

first be transformed into odds, which gives 1.35. This simply means that two countries from

the same category have a 35% higher likelihood to form a tie than countries from different

categories. The interpretation of homophily is slightly more complicated in the continuous

case. Say the estimated effect is -0.15, in this case the odds after the transformation are

0.86. This means that as the difference between two countries with respect to that continuous

variable increases by 1, the chances of forming a tie decrease by 14%. Again, more similar

countries have an increased chance of coordination, hence the term homophily. Therefore,

for continuous variables a negative effect indicates that countries more similar to each other

are more likely to coordinate, while in the categorical case a positive sign implies homophily.

Main effects, on the other hand, are interpreted in the usual fashion. Assuming the co-

efficient of a continuous variable to be 0.5, which de-logged is 1.65, this denotes that a 1

point increase in that variable increases the chance of forming a tie (with a randomly chosen

partner) by 65%. Hence, homophily is a dyadic measure which captures the probability of

coordination among countries based on similarity. Main effects, on the other hand, capture

how variations in characteristics influence the general activity of countries in the network.

3.1 The network of UNFCCC negotiations

During the UNFCCC negotiation meetings countries have the possibility to voice their views

in various segments such as the Ad hoc Working Group on Long-term Cooperative Action

under the Convention (AWG-LCA), the Ad hoc Working Group on Further Commitments

for Annex I Parties under the Kyoto Protocol (AWG-KP), the Subsidiary Body for Scientific

and Technological Advice (SBSTA), or the High Level Segment. These statements have been

registered and published by the Earth Negotiations Bulletin (ENBs, see IISD, 2009) for a

wide variety of environmental meetings and negotiations since 1992. An issue of the ENBs

is published for every day of the UNFCCC negotiations, and includes a summary (usually

about one sentence long) of the majority of statement made by the negotiation parties in

10 To attain a value for the weighting parameter I employed the procedure explained in Goodreau et al.(2009, p.111-112). The value used is 0.10, although an alteration does not change the results dramatically.

94

the publicly accessible meetings.

All the ENBs on the climate change negotiations during the 2-year period from COP 13 in

Bali (December 2007) to COP 15 in Copenhagen (December 2009) were hand-coded. Thus,

in total 11 negotiation rounds and 90 negotiation days are incorporated in the dataset from

which the network is derived. For every statement reported, four properties were recorded:

(i) who made the statement; (ii) which segment of the negotiations it was made in (e.g. COP,

AWG-KP, COP/MOP, AWG-LCA, SBI, SBSTA); (iii) the main topic of the statement (e.g.

mitigation, adaptation, finance, measuring, reporting, and verification), any subcategories,

and (if applicable) the kind of bargaining strategy used; (iv) whether the statement was

issued by a single country or as a joint statement by two or more parties, and whether it was

later supported or opposed (and if so, by whom). After the statements of each negotiation

day were coded, they were aggregated for every negotiation round, and finally combined

to obtain values for the whole 2-year period. The thus generated data not only provide

an overview of cooperation and position coordination among countries, but also provide an

indication of saliency of the different negotiation issues (how often were topics discussed by

a country).

The network serving as the dependent variable in this paper consists of all the joint

statements made over the two year period of the analysis. Hence, if two countries ever issued a

joint statement between (and including) COP 13 and COP 15, they appear in the network as

having formed a tie. The network thus constructed is represented in Figure 1.11 Statements

made by official negotiations groups were excluded from the analysis. Coalition groups are

expected to make joint statements at the outset of meetings, hence these statements merely

represent the smallest common denominator countries within a group could agree upon

and not coordinated positions. Furthermore, group statements do not allow the inclusion of

countries from outside the group, hence they cannot be interpreted as coordination as defined

above and investigated in this paper. The one exception is the European Union (EU), which

acts as a single entity in the negotiations. As the member states do not issue individual

statements during the negotiations, they are excluded from the analysis and instead the EU’s

group statements are recorded as coming from a single participant in the negotiations.12 This

11 Whether a given dyad has coordinated only once or multiple times during the period under study doesnot affect the network for the ERGMs, countries are registered as forming a tie independent of the numberof joint statements. In the network depicted in Figure 1, however, the line width of the edges depends onthe number of joint statement to give the reader a better idea of the collected data. ERGMs for valuednetworks (i.e. networks in which different values for edges representing the intensity of dyadic coordination)are currently under development at the University of Washington, Seattle (see Krivitsky, view). Futureversions of the ergm-package for R will make the implementation of such models possible for practitioners.

12 This means other attributes such as power, or democracy must be aggregated for the EU, which isdone either by averaging (e.g. democratic level) or by summating (e.g. CO2 emissions), whichever is moreappropriate.

95

Figure 1: The network of joint statements during UNFCCC negotiations

DZA

ARG

AUS

BHR

BGD

BRB

BLR

BOL

BRA

CAN

CHL

CHN

COL

HRV

CUB

EGY

EUU

ETH

GAB

GMB

GRD

GUY

ISL

INDIDN

JPN

KAZ

KWT KGZ

MWI

MYS

MHL

MEX

FSM

NZL

NOR

OMN

PAK

PAN

PER

PHL

QAT

KOR

RUSSAU

SGP

ZAF

SDNCHE

THATUR

TUV

UKR

URY

USAVEN

ZMB

African G.AOSISG77/ChinaUmbrellaLDCsOthersEUEIG

The network was constructed using the Fruchterman-Reingold algorithm. Both the size of thenodes and proximity to the center of the graph indicates network centrality. Isolate nodes, i.e.countries not forming any ties, are omitted from the figure

reflects the structure of the UNFCCC negotiations, were the EU is allowed to negotiate as

a block. Finally, negotiating parties not issuing any statements over the study period were

excluded from the analysis.

The thus generated network has 97 nodes, each representing a party to the UNFCCC. Of

the overall possible 4656 possible ties in the network 247 were realized during the two-year

period of the study. Hence, the network has a density of slightly over 0.05, meaning that

about 5% of all possible connections were realized. 38 nodes in the network are isolates, i.e.

they never formed any ties with other participants in the network, although they did issue

statements over the study period. The countries forming most ties were China (36), Saudi

Arabia and Brazil (both 23), India (21), the EU (20), Argentina (19), Canada, Japan, New

Zealand, Mexico, Norway and Russia (all 17), Australia and Singapore (both 16), South

Africa (14), as well as the US and Pakistan (both 10).

What can already be seen from Figure 1 is that countries tend to form clusters around

their negotiation group affiliation, although the coalition group statements have been deleted.

96

Figure 2: Measures of centrality

Betweenness Degree Eigenvalue

0

20

40

60

0.00 0.05 0.10 0.15 0.20 0.00 0.05 0.10 0.15 0.20 0.00 0.05 0.10 0.15 0.20value

coun

t

Furthermore, the center of the graph is crowed by big, powerful countries such as China,

India, the US, the EU, Japan, Brazil, etc., who form on average much more ties than less

powerful players, as well as a multitude of ties amongst them. Thus, H1a and H1b are

somewhat supported purely by inspecting the network. Similar inference can be drawn from

the different centrality measure depicted in Figure 2. Degree is a general measure of how

many ties an actor in the network forms, betweenness computes how often a node lies on a

geodesic, i.e. the shortest way between two not directly connected dyads, and the eigenvalue

reveals how well connected a node is to influential (i.e. well connected) actors in the network

(see e.g. Wassermann and Faust, 1994, p.167-219 for a more detailed description of the three

measures). All three figures show highly skewed distributions and are similar to each other.

This indicates that relatively few actors play a highly central role, reaching out to many less

prominent nodes in the network but also forming ties with each other. This is again evidence

in support of H1a and H1b. In Appendix 1 the three centrality measures used here are listed

for all countries in the network.

3.2 Independent variables

Power: A country’s power is operationalized in two ways. First, by the amount of green-

house gases a country emitted in the last year available, which is 2008 (see UN, 2011, yet

97

I transform the units to 100 millions of metric tons of CO2 due to model convergence is-

sues). Emissions capture power in the climate change negotiations in two ways. On the one

hand they are closely related to a country’s GDP and thus economic power.13 On the other

hand they are a proxy for the influence a country has on the changing climate, adding to

the negotiation party importance in the negotiations on climate change. As a second power

measure I apply delegation size, obtained from the official UNFCCC participants list at the

Copenhagen Climate Change Conference in December 2009 (UNFCCC, 2009). Delegation

size captures a delegation’s potential to specialize, attend meetings, and to form close rela-

tionships with a multitude of delegation members from other countries and thus represents

a softer form of power than greenhouse gas emissions.

Vulnerability: A country’s vulnerability to climate change impacts is measured using the

Environmental Vulnerability Index (EVI) developed by the South-Pacific Applied Geoscience

Commission (SOPAC) and the United Nations Environmental Programme (UNEP). In total,

the EVI measures 50 indices, 13 of which are used to construct a sub-index for climate change

vulnerability (Kaly et al., 2004).14 Although the EVI has been criticized for various reasons

(Barnett et al., 2008), - for example, on the grounds that it is impossible to quantify complex

social-ecological processes - this criticism is not particular to it and applies to all indices that

measure vulnerability.

IO membership: Membership in international organizations is operationalized by the

number of IOs of which a country is a full member as indicated by the Correlates of War

Dataset for the year 2000 (Pevehouse et al., 2004).

Democracy: For democracy scores I employ the most basic measure reported by Freedom

House (2012), rating countries either as free, partly free, or not free. In the network of 97

countries used for the models described below, 33 are rated as free, 30 are partly free, and

the remaining 34 are assessed as not free.

Culture: Whether dyads are culturally related is captured in two ways. First, official

languages are utilized to measure on the one hand colonial history, but also the ease of

communication. The following four languages are coded as dummy variables, taking on

13 When using GDP instead of emissions to capture power the results are very similar although slightlyless significant (at the 1% instead of the 0.1% level)

14 These are indices capturing climatic changes such as precipitation, droughts, etc. CO2 emissions assuch are not included, only consequences of increased global temperatures. Thus, collinearity between thevulnerability index and the greenhouse gas emissions capturing power is not problematic.

98

the value 1 if they are an official language and 0 otherwise: English, French, Spanish, and

Portuguese. A host of countries have more than only one entry, as two (or even more) of the

listed languages are officialese. On the other hand, 27 of the countries in the dataset did not

adopt any of listed languages.15 As a second measure of cultural similarity a second network

is introduced in which the value 1 indicates that two countries have a shared border.16 This

network was generated using the cshapes-package for R (see Weidmann and Gleditsch, 2010)

and then ordered to match the network of the UNFCCC negotiations.

Negotiation group: To control for coalition groups an additional variable capturing group

membership is added. Although some non-Annex 1 countries are members of more than one

negotiation group, for the purpose of this paper they are coded as belonging to only a single

coalition deemed most important for them. Almost all non-Annex 1 countries are members

of the G77/China, however, when they are also part of another group, they are coded as

members of that coalition. The same procedure is repeated for Africa, where all countries

are part of the African Group, but some in addition are members of the LDCs and are always

coded as such. All AOSIS countries are furthermore coded to belong to that group. Table 1

provides summary statistics for the independent variables.

Table 1: Descriptive statistics of the independent variables

Variable name Obs. Mean s.d. Min. Max.Emissions 97 2.95 10.00 0 70.32Delegation size 97 68.7 80.64 6 450Vulnerability 97 3.28 0.71 1.85 4.90IO membership 97 62.51 19.97 11 126Democracy 97 2.01 0.84 1 3Culture (network) 4656 0.02 0.15 0 1English 97 0.43 0.50 0 1French 97 0.16 0.37 0 1Spanish 97 0.15 0.36 0 1Portugese 97 0.02 0.14 0 1

15 As all the languages are coded as dummy variables, these 27 countries are not considered by the modelsas sharing a single language (as would be the case if languages would be coded in a single variable capturinglanguage groups, with those 27 countries coded as 0).

16 Networks can be modeled as independent variables using the dyadcov-term of the ergm-package. Theinterpretation, however, is just the same as in the homophily case, as not the attribute of single nodes isevaluated, but a characteristic that is shared between two nodes (here a tie in form of a boarder).

99

4 Results and discussion

The results of the models described in this section are summarized below in Table 2. When

running a model including only the parameters capturing the dependency structure in the

network (triad closure and mean nodal degree, model not in Table 2), the model converges

quickly and both parameters are highly significant. However, both measures for goodness

of fit in this simple model, the AIC and the BIC (1621.4 and 1634.3 respectively), are

higher than in all models including additional parameters. This indicates that all presented

models include substantive effects and are an improvement over the baseline model. In what

follows, the various hypotheses will be further evaluated against the evidence provided by

the models. Table 3 provides a summary for the odds and the percentage change in the

odds for all significant effects of Model 5 in Table 2.17 These results are repeatedly used in

the following discussion. Models 1 to 3 of Table 2 are partial models, with Model 1 being a

power model, Model 2 Model 3 capturing negotiation assets of countries (vulnerability and

connectedness via IO membership), and Model 3 acting as a cultural model. Models 4 and

5 are full models, yet the former includes all operationalized terms, while the latter only

includes the more substantial parameters for power and culture (both being operationalized

twice).18 In what follows I discuss these models with a particular focus on Model 5, as this

is the model with the best goodness of fit according to the BIC.

Power: I start with the discussion of power and turn first to the homophily effect (H1a)

which states that countries exhibit increased odds of forming ties at similar levels of power.

Greenhouse gas emissions, the proxy used for power, show a highly significant negative sign

for the homophily effect across all models, which implies that as the difference of emissions

between two countries increases, the likelihood of forming ties shrinks. More specifically,

when the gap between two countries’ emissions widens by 100 million metric tons, the prob-

ability of forming a tie decreases by about 13%. In other words, countries very similar with

regards to their emissions, like for example Kenya (10.4 million tons of CO2 emissions) and

Sri Lanka (11.8 million tons), are expected to have a higher probability to form ties amongst

them than with countries emitting considerably more (or less), such as Algeria (111.3 million

tons). In the chosen example the difference between Algeria and the other two countries is

17 The coefficients of the models are log-odds and must therefore be transformed into odds before theycan be interpreted further.

18 Dyadic dependence models such as those proposed in this paper are estimated using Markov chainsMonte Carlo (MCMC) methods. Hence, run length of the Markov chain is an important issue. For the partialmodels reveal chains with a length of 300,000 iterations are sufficient to achieve convergence. However, modeldiagnostics reveal that a run of about 3 million iterations is required to fully explore Models 4 and 5, whichis far beyond the limits of the computing power of the author’s machine. Therefore, I made use of ETH’sBrutus super-computing cluster to estimate the two full models.

100

Table 2: Results of various ERGMs

Model 1 Model 2 Model 3 Model 4 Model 5Emissions (main) 0.17∗∗∗ 0.13∗∗∗ 0.15∗∗∗

(0.03) (0.03) (0.02)Emissions (homophily) −0.15∗∗∗ −0.11∗∗∗ −0.14∗∗∗

(0.03) (0.03) (0.02)Delegation size (main) 0.17∗ 0.12

(0.08) (0.09)Delegation size (homophily) −0.15† −0.24∗

(0.09) (0.09)Vulnerability (main) 0.39∗∗∗ 0.30∗∗ 0.22∗

(0.10) (0.10) (0.09)Vulnerability (homophily) 0.00 0.13 0.13

(0.11) (0.12) (0.12)IO membership (main) 1.72∗∗∗ 1.90∗∗∗ 1.03∗∗∗

(0.29) (0.51) (0.27)IO membership (homophily) −0.87∗∗ −0.74 −0.74∗∗

(0.32) (0.52) (0.29)Democracy (main, pf) −0.38∗∗∗ −0.42∗∗∗ −0.44∗∗∗

(0.10) (0.12) (0.12)Democracy (main, nf) −0.15∗ −0.09 −0.19†

(0.07) (0.10) (0.10)Democracy (homophily) 0.24† 0.34∗ 0.30∗

(0.13) (0.15) (0.14)Culture (boarder network) 1.20∗∗∗ 0.95∗∗ 0.96∗∗

(0.27) (0.31) (0.31)English (homophily) −0.04 0.01

(0.13) (0.15)French (homophily) 0.22† 0.29

(0.13) (0.19)Spanish (homophily) −0.06 0.06

(0.12) (0.17)Portugese (homophily) −0.62∗∗∗ −0.71∗

(0.14) (0.29)Negotiation group (homophily) 0.98∗∗∗ 1.01∗∗∗ 0.81∗∗∗ 1.07∗∗∗ 1.02∗∗∗

(0.16) (0.13) (0.13) (0.15) (0.14)Nodal degree −6.00∗∗∗ −10.32∗∗∗ −5.06∗∗∗ −9.83∗∗∗ −8.29∗∗∗

(0.77) (1.09) (0.37) (1.49) (1.02)Triad closure (GWESP) 2.51∗∗ 2.51∗∗∗ 2.65∗∗∗ 2.30∗∗∗ 2.39∗∗∗

(0.76) (0.31) (0.29) (0.30) (0.30)AIC 1406.94 1471.86 1521.24 1350.19 1361.31BIC 1452.06 1516.98 1592.14 1472.66 1445.11MCMC-Length 3e+05 3e+05 3e+05 3e+06 3e+06N 4656 4656 4656 4656 4656∗∗∗p < 0.001, ∗∗p < 0.01, ∗p < 0.05, †p < 0.10

101

almost exactly 100 million tons. Hence, the chances to form ties are expected to be about

13% higher between Kenya and Sri Lanka than for either of them with Algeria. The story for

delegation size is similar, although the significance levels drops somewhat. The coefficient of

delegation size in Model 4 indicates that an increase in the size difference of two countries’

delegations by 10 decreases the propensity to form ties by slightly more than 2%. These

findings are affirmative of H1a.

The second power hypothesis (H1b - main effect) is also supported by the highly signif-

icant coefficient of CO2 emissions across the models. According to Model 5 of Table 2, an

increase of emissions by 100 million tons is congruent with a 17% increase in forming ties.

To stick to the example from above, although Algeria has a lower probability of forming a tie

with Kenya than Sri Lanka (due to homophily), the chances that Algeria coordinates with a

randomly chosen partner are about 17% higher than for either of the two others. The main

effect of delegation size tells a similar story, although the coefficient falls below significance

in the full model. An increase of the delegation size by 10 members, according to Model 1 of

Table 2 where the effect shows significance, translates into an expected increased probability

of forming ties of about 2%. As some delegations can be quite large, even this seemingly

small effect can be substantial. For example, the delegation of Japan in Copenhagen had

176 members, while India only sent 77 delegates to the conference. According to the model,

Japan therefore has an about 20% higher likelihood to form ties than India. Combined these

results suggest the validity of H1b.

Table 3: Substantive effects of robust and significant parameters

Variable Odds Percentage changein the odds

Emissions (main) 1.17 16.76Emissions (homophily) 0.87 -13.45Vulnerability (main) 1.25 24.62IO membership (main) 2.81 181.38IO membership (homophily) 0.48 -52.15Democracy (main, pf) 0.65 -35.42Democracy (main, nf) 0.83 -17.07Democracy (homophily) 1.36 35.52Boarders (network) 2.62 161.52Negotiation group (homophily) 2.78 178.17

Effects of parameters reaching significance at least at the 10% level in Model 5. Column 2

shows the odds associated with a unit change in the independent variable, and in column 3

states the percentage change in the odds is given.

102

Vulnerability: The findings for vulnerability, both the main and the homophily effect, are

probably the weakest of this paper. Yet while there is still strong evidence in favor of the

main effect (H2b), the homophily effect (H2a) is insignificant across the models and must

be rejected. Vulnerability as main effect does exhibit significance at least at the 5% level

in all models. The positive sign of the effect indicates that as vulnerability scores increase,

countries tend to grow more active and to become more involved in the negotiation process.

Specifically, a one point increase in the vulnerability scores, according to Model 5, induces

countries to be on average about 25% more active in coordinating their positions with others.

Highly vulnerable small countries, such as the SIDS, need a strategy to ensure that their views

are reflected in the negotiations. The rejection of the homophily effect of vulnerability is an

indication that such countries try to reach out to and coordinate positions with more powerful

countries to make their voice heard. Tuvalu is a good example for this sort of bargaining

behavior. Overall, the country issued joint statements with ten other negotiating parties,

among them Argentina, Australia, China, the EU, New Zealand, Norway, and Switzerland.

Micronesia’s strategy is similar, the country formed ties with inter alia Brazil, Egypt, the

EU, South Africa, and the Philippines. Thus, Jones’ (2007) notion that small, seemingly

powerless countries “exploit” stronger negotiation parties by benefitting from their strength

is somewhat supported.

IO membership: The homophily effect of IO membership (H3a) is again supported by

the models, with the exception of Model 4 of Table 2. In Model 5, the effect is solidly

significant with a p-value below 0.01. The coefficient of -0.74 indicates that as the difference

between two countries regarding the number of IOs they are a member of increases by 10,

they are about 5% less likely to form ties in the climate change negotiations, presumably

because they know each other less well from other international forums. The minimum

number for IO membership in the data is 11 (Palau), while the maximum is 126 (EU19).

Hence, the probability of these two actors to form a tie is about 60% lower than for two

countries who are members in about the same number of IOs. H3a is thus support. Indeed

very well integrated countries into the international community tend to form ties with other

well connected states, e.g. the EU, Australia, the US, China, and Japan exhibit a very high

propensity of forming ties among each other. Conversely, countries who are members of only

relatively few international organizations tend to be much less active, yet this is captured

(and thus controlled for) by the main effect of IO membership. When such relatively isolated

countries do form ties, however, they have an increased propensity to connect to other states

who also suffer an integration deficit. For example, Micronesia forms inter alia connections

19 For the EU every IO is counted in which at least two member countries enjoy full membership status.

103

to countries such as Ethiopia, Bangladesh, Palau, Singapore, and Tuvalu, all of which are in

the lower third of countries in terms of IO membership. Another example is Bahrain, who

formed ties with Kuwait, Saudi Arabia, Singapore, and the Gambia, again countries not very

well connected in the international system, but at a similar level as Kuwait is. Hence, this

homophily effect seems to be equally valid for well integrated and more isolated countries.

The evidence in favor of H3b (main effect of IO membership) is even stronger, as the

coefficient in question consistently is highly significant. This means that countries who are

members in a higher number of IOs are in general more active and form more ties than

countries less well integrated into the international system. More specifically, if full mem-

bership status in IOs is increased by 10, the propensity of forming ties in the climate change

negotiations rises by about 18%. This considerably large effect is a rather strong indication

for the validity of H3b. This is not particularly surprising. Among the best connected nego-

tiating parties are the EU (126 IOs), the US (92), Brazil (87), Canada (85), India (84), or

Japan (82), most of which are very active in the negotiations and form a substantial number

of ties. However, as the power effect of these countries is already accounted for, their still

remains a significant and not too small effect signaling that beyond sheer size and power

there is something else which induces these countries to be more active than others during

negotiations.

Democracy: The results for a country’s democratic status indicate increased coordination

at similar levels of democracy, as H4a suggests. Although only significant at the 10% level

in the partial model, the significance increases in both full models and is thus corroborated.

The homophily effect of 0.30 in Model 5 of Table 2 translates to odds of 1.36, which implies

that countries sharing the same Freedom House rating have an higher likelihood to issue joint

statements than countries with different ratings. The coefficient indicates that two countries

from the same category are about 36% more likely to from a tie in the network than two

countries from different categories. This statement is valid for all three possible categories

free, partly free, and not free and supports H4a.

Next I discuss the main effect of democracy (H4b), which is the propensity of countries

to form ties based on their democratic status alone. As there are only three possible rating,

separate parameters are estimated in the models for each category, with free serving as the

baseline group. The effect of partly free is significant and negative in all models, hence

partly free countries tend to be less active than their more democratic counterparts. The

same is true for countries with the rating not free, yet the effect is smaller, on the fringe

of significance and in some Model 4 even insignificant. The odds of partly free countries to

form a tie with a randomly chosen country are about 35% lower than for free countries. For

104

not free countries these odds are only 17% below those of free countries (employing Model

5 coefficients). This result is unexpected. According to theory, states labeled as not free

should be expected to be less active than partly countries. As it turns out, some particularly

active states such as China, Saudi Arabia, or Russia are in the not free segment, producing

effects which reveal that the not free countries are more active than their partly free peers.

Power is already controlled for in the models, hence it cannot be the hidden driver behind

these unexpected results. Hence, H4b is partially contradicted, although evidence is found

that free countries are more active than countries from either of the two other categories.

Culture: Next I focus on the effects of culture and H5. Regarding shared official languages

the results are rather disappointing. The only significant effect is found for Portuguese,

however, the sign of the coefficient is negative and countries sharing this language exhibit

a decreased probability of forming ties vis--vis a randomly chosen country pair. I suspect

the reason for this unexpected result to be the rather big geographical distance between

the Portuguese speaking countries. These results also indicate that common languages,

introduced by former colonial powers, are not very good proxies for culture.

Shared boarders, on the other hand, which imply at least some cultural exchange and

a shared history, exhibit much stronger and significant effects in favor of H5. Utilizing

again the coefficient of Model 5 in Table 2, neighboring countries are 161% more likely to

form ties during the climate change negotiations than countries without a shared boarder.

This seems like a rather sizeable effect, yet recall that the baseline propensity of countries to

coordinate positions is only about 5%. Bordering countries thus exhibit a probability to form

ties of around 13%, much higher than the baseline probability, but still within the limits of

expectations. Culture based on common boarders is a concept which can easily be criticized.

For example, it has been suggested that instead of culture, common boarders measure to some

extend two countries’ similarity in terms of vulnerability to climate change impacts. In my

view, this is rather unrealistic, as on the one hand I control for vulnerability in the models, on

the other hand bordering countries can exhibit vastly different vulnerability scores. However,

common borders might still measure something I do not want to capture instead of culture,

yet lacking a better measure for culture I have to rely on that measure and find evidence

in support of H5. To control for the potential influence the inclusion of common borders

has on the other coefficients I computed Model 5 again but excluded the proxy for culture.

The resulting estimates were very similar to those reported in Table 2. Hence, although the

findings for culture are questionable due to undeniable operationalization problems, other

findings in the paper are not polluted by the inclusion of common borders.

Finally, note that the effect of the control variable negotiation group is highly significant

105

across all models. The direction and size of the effect confirms that countries of a coalition

group tend to work much closer together than two randomly chosen countries, even after

discounting the purely group specific statements. Overall, coalition partners have an almost

three times higher probability of issuing joint statements than two countries from different

groups. Although not very surprising, this result shows that coalitions indeed play a pivotal

role in the climate change negotiations.

5 Conclusion

The results of the models described and discussed in this paper clearly indicate that coordi-

nated behavior in the form of issuing joint statements during the climate change negotiations

strongly depends on countries’ interests and characteristics. ERGMs provide an ideal oppor-

tunity to empirically test hypotheses derived from the IR literature on a network including

a big number of the worlds’ countries. This novel approach used in this paper, modeling

country characteristics on a network of ties in the climate change negotiations, helps to

shed more light on the question why countries coordinate their negotiation positions and

how cooperation in a multiparty environment works. In particular, the paper shows that

a country’s power, its democratic status, the integration into the international system of

states, culture (although only crudely measured by shared borders), and to a lesser extend

also vulnerability to climate change impacts all play a role in determining the coordinative

behavior of parties to the UNFCCC. The statistical models discussed above show that both

homophily as well as main effects are important in explaining the structure of the network

formed in the climate change negotiations through joint statements.

In the introduction of this paper I argued that explaining the structure of the network

formed during the UNFCCC negotiations is important to gain a better understanding of

regime formation. Coordination of negotiation positions reduces the complexity of the nego-

tiations and thus facilitates overcoming the Prisoner’s Dilemma. This paper demonstrates

that coordination is indeed occurring. Particularly countries knowing each other through

other international forums were they presumably formed some level of mutual trust, as well

as countries at similar democratic levels tend to form collaborative ties. In addition, unsur-

prisingly, a considerable share of positioning coordination is taking place at the negotiation

coalition level. On the other hand, countries particularly vulnerable to climate change im-

pacts do not primarily seek partnership with other climate sensitive countries, but instead

try to voice their concerns by teaming up with often powerful and thus influential negoti-

ating parties. Thus, the overall network shows clear patterns of coordination. On the one

hand countries seek to maximize their gains by influencing the outcome of the negotiations

106

in their favor. On the other hand these structures shed light on the inner process of regime

formation, as clear battle lines between democratic and non-democratic states, developed

and developing countries (the coalition groups), or more and less well integrated countries

can be distinguished.

As examples for these homophily effects consider Australia and the EU. The former has

14 ties in the network, eleven of which are joint statements with other democratic parties.

For the EU this ratio is 13 out of 17. In addition, both the EU and Australia have made joint

statements with China and Russia, which exemplifies the increased likelihood of powerful

states to form connections. Examples for not free states working closely together is the

cluster Russia, Kazakhstan, Belarus, Kyrgyzstan, Thailand, Gambia, Bahrain, Ethiopia,

and Saudi Arabia. These countries form regular connections among them, yet have only

infrequent contacts to more democratic countries. If they form connections with countries

not labeled not free, they are almost always from the partly free group. The exceptions

are Saudi Arabia and Russia, who also form regular connections with democratic, mostly

powerful countries. This again confirms the power hypothesis.

Regarding the main effects is has already been pointed out that highly vulnerable coun-

tries such as Micronesia and Tuvalu are very active. In addition, there are a host of other

rather active, highly vulnerable countries in the network, e.g. Japan (17 ties), Singapore

(16), Kuwait (9), South Korea (8), Iceland (6), or Bahrain (4). The same is evidently true

for power, as can be seen in the graphic depiction of the network, where parties such as

China, India, Brazil, the EU, Japan, or the US take on highly central, well connected po-

sitions. The main effect for democracy might not be as obvious. Yet there are at least as

many free countries in the center of the graph (which indicates high connectivity) as partly

free and not free nodes combined. This is a clear indication that democracies tend to form

ties more frequently, while the distinction between not free and partly free countries is much

less clear.

What do these results indicate for the future of the negotiations? First, I would suggest

that if there really is a fourth wave of democratization ongoing in the Arab World, which

is still debated (see e.g. Diamond, 2011), this might also have positive implications for the

climate change negotiations. If newly formed democracies really behave more cooperatively,

as the results in this paper suggest, then a more democratic world would significantly improve

the chances of finding an agreement. Another interesting finding is that powerful states and

those well integrated into the international system do form considerably more ties. As I

had the chance to visit negotiation meetings repeatedly, this corresponds to my impression

that resistance against possible agreements often comes from smaller states with particular

interests (e.g. Bolivia, Venezuela, but also Tuvalu, or the oil exporting countries). These

107

results might imply that a good way towards progress in the negotiations could be increased

discussions in other international fora such as the G20. Potential agreements found in such

a smaller setting could then be transferred to the bigger UNFCCC negotiations. Finally,

the findings that countries with cultural similarities more often coordinate their positions

indicates that if a global agreement fails to materialize, smaller regional negotiations and

treaties might be a way to go ahead. Although such an approach is of course only the

second best option, regional agreements could be an option for those countries willing to

combat global warming and to show their commitment via a treaty. An example for such an

approach is the EU, which has pledged to reduce emissions further in a second Kyoto period

even if the negotiations should fail (see e.g. Bohringer et al., 2009).

To conclude, networks in international negotiations do not form randomly. This paper

is a first attempt to model the structure of such a network for the climate change case. A

better understanding of what drives countries to coordinate their positions might help to

find a solution to the Prisoner’s Dilemma underlying the climate change negotiations, and

is therefore an interesting topic to study. More research in this direction is thus desirable.

108

References

Adair, W. L. and J. M. Brett (2004). Culture and negotiation processes. In M. J. Gelfandand J. M. Brett (Eds.), The Handbook of Negotiation and Culture, pp. 158–176. Stanford,CA: Stanford Business Books.

Axelrod, R. (1967). Conflict of interest: An axiomatic approach. Journal of Conflict Reso-lution 11 (1), 87–99.

Axelrod, R. and W. D. Hamilton (1981). The evolution of cooperation. Science 211 (4489),1390–1396.

Bailer, S. (2004). Bargaining success in the European Union: The impact of exogenous andendogenous power resources. European Union Politics 5 (1), 99–123.

Barnett, J., S. Lambert, and I. Fry (2008). The hazards of indicators: Insights from the Envi-ronmental Vulnerability Index. Annals of the Association of American Geographers 98 (1),102–119.

Barrett, S. (2001). International cooperation for sale. European Economic Review 45 (2001),1835–1850.

Bennett, D. S. (1997). Testing alternative models of alliance duration, 1816-1984. AmericanJournal of Political Science 41 (3), 846–878.

Bernauer, T., A. Kalbhenn, V. Koubi, and G. Spilker (2009). A comparison of interna-tional and domestic sources of global governance dynamics. British Journal of PoliticalScience 40 (3), 487–508.

Betzold, C., P. Castro, and F. Weiler (2012). AOSIS in the UNFCCC negotiations: Fromunity to fragmentation? Climate Policy 12 (5), 591–613.

Bohringer, C., A. Loschel, U. Moslener, and T. Rutherford (2009). EU climate policy up to2020: An economic impact assessment. Energy Economics 31 (Supplement 2), S295–S305.

Butts, C. T. (2008). network: A package for managing relational data in R. Journal ofstatistical software 24 (2), 1–36.

Buys, P., U. Deichmann, C. Meisner, T. T. Thao, and D. Wheeler (2009). Country stakesin climate change negotiations: Two dimensions of vulnerability. Climate Policy 9 (3),288–305.

Carraro, C. and D. Siniscalco (1993). Strategies for the international protection of theenvironment. Journal of Public Economics 52, 309–328.

Diamond, L. (2011). A Fourth Wave or False Start? Democracy After the Arab Spring.Foreign Affairs, May 22.

109

Dillenbourg, P., M. J. Baker, A. Blaye, and C. O’Malley (1995). The evolution of researchon collaborative learning. In P. Reimann and H. Spada (Eds.), Learning in humans andMachines. Towards an Interdisciplinary Learning Science, Volume 189-211. London: Perg-amon.

Dixon, W. J. (1994). Democracy and the peaceful settlement of international conflict. Amer-ican Political Science Review 88 (1), 14–32.

Dupont, C. (1994). Coalition theory: Using power to build cooperation. In I. W. Zartman(Ed.), International Multilateral Negotiations. Approaches to the Management of Com-plexity, pp. 148–177. San Francisco: Jossey-Bass Publishers.

Dupont, C. (1996). Negotiation as coalition building. International Negotiation 1 (1), 47–64.

Fearon, J. D. (1994). Domstic political audiences and the escalation of international disputes.American Political Science Review 90 (7), 15–35.

Fearon, J. D. (1997). Signaling foreign policy interests: Tying hands versus sinking costs.The Journal of Conflict Resolution 41 (1), 68–90.

Fredriksson, P. G. and N. Gaston (2000). Ratification of the 1992 climate change convention:What determines legislative delay? Public Choice 104 (3), 345–368.

Freedom House (2012). Country ratings and status, fiw 1973-2009, available athttp://www.freedomhouse.org/report-types/freedom-world.

Gelfand, M. J., L. H. Nishii, K. M. Holcombe, N. Dyer, K.-I. Ohbuchi, and M. Fukuno(2001). Cultural influences on cognitive representations of conflict: Interpretations ofconflict episodes in the united states and japan. Journal of Applied Psychology 86 (6),1059–1074.

Goodreau, S. M., M. S. Handcock, D. R. Hunter, C. T. Butts, and M. Morris (2008). Astatnet tutorial. Journal of statistical software 24 (9), 1–27.

Goodreau, S. M., J. A. Kitts, and M. Morris (2009). Birds of a feather, or friend of afriend? using exponential random graph models to investigate adolescent social networks*.Demography 46 (1), 103–125.

Goodwin, J. and M. Emirbayer (1994). Network analysis, culture, and the problem of agency.The American Journal of Sociology 99 (6), 1441–1454.

Grundig, F., H. Ward, and E. P. Zorick (2001). Modelling Global Climate Negotiations,Volume 153-182 of International Relations and Global Climate Change. Cambridge, MA:MIT Press.

Hafner-Burton, E. M., M. Kahler, and A. H. Montgomery (2009). Network analysis forinternational relations. International Organization 63 (3), 559–92.

Henry, A. D. (2011). Ideology, power, and the structure of policy networks. Policy StudiesJournal 39 (3), 361–383.

110

Hinich, M. and M. Munger (1997). Analytical Politics. Cambridge, New York, Melbourne:Cambridge University Press.

Hopmann, P. T. (1996). The Negotiation Process and the Resolution of International Con-flicts. Columbia: University of South Carolina Press.

Hunter, D. R., M. S. Handcock, C. T. Butts, S. M. Goodreau, and M. Morris (2008). ergm:A package to fit, simulate and diagnose exponential-family models for networks. Journalof statistical software 24 (3), 1–29.

IISD (2007-2009). Earth negotiation bulletin. http://www.iisd.ca/vol12.

Jensen, T. and T. Winzen (2012). Legislative negotiations in the european parliament.European Union Politics 13 (1), 118–149.

Jervis, R. (1978). Cooperation under the security dilemma. World politics 30 (2), 167–214.

Jones, P. (2007). Colluding victims: A public choice analysis of international alliances. PublicChoice 132 (3), 319–332.

Jonsson, C. (1981). Bargaining power: Notes on an elusive concept. Cooperation and Con-flict 16 (4), 249–57.

Jonsson, C., B. Bjurulf, O. Elgstrom, A. Sannerstedt, and M. Stromvik (1998). Negotiationsin networks in the european union. International Negotiation 3 (3), 319–344.

Kaly, U., C. Pratt, and J. Mitchell (2004). The demonstration environmental vulnerabilityindex (EVI) 2004. Technical report, SOPAC Technical Report 384.

Keohane, R. O. (1984). After Hegemony: Cooperation and Discord in the World PoliticalEconomy. Princeton: Princeton Universtiy Press.

Krivitsky, P. N. (under review). Exponential-family random graph models for valued net-works. Arxiv preprint (arXiv:1101.1359), available at http://arxiv.org/pdf/1101.1359.pdf.

Lumsdaine, D. H. (1993). Moral vision in international politics: the foreign aid regime,1949-1989. Princeton: Princeton University Press.

Maliniak, D. and M. Plouffe (2011). A network approach to the formation of diplomatic ties.

Maoz, Z. and B. Russett (1993). Normative and structural causes of democratic peace,1946-1986. American Political Science Review , 624–638.

Morgenthau, H. (1948). Politics Among Nations. The Struggle for Power and Peace. NewYork: Alfred A. Knopf.

Morris, M., M. S. Handcock, and D. R. Hunter (2008). Specification of exponential-familyrandom graph models: Terms and computational aspects. Journal of statistical soft-ware 24 (4), 1–24.

111

Morrow, J. D., R. M. Siverson, and T. E. Tabares (1998). The political determinants ofinternational trade: The major powers, 1907-90. American Political Science Review , 649–661.

Naurin, D. (2008). Choosing partners. coalition-building in the Council of the EU. Paperpresented at the American Political Science Association Annual Meeting, Boston August28-31, and at the ECPRs Fourth Pan-European Conference on EU Politics, Riga, Septem-ber 25-27, 2008.

Neumayer, E. (2002). Do democracies exhibit stronger international environmental commit-ment? A cross-country analysis. Journal of Peace Research 39 (2), 139–164.

Olson, M. (1965). The Logic of Collective Action. Public Goods and the Theory of Groups.Cambridge, London: Harvard University Press.

Ostrom, E. (1990). Governing the commons. The evoluton of institutions for collective action.Cambridge: Cambridge University Press.

Oye, K. A. (1986). Explaining cooperation under anarchy: hypotheses and strategies. WorldPolitics 38 (1), 1–24.

Pevehouse, J. C., T. Nordstrom, and K. Warnke (2004). The correlates of war 2 inter-national governmental organizations data version 2.0. Conflict Management and PeaceScience 21 (2), 101–119. Data available at http://www.correlatesofwar.org/COW2

Robins, G., P. Pattison, Y. Kalish, and D. Lusher (2007). An introduction to exponentialrandom graph (p*) models for social networks. Social networks 29 (2), 173–191.

Robins, G., T. A. Snijders, P. Wang, M. S. Handcock, and P. E. Pattison (2007). Re-cent developments in exponential random graph (p*) models for social networks. Socialnetworks 29 (2), 192–215.

Rubin, J. Z. and W. C. Swap (1994). Small group theory: Forming consensus through groupprocesses. In I. W. Zartman (Ed.), International Multilateral Negotiation: Approaches tothe Management of Complexity, pp. 132–147. San Francisco: Jossey-Bass.

Russett, B., J. R. Oneal, and D. R. Davis (1998). The third leg of the kantian tripodfor peace: International organizations and militarized disputes, 195085. InternationalOrganization 52 (3), 441–467.

Sabatier, P. A. and H. C. Jenkins-Smith (1993). Policy change and learning: An advocacycoalition approach. Boulder, CO: Westview Press.

Schelling, T. C. (1960). The Stategy of Conflict. Cambridge, London: Harvard UniversityPress.

Schelling, T. C. (2002). What makes greenhouse sense? time to rethink the kyoto protocol.Foreign Affairs 81 (3), 2–9.

112

Snijders, T. A., P. E. Pattison, G. L. Robins, and M. S. Handcock (2006). New specificationsfor exponential random graph models. Sociological Methodology 36 (1), 99–153.

Sprinz, D. and T. Vaahtoranta (1994). The interest-based explanation of international envi-ronmental policy. International Organization 48 (1), 77–105.

Stern, N. (2007). The Economics of Climate Change: The Stern Review. Cambridge: Cam-bridge University Press.

Thompson, W. R. and R. M. Tucker (1997). A tale of two democratic peace critiques.Journal of Conflict Resolution 41 (3), 428–454.

UNFCCC (2009). Copenhagen accord.

Waltz, K. N. (1979). Theory of International Politics. New York: Random House.

Wassermann, S. and K. Faust (1994). Social network analysis: Methods and applications.Cambridge, New York: Cambridge University Press.

Weible, C. M. (2005). Beliefs and perceived influence in a natural resource conflict: Anadvocacy coalition approach to policy networks. Political Research Quarterly 58 (3), 461–475.

Weidmann, N. B. and K. S. Gleditsch (2010). Mapping and measuring country shapes. TheR Journal 2 (1), 18–24.

Weiler, F. (2012). Determinants of bargaining success in the climate change negotiations.Climate Policy 12 (5), 552–574.

Wendt, A. (1992). Anarchy is what states make of it: the social construction of powerpolitics. International Organization 46 (2), 391–425.

Zartman, I. W. (1994). Introduction: Twos company and moresa crowd: the complexitiesof multilateral negotiation. In I. W. Zartman (Ed.), International Multilateral Negotia-tions. Approaches to the Management of Complexity, pp. 1–12. San Francisco: Jossey-BassPublishers.

113

Appendix 1: Various centrality measures

Country Betweenness Degree Eigenvalue1 China 0.22 0.07 0.062 Russian Federation 0.11 0.03 0.033 Saudi Arabia 0.10 0.05 0.044 South Africa 0.05 0.03 0.035 Mexico 0.05 0.03 0.036 Brazil 0.05 0.05 0.047 EU 0.05 0.04 0.048 Argentina 0.04 0.04 0.049 Australia 0.04 0.03 0.03

10 Japan 0.03 0.03 0.0411 Canada 0.03 0.03 0.0412 India 0.03 0.04 0.0413 Singapore 0.03 0.03 0.0314 Norway 0.03 0.03 0.0415 Bahrain 0.02 0.01 0.0116 New Zealand 0.02 0.03 0.0417 Kuwait 0.01 0.02 0.0218 Philippines 0.01 0.02 0.0319 Tuvalu 0.01 0.02 0.0220 Pakistan 0.01 0.02 0.0221 Thailand 0.01 0.01 0.0022 Micronesia 0.01 0.02 0.0223 Switzerland 0.01 0.02 0.0224 US 0.01 0.02 0.0325 Bolivia 0.00 0.02 0.0226 Oman 0.00 0.02 0.0127 Ethiopia 0.00 0.02 0.0228 Iceland 0.00 0.01 0.0129 Bangladesh 0.00 0.01 0.0130 Republic of Korea 0.00 0.01 0.0131 Egypt 0.00 0.01 0.0132 Gambia 0.00 0.00 0.0033 Venezuela 0.00 0.01 0.0134 Peru 0.00 0.00 0.0135 Barbados 0.00 0.00 0.0036 Algeria 0.00 0.01 0.0237 Antigua and Barbuda 0.00 0.00 0.0038 Belarus 0.00 0.00 0.0039 Belize 0.00 0.00 0.0040 Benin 0.00 0.00 0.0041 Bhutan 0.00 0.00 0.00

114

Country Betweenness Degree Eigenvalue42 Burkina Faso 0.00 0.00 0.0043 Burundi 0.00 0.00 0.0044 Cambodia 0.00 0.00 0.0045 Cameroon 0.00 0.00 0.0046 Chile 0.00 0.00 0.0047 Colombia 0.00 0.01 0.0148 Cook Islands 0.00 0.00 0.0049 Costa Rica 0.00 0.00 0.0050 Croatia 0.00 0.00 0.0051 Cuba 0.00 0.01 0.0152 Ecuador 0.00 0.00 0.0053 Gabon 0.00 0.01 0.0154 Ghana 0.00 0.00 0.0055 Grenada 0.00 0.00 0.0056 Guatemala 0.00 0.00 0.0057 Guyana 0.00 0.01 0.0158 Indonesia 0.00 0.01 0.0159 Iran 0.00 0.00 0.0060 Jamaica 0.00 0.00 0.0061 Kazakhstan 0.00 0.00 0.0062 Kenya 0.00 0.00 0.0063 Kyrgyzstan 0.00 0.00 0.0064 Liberia 0.00 0.00 0.0065 Malawi 0.00 0.00 0.0066 Malaysia 0.00 0.01 0.0167 Maldives 0.00 0.00 0.0068 Mali 0.00 0.00 0.0069 Marshall Islands 0.00 0.00 0.0070 Mauritania 0.00 0.00 0.0071 Nepal 0.00 0.00 0.0072 Nigeria 0.00 0.00 0.0073 Palau 0.00 0.00 0.0074 Panama 0.00 0.01 0.0175 Papua New Guinea 0.00 0.00 0.0076 Paraguay 0.00 0.00 0.0077 Qatar 0.00 0.00 0.0078 Republic of Congo 0.00 0.00 0.0079 Rwanda 0.00 0.00 0.0080 Saint Lucia 0.00 0.00 0.0081 Saint Vincent and the Grenadines 0.00 0.00 0.0082 Samoa 0.00 0.00 0.0083 Senegal 0.00 0.00 0.00

115

Country Betweenness Degree Eigenvalue84 Sierra Leone 0.00 0.00 0.0085 Solomon Islands 0.00 0.00 0.0086 Sri Lanka 0.00 0.00 0.0087 Sudan 0.00 0.01 0.0188 Tajikistan 0.00 0.00 0.0089 Tanzania 0.00 0.00 0.0090 Togo 0.00 0.00 0.0091 Turkey 0.00 0.00 0.0092 Uganda 0.00 0.00 0.0093 Ukraine 0.00 0.00 0.0094 United Arab Emirates 0.00 0.00 0.0095 Uruguay 0.00 0.01 0.0196 Zambia 0.00 0.00 0.00

116

AOSIS in the UNFCCC negotiations: Fromunity to fragmentation?

Carola Betzolda, Paula Castrob & Florian Weilera

a Center for Comparative and International StudiesFederal Institute of Technology, Zurich

b Department for Political ScienceUniversity of Zurich

Paper published in Climate Policy 12(5)

Abstract

Small island states were able to obtain some remarkable achievements in the climatechange negotiations by building a cohesive coalition, the Alliance of Small Island States(AOSIS). The cohesion of the Alliance, however, has been affected by changes in theUNFCCC process. The multiplication of issues on the climate agenda and the in-creasing number of negotiation groups may help or hinder compromise and findingcommon ground. To track how AOSIS has fared in the climate change regime, thispaper compares the activities and positions of AOSIS as a group, and of individualAOSIS members over three distinct periods in the climate change regime: its earlyphase from 1995 to 2000; an implementation phase from 2001 to 2005; and the morerecent period from 2006 to 2011. Over time, group activity has declined in relativeterms, with some issues such as forestry receiving particular attention from individualAOSIS members. Despite controversies in some areas, AOSIS has remained a tighlycoordinated and cohesive alliance that continues to be a key player in global climatepolicy.

Keywords: Alliance of Small Island States (AOSIS), coalitions, fragmentation, cli-mate change negotiations, UNFCCC

1 Introduction

When, in 1990, island countries worldwide recognised the disproportionate vulnerability

of their territories and populations to the negative consequences of climate change, they

came together in a negotiating group, the Alliance of Small Island States (AOSIS). AO-

SIS’s main purpose is to defend island interests in the international negotiations under the

United Nations Framework Convention on Climate Change (UNFCCC), where it can point

to some remarkable accomplishments. Despite the smallness and lack of political clout of

its members, AOSIS has become one of the key players in the UNFCCC negotiations. This

recognition itself is a notable success for island microstates. Further achievements include the

specific small island developing states (SIDS) seat on the various bodies established under

the Convention and its 1997 Kyoto Protocol, or, more recently, consideration to strengthen

the goal of keeping global temperature rises below 1.5◦C. Much of this success is related to

SIDS forming a tight coalition that allowed members to overcome some of their individual

limitations and make their voice heard (Ashe, et al., 1999; Betzold, 2010; McMahon, 1993).

Since the foundation of AOSIS, however, the UNFCCC process has undergone profound

changes. Not only are more and more issues placed under the ever-growing climate change

agenda; also, more and more country groups are formed in the negotiations, with diverging

positions on the various agenda items. By now, a plethora of overlapping country groups

exist in the negotiations (see Figure 1), from single-issue coalitions like the Coalition of

Rainforest Nations founded in 2005, to the leftist Bolivarian Alliance of the Peoples of our

Americas (ALBA) created in 2004.

Presumably, this growth in coalitions makes it more difficult for any one of them to get

their voice heard; similarly, the multiplication of issues has implications for the coordination

among coalition members. On the one hand, it may be more difficult to find common

ground as individual interests and concerns on specific agenda items become more visible.

On the other hand, and quite to the contrary, the multiplication of topics may facilitate

compromising through issue linkages and side payments.

This paper hence takes the fragmentation of the negotiating process as its starting point

and asks to what extent the multiplication of issues as well as country groups has affected

AOSIS. Has the cohesiveness of the Alliance, one of its key characteristics and strengths,

diminished over time, as issues multiplied and differences among members may have become

more visible? Or, to the contrary, has group cohesion remained stable or even increased as a

broader agenda has given compromising more space? Or alternatively, does group cohesion

reassert itself on the most fundamental questions such as mitigation commitments, even

when more complex relationships are at play?

118

Figure 1: Country groups in the climate change negotiations

Bahrain Bosnia and Herzegovina Brunei Darussalam Colombia D. P. R. Korea Jordan Lebanon Mongolia Oman Peru Philippines Sri Lanka Syrian Arab Republic

Azerbaijan The former Yugoslav Republic of Macedonia Israel Kyrgyzstan Montenegro San Marino Serbia Tajikistan

African Group

ALBA

Coalition of Rainforest Nations*

Belarus Croatia Monaco Turkey

Botswana Côte d'Ivoire Egypt Morocco Namibia Swaziland Tunisia Zimbabwe

Cameroon Congo Gabon Ghana Kenya

Afghanistan Bhutan Cambodia Lao P. D. R. Myanmar Nepal Yemen

Mexico South Korea

Kiribati Tuvalu Samoa Solomon Islands Vanuatu Haiti Maldives Timor-Leste

Cook Islands Nauru Niue Palau Dominican Republic Fiji Guyana Papua New Guinea Suriname Bahamas Barbados Micronesia Grenada Jamaica Marshall Islands Singapore Saint Kitts and Nevis Saint Lucia Tonga Trinidad and Tobago Antigua and Barbuda Cuba Saint Vincent and the Grenadines Dominica

Benin Burkina Faso Burundi Chad Djibouti

Eritrea Ethiopia Gambia Guinea Malawi Mali Mauritania Mozambique Niger Rwanda Senegal Somalia Sudan Tanzania Togo Zambia

Central African Republic D. R. Congo Equatorial Guinea Lesotho Liberia Madagascar Sierra Leone Uganda

Comoros Guinea-Bissau Sao Tome and Principe

Cape Verde Mauritius Seychelles

Qatar Saudi Arabia United Arab Emirates

Albania Armenia Georgia Kazakhstan Republic of Moldova Uzbekistan

Costa Rica El Salvador Guatemala Honduras Panama

Bolivia

AOSIS

Nigeria Algeria Libya

LDCs

OPEC

CACAM

G77 / China

Liechtenstein Switzerland

Australia Canada Iceland Japan New Zealand Norway Russia Ukraine United States

Annex I

Non-Annex I

Environmental Integrity Group

Umbrella

SICA

Argentina Chile Malaysia Indonesia Pakistan Paraguay Thailand Uruguay Viet Nam

Bangladesh

EU

Turkmenistan n

Austria Belgium Bulgaria Cyprus Czech Republic Denmark Estonia Finland France

Germany Greece Hungary Ireland Italy Latvia Lithuania Luxembourg Malta

Netherlands Poland Portugal Romania Slovakia Slovenia Spain Sweden United Kingdom

Nicaragua

* countries in italics form part of the Coalition of Rainforest Nations.

Belize

Brazil China India

Angola Venezuela

Iran Iraq Kuwait

South Africa

BASIC

Ecuador

119

In order to map common positions as well as differences in views and priorities over time,

at least as far as they appear to wider audiences, this article relies on public data available for

the entire period of analysis, including official submissions from AOSIS members; reports of

the negotiations in the Earth Negotiations Bulletin; and the lists of participants to selected

meetings. It compares these sources over three distinct periods in the climate change regime:

its early phase from 1995 to 2000; an implementation phase from 2001 to 2005; and the recent

period from 2006 to 2011 focusing on a follow-up to the 1997 Kyoto Protocol and its first

commitment period. Information obtained from interviews with delegates backs up some of

the findings for the most recent period.

The data indicate changes over time. First, submissions and interventions as a group have

decreased relative to individual activities. Differences in positions become more evident when

looking at specific issue areas, particularly those related to Land Use, Land-Use Change and

Forestry (LULUCF) and Reduced Emissions from Deforestation and Forest Degradation

(REDD). Possibly, this suggests that, while AOSIS still remains a fairly tight negotiating

coalition, it has become more difficult to uphold unity.

The next section briefly surveys existing literature on AOSIS in the climate change ne-

gotiations; followed by insights on coalition and group cohesion from negotiation theory and

an overview over the data and methods used. Section 5 compares then AOSIS’s positions

generally, as well as with regard to adaptation, mitigation, LULUCF and REDD, over the

three periods outlined above. Section 6 summarizes and concludes.

2 AOSIS’ sources of bargaining success

Despite important differences in terms of culture, language, and geography, SIDS face com-

mon challenges, including their disproportionate vulnerability to the adverse effects of climate

change (Kelman and West, 2009; Mimura et al., 2007; Wong, 2011). Early on, island states

worldwide recognized this commonality, as well as the need for inter-regional cooperation,

given their very limited individual economic and political clout. Consequently, under the

leadership of the Maldives, Vanuatu and Trinidad and Tobago, 24 island states from all UN

regions formed AOSIS in 1990 as a trans-regional, informal coalition in the negotiations for

the UNFCCC (Chasek, 2005; Heileman, 1993; Taplin, 1994).

Since then, membership has increased to currently 39 full members (AOSIS, 2011; Fry,

2005) that work together largely based on consultation and coordination (Honore, 2004,

p.7).1 Although AOSIS has somewhat broadened its scope (see Chasek, 2005; Fry, 2005), its

main focus remains on the climate change negotiations. Here, AOSIS is by now recognized as

1 Interview with delegate from an AOSIS member country, 4th June 2010, Bonn.

120

a major player (Yamin and Depledge, 2004) - no small feat for these microstates that, even

combined, have less than 1% each of world territory, population, GDP, and greenhouse gas

emissions.2 Beyond recognition, the Alliance can point to some remarkable accomplishments.

Most prominently, SIDS obtained a seat on the Bureau, a position that until then had been

the privilege of the five UN regional groups.3. AOSIS has managed to perpetuate this

key achievement, and managed to obtain a SIDS seat in other UNFCCC bodies, such as

the Executive Board of the Clean Development Mechanism (CDM) or the boards of the

Adaptation Fund or the Green Climate Fund.4

Early studies on the UNFCCC process thus ascribe considerable influence to AOSIS.

Davis (1996, p.18), for instance, argues that “these small and relatively powerless developing

states have managed to exert a profound and continuing impact on global climate policy”

while former AOSIS negotiators Ashe et al. (1999, p.209) even claim that the UNFCCC

“represented a singular triumph [for AOSIS]” (see also Betzold, 2009, 2010; Shibuya, 1996;

Taplin, 1994).

Several factors have been identified as important in explaining the remarkable influence

of these otherwise fairly powerless countries. Davis (1996) lists four main factors: the “truth

and justness of its cause” (p. 19), the support by the best available scientific evidence, the

Alliance’s sense of unity due to the common threat of climate change, and the strong and

skilled leadership by AOSIS’s first chair, ni-Vanuatu ambassador Robert Van Lierop. What

Davis calls “truth and justness” is generally referred to as vulnerability. This extreme sen-

sitivity of small islands to the consequences of climate change gives AOSIS moral leverage.

Larson (2003, 2005) hence argues that AOSIS successfully highlighted their strong exposure

to changing climatic conditions, as well as the negative effects of climate change for all coun-

tries worldwide, which helped to forge coalitions with more powerful groups of countries,

especially the EU and more progressive countries within the G-77 and China. In a similar

vein, the group’s former vice-chair Tiuloma Neroni Slade (2003, p.534) underlines the coop-

erative nature and consensus orientation of small island state diplomacy more generally, as

well as the inclination toward coalitions and like-minded countries. He notes that islands

instinctively [...] recognise strength in acting together, whether as regional sub-

groups of the Caribbean or Pacific countries, or as the larger Alliance of Small

Island States. (p.534)

2 Figures are for 2009, and for 2005 for emissions, see Betzold (2010).3 See rule 22.1 of the draft Rules of Procedure (FCCC/CP/1996/2)4 These seats are hard fought for, as two anonymous reviewers stressed. See CMP1 decisions

(FCCC/KP/CMP/2005/8/Add.1), decision 1/CMP.4 (FCCC/KP/CMP/2008/11/Add.2) and decision1/CP.16 (FCCC/CP/2010/7/Add.1).

121

These soft negotiation strategies also figure prominently in Betzold (2010). According

to her analysis of the climate regime up to the 1997 Kyoto Protocol, AOSIS managed to

highlight common interests, raise moral concerns, as well as “play by the rules”. AOSIS as

a group early on participated very actively in the process, was well prepared and enjoyed

a first-mover advantage vis-a-vis other groups (see also Ashe et al., 1999; McMahon, 1993).

This early full participation, however, was only possible by forming a coalition and pooling

resources, since SIDS individually have limited negotiating capacity, with many of their

delegations consisting only of one or two representatives (Chasek, 2005; McMahon, 1993;

McNamara and Gibson, 2009).

Participation as a bloc is important for AOSIS’s influence, but it is not always easy

to find a common denominator among 39 countries. Despite their common vulnerability,

small island states are threatened by climate change in different ways. Whereas some states

that consist exclusively of low-lying atolls such as the Maldives, Kiribati or Tuvalu, have to

worry about their very existence as states (Yamamoto and Esteban, 2010), other countries

face serious impacts in coastal zones, but may be able to adapt, such as Belize or Cuba.

Similarly, climate regulations affect AOSIS members differently. With large tropical forest

covers, countries like Papua New Guinea, Suriname or Guyana are interested in compensation

payments as part of REDD, while others, in particular Singapore with its large harbour, have

a special interest in bunker fuels and maritime transport. In other words, as the UNFCCC

process increases in scope and complexity, different and potentially diverging interests should

become more pronounced.

3 Coalitions in multilateral negotiations

What do such lines of divergence imply from a theoretical perspective? The literature on

coalitions in multilateral negotiations highlights two opposite effects of an increase in the

number of issues and interests on coalition cohesion. On the one hand, a broader agenda

provides more opportunities for divergent interests to appear, and thus hinders reaching

common ground (e.g. Costantini et al., 2007). On the other hand, it has been argued,

adding issues might in fact facilitate compromising by allowing for issue linkages and side

payments (Sebenius, 1983).

Coalitions are a defining feature of multilateral negotiations. As soon as there are more

than two parties, negotiators start forming groups (Dupont, 1994, 1996). The main purpose

of such groups is to increase the individual members’ negotiating power and thus their

potential gains (Starkey et al., 2005). The increase in bargaining power, however, comes at a

price. Since the coalition’s position is a compromise of the positions of all coalition members,

122

this price can be relatively high when an individual coalition member’s ideal policy is far

from the coalition’s joint position. In contrast, the cost will be lower the closer individual

preferences are to the overall common position. Building and maintaining a coalition is thus

easier among homogeneous members (Axelrod, 1970; Costantini et al., 2007; Garrett and

Tsebelis, 1996).

In single-issue negotiations, it should be relatively easy to identify common interests

and agree on a common position. In contrast, more issues provide more opportunities for

diverging interests to appear among coalition members, and thus make it more difficult to

hold the collective together. From this perspective, it might be expected that it has, over

time, become more difficult for AOSIS to uphold its unity. Since AOSIS countries differ

in how climate change affects them, they value certain issues very differently, and their

individual interest may thus be relatively far from the coalition position, at least in certain

areas. Furthermore, with a better understanding of climate change and its implications as

well as of the negotiation process, individual states may be better aware of their interests

and how they relate to group positions. Lines of divergence may thus be expected to be more

visible now as compared to the early years of the climate change regime, when uncertainty

was even more prevalent.5

On the other hand, it has been proposed that adding issues may in fact help compromis-

ing. More issues that are negotiated simultaneously provide opportunities for issue linkages

and side payments. Thus, if country A is reluctant to agree to the joint position on one

issue, the coalition might be able to get that support by in return promising A to support it

on another issue that is valued highly by A (Sebenius, 1984).

Such exchanges, however, are only possible if there are many, differently valued issues on

the agenda. Hence, cooperation across many issues is used to explain why very heterogeneous

groups like the Group of 77 (G77) have been able to maintain cooperation despite diverse

interests (e.g. Najam, 2004; Vihma et al., 2011).

According to this line of reasoning, then, AOSIS unity should not have suffered from the

multiplication of issues on the UNFCCC agenda. Because AOSIS members value different

items differently, adding them onto the agenda opens up room for compromise, and hence fa-

cilitates coalition maintenance. Further, the growing certainty of climate change may, rather

than highlight divergences, in fact serve to emphasize the overarching common interest: a

strong climate change regime in the face of island vulnerability.

5 We thank Joyeeta Gupta for bringing up this point.

123

4 Data and methods

The empirical analysis compares three distinct periods of negotiations:

• A first period from 1995 to 2000 centred on the design of the Kyoto Protocol.

• A second period from 2001 to 2005, starting with the Marrakesh Accords that focused

negotiations on the detailed rules and operationalization of the Kyoto Protocol and its

flexibility mechanisms.

• A third period from 2006 to 2010, in which the focus shifted to negotiations about a

second commitment period for the Kyoto Protocol and an eventual new protocol.

For such a comparison, data on the negotiations since 1995 is needed. Therefore, the

paper relies on submissions by AOSIS and its member states to the UNFCCC as well as lists

of participants to key negotiation sessions. This material is supplemented with negotiation

summaries as published in the Earth Negotiations Bulletin (ENB) as well as interviews with

AOSIS negotiators.

Governments, usually upon request, provide submissions to share views and perspectives

on specific topics, and to allow chairs and the Secretariat to compile text for negotiations.

All submissions by AOSIS and its member states were manually coded in terms of their

author(s), possible co-authors, as well as content based on the general topic and word counts

of specific markers as listed in Table 1.6

For UNFCCC meetings in years in which major stepping stones in the climate regime were

achieved (Table 2)7, information about the composition of AOSIS delegations was extracted.8

For each AOSIS delegate, information on the type of their affiliation (government or non-

governmental) as well as their detailed background (e.g. type of ministry for governmental

delegates) was coded.

For the period between COP13 in Bali (December 2007) and COP15 in Copenhagen

(December 2009), summaries of the open negotiation sessions from the ENB (International

Institute for Sustainable Development, IISD, 2009) were also hand-coded. Count variables

were created that provide information on how often a country made an intervention on a

6 As two of the anonymous reviewers pointed out, submissions are often called for on technical issues orwhere progress is difficult to obtain, so comparisons across topics may be biased. Despite this, submissionscan still highlight differences in priorities or perspectives among parties on the issues on which they areavailable.

7 See the UNFCCC website at http://unfccc.int/meetings/archive/items/2749.php.8 All lists of participants are available online from the UNFCCC website at

http://unfccc.int/documentation/documents /items/3595.php.

124

specific negotiation topic, and how often statements were supported or opposed by another

country (see Michaelowa et al., 2011).9

Table 1: Negotiation topics and related keywords

Negotiation topic KeywordsAdaptation Adapt

VulnerMitigation Mitig

ReducCommitmTarget

LULUCF and REDD LULUCFREDDForest

Market mechanisms MarketFinance and support Support

FinancFund

Technology transfer, capacity building TechnolCapacity

Impact of response measures Responsemeasure

Finally, interviews conducted in the context of the wider study on climate change ne-

gotiations (Weiler, 2012), were analysed to get some more insight into individual country

positions in the current round of negotiations and their relationship to AOSIS.

A comparison of the group with the individual country level can map differences in views

and priorities, as well as changes over time and across issue areas, at least as portrayed to

wider audiences in the submissions and interventions. It is clear that this material does not

convey information on internal processes, with much of the negotiations occurring behind

closed doors, nor does it provide insights on motivations behind observed changes. Nonethe-

less, the picture obtained through this analysis provides a useful starting point for tracking

the evolution of AOSIS over time.

9 The ENB can only report on negotiation sessions open to observers. Our coding thus assumes that thepositions and behaviour revealed in these open sessions are good proxies for the overall negotiation behaviourof parties.

125

Table 2: Negotiation meetings in which the participant lists were coded

Meeting Location/Date ImportanceCOP1 Berlin, April 1995 First COP, UNFCCC entered into forceSB6 Bonn, August 1997 Year in which Kyoto Protocol was adoptedCOP3 Kyoto, December 1997 Adoption of the Kyoto ProtocolSB12 Bonn, June 2000 Negotiations on the detailed rules of the

Kyoto ProtocolCOP6 The Hague, November 2000 Negotiations on the detailed rules of the

Kyoto ProtocolCOP6bis Bonn, July 2001 Negotiations on the detailed rules of the

Kyoto ProtocolCOP7 Marrakesh, October 2001 Adoption of the Marrakesh Accords (de-

tailed rules of the Kyoto Protocol)SB22 Bonn, May 2005 Year in which the Kyoto Protocol entered

into forceCOP11 Montreal, December 2005 The Kyoto Protocol enters into force; ini-

tiation of the negotiations towards a sec-ond commitment period (Ad Hoc WorkingGroup on Further Commitments for An-nex I Parties under the Kyoto Protocol,AWG-KP)

COP13 Bali, December 2007 Adoption of the Bali Action Plan; initia-tion of the negotiations towards a compre-hensive long-term climate agreement (AdHoc Working Group on Long-term Co-operation under the Convention, AWG-LCA)

SB28 Bonn, June 2008 AWG-KP and AWG-LCA continueCOP14 Poznan, December 2008 AWG-KP and AWG-LCA continueSB30 Bonn, June 2009 AWG-KP and AWG-LCA continueCOP15 Copenhagen, December 2009 AWG-KP and AWG-LCA are supposed to

finish their work; Copenhagen AccordSB32 Bonn, June 2010 AWG-KP and AWG-LCA continueCOP16 Cancun, December 2010 Cancn Agreements

5 Results and discussion

5.1 AOSIS’ and AOSIS members’ interests over time

5.1.1 Written submissions

Figure 2 shows developments in the amount of written submissions sent by AOSIS and

its member countries to the UNFCCC in the three periods of analysis. Three types of

126

Figure 2: Count of AOSIS and AOSIS member written submissions

VanuatuTuvalu

Trinidad and TobagoTimor−Leste

SurinameSolomon Islands

SingaporeSeychelles

SamoaPapua New Guinea

NauruMicronesia

MauritiusMarshall Islands

HaitiGuyana

GrenadaFiji

Dominican RepublicCuba

Cook IslandsComoros

Cape VerdeBelize

BarbadosAntigua and Barbuda

AOSIS

5 10 15 20 25

1995−2000

5 10 15 20 25

2001−2005

5 10 15 20 25

2006−2011

Individual submissionsJoint submissionsGroup submissions

submissions were differentiated: those made by AOSIS as a group; those made by individual

AOSIS members; and those by AOSIS members jointly with other countries (which may or

may not be AOSIS members themselves).10 In the first two periods (i.e. between 1995 and

2005), most submissions were made by AOSIS as a group, with relatively few individual

or joint submissions. From 2006 on, however, the majority of AOSIS countries have made

at least one submission independently of AOSIS. While AOSIS group submissions are still

high in number, their proportion, when compared to the individual or joint submissions, has

declined notably.

Most active are Belize, Papua New Guinea, Singapore and Tuvalu, with several countries

also having a relatively high amount of joint submissions.11 The Dominican Republic, for

example, frequently makes submissions with other Latin American countries outside AOSIS,

and did so already in the 1990s.

10 Group submissions are typically submitted by the Chair of the Alliance on behalf of AOSIS andwere hence counted as a submission by AOSIS and not as a submission by the country holding the Chair.Submissions made by two or more AOSIS members jointly were counted more than once.

11 It should be noted that among the most active AOSIS countries tend to be those who invite highlyskilled external experts to join their delegation. Tuvalu’s activism, for example, can be attributed to its chiefnegotiator Ian Fry, an Australian-born former Greenpeace activist. Other examples are Kevin Conrad, PapuaNew Guinea’s UN Special Envoy and Ambassador for Climate Change and Environment, or a representativeof the Foundation of International Environmental Law and Development serving on Micronesia’s delegation.Thus, it seems that skilled leadership and outside expertise play an important role for small island states inthe climate change negotiations.

127

Figure 3: Main topics of AOSIS and AOSIS member written submissions (percentage oftotal submissions within the period)

Other issues

Multiple issues

Protocol

REDD

LULUCF

CDM

Finance &Technology

Mitigation

Adaptation

10 20 30 40 50

1995−2000

10 20 30 40 50

2001−2005

10 20 30 40 50

2006−2011

With regard to content, the main topic of the submissions as well as an analysis of

keywords12 yield a similar picture. Figure 3 shows changes in the relative importance of

the different main topics over time. In general, topics related to climate mitigation were

very important in the 1990s in the run-up to Kyoto and again from 2006 on. Among these

topics, LULUCF and the CDM, that is, the detailed rules about how to operationalize

the Kyoto Protocol, were more important in the 1990s, while in 2006-2011, more general

mitigation targets and REDD have seen most submissions. The topic of adaptation, in theory

very important for the subsistence of small island states, is generally less prominent in the

submissions than mitigation, probably because it is a less contentious topic than mitigation

targets. Surprisingly, finance and technology appear to have been more important in the

two first periods than in the last one in relative terms, although in recent years negotiations

on a new financial mechanism of the Convention have gained in relevance. Not surprisingly,

submissions regarding a protocol were important in the 1990s (towards Kyoto) and from

2006 on (new protocol, or reform of Kyoto).

Overall, this descriptive analysis hints towards a reduced importance of AOSIS group

submissions in the latest negotiation round, while at the same time the different negotiation

topics have varied in importance, or new topics have emerged. Does the decrease in group

submissions simply reflect shifts in the climate change agenda, or is there a genuine time

trend toward individual rather than group activity?

A more detailed analysis of the topics for which group or individual submissions predomi-

nate can shed light on this question. Keyword counts in submissions by countries reveal that

AOSIS as a group remains prominent in submissions related to adaptation or vulnerability,

financial support, and technology or capacity building. If, in contrast, individual submis-

sions are considered as an indicator of possible diverging positions within the group, some

12 The data and detailed analysis are not shown due to space reasons, but are available from the authorson request.

128

topics show a larger dispersion of interests: with respect to mitigation commitments, Tuvalu,

Papua New Guinea, Vanuatu and Dominican Republic are the most active AOSIS countries,

followed by Solomon Islands and Singapore, mainly in the period 2006-2011. However, this

does not indicate divergence. A more detailed analysis (see below) reveals that many of

these individual submissions are used to reiterate and reinforce group positions, such as

the demand for emissions cuts in the order of 40% compared to 1990 levels. Land-use and

forestry issues were mentioned most frequently by Tuvalu, followed by Vanuatu, Papua New

Guinea, Dominican Republic, Solomon Islands, Belize, and Singapore. The interest of most

of these countries in the forestry sector seems to have started only during the 2006-2011

period, which points towards a special focus on REDD - and here, individual submissions in-

deed point toward different viewpoints, with different countries proposing divergent ways of

dealing with forests. The word “market” follows a very similar pattern to the terms related

to forestry. Interestingly, both for forestry and markets, some individual countries appear

to be more active than AOSIS as a group, as revealed by the fact that the word counts are

larger for these individual countries than for AOSIS group submissions (see e.g. Figure 6

below).

5.1.2 Oral interventions

The analysis of the oral interventions in the negotiations, as reported in the ENBs, shows

similar patterns. Table 3 compares the topics that, according to the ENB coding, were

most relevant for AOSIS as a group and for the AOSIS countries that intervened more

than ten times in the period between Bali and Copenhagen. While AOSIS as a group

has participated repeatedly on topics such as adaptation, mitigation, finance and capacity

building or technology transfer, which are of general interest to all vulnerable countries,

it has made very few group interventions on LULUCF and REDD. Some individual AOSIS

members, however, have participated actively in the LULUCF and REDD discussions, among

them Tuvalu, Papua New Guinea, Guyana, Singapore and Micronesia.

The coding of the ENBs also shows some instances in which AOSIS member countries

have openly held opposing positions in the negotiations. In the Bonn meeting in August

2009, Papua New Guinea and Tuvalu were reported to have opposing views on LULUCF

accounting and on LULUCF eligibility under the CDM (Fry, 2008). The forestry sector

thus appears to be one of the contentious issues among SIDS. But other issues have also

generated disagreement: In the Bangkok meeting in October 2009, Singapore joined some

non-AOSIS countries in proposing that the International Civil Aviation Organization and

the International Maritime Organization take the lead in regulating emissions from aviation

129

Table 3: Number of oral interventions of most active AOSIS countries per negotiation topic,December 2007 - December 2009

Country A&V M&C KFM SM&NP MRV LU RE FIN CB CCP SVAOSIS 51 76 26 1 12 3 1 65 32 8 14Tuvalu 7 40 27 7 2 20 16 16 0 4 3Singapore 1 23 5 3 1 0 2 0 0 3 1Micronesia 4 17 8 4 4 1 0 1 0 0 1PNG 0 5 3 0 0 10 18 3 1 0 0Guyana 1 3 0 0 0 1 15 2 2 0 1Barbados 2 3 0 1 0 0 0 4 2 4 2

A&V = Adaptation and vulnerabilityM&C = Mitigation and complianceKFM = Kyoto flexibility mechanismSM&NP = Sectoral mechanisms and national policiesMRV = Monitoring, reporting and verificationLU = Land Use, Land Use Change, and Forestry (LULUCF)RE = Reducing Emissions from Deforestation and Forest Degradation (REDD+)FIN = FinanceCB = Capacity building, technology transfer, R&DCCP = Consequences of climate policySV = Shared visionPNG = Papua New GuineaSource: Earth Negotiation Bulletin (IISD, 2007-2009), own coding.

and maritime transport (which was later supported by Cook Islands13), whereas Tuvalu and

Micronesia suggested that such regulations need to be guided by the Convention. At COP15,

Papua New Guinea reportedly stated that they did not support the AOSIS proposal for a

continuation of the Kyoto Protocol and an additional protocol to enhance action under the

Convention.

5.1.3 Country delegations

Over time, the delegations of small island states to the UNFCCC meetings have grown im-

portantly in size, as shown in Table 4. Especially for COP meetings in which important

decisions are expected, the aggregated AOSIS delegation has become quite large. If coordi-

nation among AOSIS members is high, such a delegation is an important resource for small

island states. Closer analysis shows that the growth in delegation size has not been equal

across AOSIS members - Singapore, Papua New Guinea, Samoa, Micronesia and Tuvalu are

the countries that have had the largest delegations at some point and are thus the focus of

the analysis below.

13 See submission by Cook Islands in FCCC/AWGLCA/2010/MISC.2/Add.1, p.8-9.

130

These differences across AOSIS members could be due simply to different economic or

human resources, or different ways of dealing with national delegations14, but they could

also signal a diversification of interests within some AOSIS members as the negotiations

progressed.15 Broadly, the core of the delegations should be composed of representatives from

agencies related to environment, climate change and meteorology, and the foreign service,

which have traditionally negotiated the climate change issue. Many representatives from the

ministry of finance, economy or development may be an indicator of concerns about how

to finance climate-related action. Representatives from other governmental sectors or from

business may indicate the existence of other interests.

Table 4: Negotiation meetings in which the participant lists were coded

Meeting Date Total AOSIS Mean Min Max s.d. MostAOSIS share (AOSIS) (AOSIS) (AOSIS) delegates

COP1 April 1995 67 8.85% 2.09 1 5 1.18 MicronesiaSB6 August 1997 31 n/a 1.55 1 5 1.02 SingaporeCOP3 December 1997 115 7.50% 3.83 1 15 2.98 MicronesiaSB12 June 2000 39 4.84% 1.56 1 5 0.98 SamoaCOP6 November 2000 153 6.97% 4.25 1 12 2.49 MicronesiaCOP6bis July 2001 117 6.45% 3.34 1 9 2.19 SamoaCOP7 October 2001 61 2.53% 2.26 1 6 1.35 SamoaSB22 May 2005 45 4.86% 1.61 1 5 1.08 TuvaluCOP11 December 2005 137 4.89% 3.91 1 15 3.13 PNGCOP13 December 2007 344 9.81% 9.05 1 61 11.80 SingaporeSB28 June 2008 94 7.15% 2.76 1 17 2.67 SingaporeCOP14 December 2008 220 5.56% 5.64 1 27 5.56 SingaporeSB30 June 2009 121 6.92% 3.36 1 19 3.71 SingaporeCOP15 December 2009 638 6.03% 16.36 5 82 14.43 PNGSB32 June 2010 143 8.57% 3.86 1 28 4.62 SingaporeCOP16 December 2010 418 8.06% 11.00 3 41 9.33 Singapore

Source: Participant lists to UNFCCC meeting.

14 While some countries strictly include only members of government in their national delegations, othersare open to including representatives of civil society or NGOs even in cases where these do not contributedirectly to the negotiations. Hence, the size of the delegation is by itself not a good indicator of bargainingresources of the party.

15 An alternative explanation could be that, faced with limited resources, countries within the AOSIScoalition coordinate the composition of their national delegations so that overall they have experts in allnegotiation topics across all SIDS, who can inform each other about the progress in each topic. Even inthis case, having the experts for one particular topic may be a sign of salience of this topic for a particularcountry.

131

Figure 4: Composition of selected AOSIS member delegations, by sector (percentage oftotal delegates in analysed meetings)

Micronesia

Papua New Guinea

Samoa

Singapore

Tuvalu

20 40 60 80

1995−2000

20 40 60 80

2001−2005

20 40 60 80

2006−2011

Environment/Foreign affairsFinance & BusinessEnergyAgricultureOther governmentInternational cooperation & NGOs

Note: See Appendix for a description of how sectors were codedSource: Participant lists to UNFCCC meetings

Figure 4 shows our findings in terms of the composition by sector of the delegations of

the five countries mentioned above. Changes over time and differences across delegations

become evident. Indeed, delegates from “environment or foreign affairs” make up the largest

part of the delegations analysed. Concerns about climate change impacts and reliance on

career diplomats explain a large part of the selected countries’ delegations. From the other

governmental sectors, Singapore is the only country in the sample that includes representa-

tives of the energy sector (since 2001), and Papua New Guinea and Singapore the only ones

with representatives from agriculture (since 2006).

Looking beyond sectors, Figure 5 shows the composition of the five delegations during

2006-2011. The differences across countries become more evident. Specialists on forestry,

the CDM and carbon markets appear only in the delegation of Papua New Guinea, while ref-

erences to energy and aviation or maritime transport seem important only for the delegation

of Singapore.

Such differences confirm differenct priorities across AOSIS member countries. While

climate change and environmental considerations are still the most important topic among

all delegations, more specific issues such as carbon markets, forests and emissions from energy

and transport seem to be relevant agenda items for certain countries, among them Papua

New Guinea and Singapore.

In summary, while some topics appear thus to be negotiated by AOSIS as a group,

others seem to be negotiated by individual member countries. However, this analysis still

cannot show whether these observations reflect a divergence in interests, or a strategy of

132

Figure 5: Representation of interest groups in selected AOSIS member delegations (keywordcounts in analysed meetings)

Micronesia

Papua New Guinea

Samoa

Singapore

Tuvalu

20 40 60 80 100 120

2006−2011

Climate change, meteorology, vulnerabilityCDM, carbon marketsEnergyAviation, maritime transp.Forestry

Note: See Appendix for a description of how sectors were codedSource: Participant lists to UNFCCC meetings

specialization between them. A closer look at the issue areas of adaptation, mitigation,

LULUCF and REDD help to better understand the implications of the observed changes.

5.2 Positions on adaptation and mitigation

Unsurprisingly, adaptation and mitigation figure prominently among the issues of relatively

high importance to AOSIS. Of a total of 176 submissions produced by the group or its mem-

bers since 1995, 17% are dedicated to mitigation, and 11% to adaptation. If one considers

that the early AOSIS protocol proposals were mainly focused on mitigation, and the more

recent ones have very important components of both adaptation and mitigation, then these

figures would grow further. Whether these figures suggest a stronger interest of AOSIS for

mitigation than for adaptation measures, or whether mitigation has simply been a more

contested issue due to the evolution of the negotiations, cannot be concluded clearly from

the analysis of submission counts. However, some evidence does point out that the relative

importance of mitigation versus adaptation varies across AOSIS member countries, which

supports the idea that the SIDS are not an entirely homogeneous block. For example, a

Maldives representative specified in an interview that more money contributed by Annex

I countries should be earmarked for adaptation, where there is no market.16 On the con-

16 Interview with delegate from the Maldives, 10th June 2010, Bonn. This view is also reflected in theonly individual submission discussing mitigation made by the Maldives, which states that “the required levelof financial resources [for adaptation] should be assessed in light of other elements of the proposed outcome

133

trary, some countries such as Papua New Guinea, Grenada, or Vanuatu do not concede

much space to adaptation in their individual submissions or those jointly with non-AOSIS

countries. Papua New Guinea, for example, gives adaptation only room for 0.25% of its

statements, while mitigation gets much more attention (31%; forest receives most attention,

with 40% of statements). This echoes an interview with a delegation member of Papua New

Guinea, who ranked the contact group on enhanced action on mitigation as the most impor-

tant for the country, while the contact group on enhanced action on adaptation is not among

the three top-priority contact groups.17 At the opposite end of the spectrum is Comoros,

which does not consider mitigation in its individual submissions at all, while 63% of the

statements have adaptation as the central topic.

This is a first indication that differences regarding mitigation and adaptation exist wtihin

AOSIS. As described above, there is also a change in the relative importance of group and

individual submissions on mitigation and adaptation over time. During the first period

from 1995 to 2000, 89% of references in the written submissions to adaptation and 71% of

references to mitigation were made on behalf of the group (figures based on keyword counts).

The picture for the second time period, 2001 to 2005, seems relatively stable, with 72% of all

statements in the submissions concerning mitigation and 64% of those concerning adaptation

made on behalf of the group. A downward trend in group submissions becomes clearer in the

third period from 2006 to 2011. A division is particularly pronounced for mitigation, with

only 19% of references in written submissions made on behalf of the group. Tuvalu (21%)

and Papua New Guinea (11%) are leading with regard to the number of individual and joint

statements. During this period, 23 of AOSIS members submitted their individual views to

the UNFCCC on mitigation issues, while during the 2001 to 2006 negotiation stage only

four members felt the need to draft individual submission. The same trend, although less

pronounced, is observed for adaptation, with 43% of references made on behalf of the group

in the final negotiation phase. Again Tuvalu, accounting for 31% of all statements made

by AOSIS members during that time period, is leading the pack. Regarding adaptation, 19

AOSIS members decided to express their views in individual submissions between 2006 and

2011 (significantly more than between 2001 and 2005, when only 7 members made individual

submission).

Do these results imply the decline of within-group unity regarding mitigation and adap-

for Cancun including the expected global goal, Annex I mitigation efforts and the likely resulting impactson developing countries” (see document FCCC/AWGLCA/2010/MISC.2, p. 69). A higher mitigation effortin the developed world, thus, would lead to lower financial needs for adaptation. Given that the Maldivescall for higher targets that the rest of AOSIS (45% instead of 40%), this is congruent with the statementmade in the interview that mitigation should be prioritized over adaptation.

17 Phone interview with delegate from an AOSIS member country, 13th October 2009.

134

tation? A deeper analysis of the positions displayed by SIDS on emission reduction targets

for Annex I countries and on adaptation measures does not support such a divergence of

positions. Analysing the last negotiation period, AOSIS as a group calls for an aggregate

emission reduction of at least 40% by 2020 in the developed world, a view which is reflected

in most individual submissions of AOSIS members, although the Maldives were calling for

even more stringent targets of 45%.18 There also seems to be broad agreement that mitiga-

tion efforts should be based on historic responsibilities. On mitigation, therefore, a higher

level of fragmentation during the last negotiation period cannot be deduced. Individual sub-

missions are either used to reiterate the view of the whole group, or to promote particular

ideas. An example for the latter would be Micronesia’s repeated submissions on “fast start

mitigation strategies”.19 Finally, Tuvalu used its individual submissions on mitigation inter

alia to raise the pressure on Annex I parties by illustrating that these countries contributed

approximately 75% of all anthropogenic CO2 emissions to date.20 In terms of adaptation,

the positions displayed in group and in individual submissions also evidence high agreement

among AOSIS members. Sometimes individual submissions are at the forefront of positions

that are later adopted by the whole group, such as in the case of extending the share of pro-

ceeds to finance adaptation also to Joint Implementation and to Emissions Trading, which

has been pursued by Tuvalu, supported then by individual SIDS and later taken up by the

whole group.21 Sometimes, individual submissions reinforce what has been already proposed

in group submissions, or provide more detail on specific aspects, such as on the institutional

framework for adaptation or on the insurance mechanism. Thus, individual submissions on

mitigation and adaptation seem to reflect an extra effort of small island states to corroborate

their positions, but do not back the hypothesis of increased disunity within AOSIS.

A special case, however, exists on submissions about how to treat the Copenhagen Ac-

cord. Submissions by individual SIDS in 2010 evidence strong disagreements on whether

the text of the Copenhagen Accord should be used for future negotiations under the Con-

vention: while several countries (Barbados, Belize, Maldives, Marshall Islands, Singapore)

mention that contents of the Copenhagen Accord should flow into the negotiations (albeit

18 Submission by the Maldives in document FCCC/AWGLCA/2010/MISC.2, p. 69.19 See for example submission by Micronesia in document FCCC/AWGLCA/2008/MISC.1, p. 41.20 Submission by Tuvalu in document FCCC/KP/AWG/2009/MISC.1/Add.1, p. 10-14.21 The CDM, Joint Implementation and Emissions Trading are instruments of the Kyoto Protocol that

provide flexibility in terms of where to achieve emission reductions. Currently, a 2% share of proceeds fromthe CDM is used to finance adaptation, but such a levy is not applied to Joint Implementation or EmissionsTrading. See proposal by Tuvalu for an International Blueprint on Adaptation in FCCC/CP/2007/Misc.2,and subsequent submissions supporting the share of proceeds expansion in FCCC/SBI/2008/MISC.10,FCCC/AWGLCA/2009/MISC.4, FCCC/AWGLCA/2009/MISC.8 and FCCC/CP/2010/3. It should benoted however that having a share of proceeds for adaptation from all three Kyoto mechanisms was al-ready an AOSIS position during the negotiations towards the Marrakesh Accords in 2001.

135

with improvements), others, such as Cuba and Tuvalu are strongly against it:

It is Tuvalu’s firm view that the Copenhagen Accord should not be the basis for,

or have any influencing role, on the Chair’s text. The Copenhagen Accord is a

fundamentally flawed document.22

Indeed, in an interview in 2010, an AOSIS delegate, commented that there may be “some

degree of concern amongst the AOSIS members that maybe there was some betrayal, maybe

there was some breach of the common trust” when some members associated with the Copen-

hagen Accord.23 However, the same delegate explained that AOSIS as a group had to move

on from such disagreement and keep firm on its main negotiation goals.

5.3 Positions on LULUCF and AOSIS

The general analysis of submissions by AOSIS countries over the period 1995-2011 has re-

vealed a noticeable evolution in the importance of the forestry negotiation topics for this

group of countries. The forestry negotiations encompass rules for how industrialized coun-

tries should account for the sequestration or emission of greenhouse gases from forests and

other land-use activities in their emission inventories and in their emission reduction targets

(negotiations on LULUCF), rules for what types of forestry and land-use activities should

be included in the CDM (LULUCF in the CDM), and, more recently, rules on a possible

new mechanism to address emissions from deforestation and land degradation in developing

countries (REDD negotiations24). As explained above and shown in Figure 6, individual

SIDS seem to be more active than AOSIS as a group in the discussions about forestry issues,

particularly in the period from 2006 on.

The negotiations on forestry-related issues reveal a divide within the AOSIS members,

which started to exist already in the early negotiations in the 1990s. Between 1998 and

2002, AOSIS as a group made five written submissions related to LULUCF, which reveal a

consistently strict position regarding how land-use and forestry activities should be consid-

ered both by the industrialized countrie as part of their mitigation efforts, and by developing

countries under the CDM. Two quotes make this clear:

22 Submissions contained in documents FCCC/AWGLCA/2010/MISC.1,FCCC/AWGLCA/2010/MISC.2, FCCC/AWGLCA/2010/MISC.2/Add.1 andFCCC/AWGLCA/2010/MISC.2/Add.2. Tuvalu’s quote is from the last document listed, p.6.

23 Interview with delegate from an AOSIS member country, 11th April 2010, Bonn.24 The REDD negotiations have been expanded to include also negotiations on the conservation and

enhancement of forests and on sustainable forest management, which is usually known as REDD+. Somecountries also support the inclusion of other land-related activities in the REDD mechanism, such as agricul-ture and related soil carbon content, which is known by experts as REDD++. For simplicity, in this articlewe will generally refer to all these topics as REDD negotiations.

136

AOSIS is in favour of very strict considerations to be met if land use change and

forestry activities are to be included in the mitigation efforts of the industrialised

countries.25

The primary priority should rest with the reduction of emissions and that en-

hancement of sinks is an additional activity in the short term.26

Figure 6: Word counts in AOSIS and AOSIS member submissions (LU-LUCF/REDD/forest), per period

VanuatuTuvalu

Trinidad and TobagoTimor−Leste

SurinameSolomon Islands

SingaporeSeychelles

SamoaPapua New Guinea

NauruMicronesia

MauritiusMarshall Islands

HaitiGuyana

GrenadaFiji

Dominican RepublicCuba

Cook IslandsComoros

Cape VerdeBelize

BarbadosAntigua and Barbuda

AOSIS

200 400 600 800

1995−2000

200 400 600 800

2001−2005

200 400 600 800

2006−2011

A joint submission by Samoa and Tuvalu and an individual submission by Tuvalu, both

from 2000, support this strictness. In addition, Tuvalu asks for limited acceptability of

LULUCF activities as Joint Implementation projects, and for no LULUCF activities in the

CDM during the first commitment period, due to concerns about environmental integrity,

accounting and institutional issues.27 On the other hand, the Dominican Republic, with a

group of Latin American countries, made two submissions proposing which forestry activities

should be included in the CDM. These proposals were much more lenient than those of AOSIS

as a group: they not only state that LULUCF activities should be eligible as CDM projects,

but also ask for an inclusion of activities that slow, reduce or avoid deforestation, including

forest management.28 These submissions thus point toward a certain fragmentation, and

25 Submission by AOSIS in document FCCC/CP/1998/MISC.1, p. 47.26 Submission by AOSIS in document FCCC/SBSTA/1999/MISC.2, p. 47.27 Submission by Tuvalu in document FCCC/SB/2000/MISC.1/Add.2.28 Submissions in documents FCCC/SB/1999/MISC.10/Add.3 and FCCC/SB/2000/MISC.1/Add.2.

137

indicate that individual self-interests may dominate group cohesion on this issue.

The division becomes clearer in the later submissions regarding LULUCF from 2009 on:

in this period, no joint AOSIS submission exists on the topic; instead, there are a host of

individual submissions by Belize, Tuvalu, Singapore and Papua New Guinea,, as well as a

joint submission by Guyana and Papua New Guinea with a large group of other (non-AOSIS)

non-Annex I countries. These submissions point towards diverging interests and opinions.29

It appears likely that AOSIS countries could not agree on a group submission about LULUCF

after 2009, so that individual countries have submitted their positions independently from

each other.

With regard to REDD, the fragmentation of opinions within AOSIS is even more pro-

nounced. The concept of reducing emissions from deforestation was first introduced in the

negotiations jointly by Papua New Guinea and Costa Rica at COP11 in Montreal in 2005.30

Parties agreed to start discussing the topic as a new agenda item, and launched a 2-year

consultation process. At COP13, reducing emissions from forest degradation was also in-

cluded in the discussions, giving place to REDD. Since then, negotiations have continued

on how to address the methodological issues required to measure emission reductions from

deforestation and forest degradation, and on how to generate positive incentives to halt these

emissions (Fry, 2008; Sanz-Sanchez, 2011).

All submissions from SIDS regarding this topic have been made either by individual

countries or by distinct groups of countries. No group AOSIS submission exists on REDD.

Diverging opinions mainly concern questions about whether emission reduction from REDD

activities should be used as offsets in the carbon market in a CDM-type or a sectoral mecha-

nism, whether and how early action by countries that have already made efforts to preserve

their forests should be recognized, and how to address the balance of supply and demand for

carbon credits in the market (on REDD, see Martinet and Christovam, 2009; Verchot and

Petkova, 2009). Belize, the Dominican Republic, Guyana, Papua New Guinea, Singapore,

Solomon Islands, Suriname and Vanuatu are generally pro-markets, pro-recognition of early

action and concerned about prices for carbon offsets. Tuvalu, on the other hand, makes

clear in several submissions that it is against the inclusion of REDD activities in the carbon

market, even in the form of pilot projects, and against granting credits for early action.

Instead, it made a proposal for a non-market REDD mechanism.31

29 The topics of these submissions are mostly technical, e.g. how to better account for LULUCF emissions,what types of activities should be included in LULUCF (in general and in the CDM), and what referencelevels should be used to determine LULUCF emissions.

30 Submission by Papua New Guinea and Panama in document FCCC/CP/2005/MISC.1, p. 2-11.31 Submissions by Tuvalu in documents FCCC/SBSTA/2007/MISC.2/Add.1 and

FCCC/SBSTA/2009/MISC.1/Add.1.

138

With regard to forestry, then, there is a divide between AOSIS members. The number

of individual submissions indicates disagreement and fragmentation, rather than serving to

strengthen a common position as was the case for mitigation.

6 Concluding remarks: AOSIS’ role in the future, unity

versus fragmentation?

While tensions clearly exist, AOSIS remains a tightly coordinated negotiation coalition in

the climate change process. Its members are acutely aware of their need for a strong unified

voice to convince other, larger countries of ambitious action on climate change. As one

interviewee emphasizes, “we can’t fight amongst ourselves, because we are not the enemy.”32

Nonetheless, AOSIS member states are affected by climate change and climate policies in

different ways. It is thus not surprising to note that different AOSIS countries accord different

priorities to different agenda items, as for instance mitigation compared to adaptation or

forestry-related issues.

As the climate change agenda has grown since COP1, AOSIS member states increasingly

participate in the negotiations as individual parties rather than on behalf of the coalition.

In particular, some areas such as LULUCF and REDD, are contentious within AOSIS and

sometimes even provoke open confrontation. At first glance, this may suggest that AOSIS

has become less cohesive and more fragmented over time. A more detailed analysis, however,

indicates that many of the individual contributions reiterate and reinforce group positions.

I sum, then, the Alliance has been able to uphold unity. Although interviewees comment

on internal controversies and criticism, they seem to feel overall that SIDS are a relatively

homogenous group with little disagreement.33 Differences in priorities and capacities are

even harnessed, as the Samoan interviewee explains. Some low-lying atoll countries like

Tuvalu are more vulnerable than Samoa, he says, so “the best we can do for Tuvalu is to

give them their space. Because people will listen more to Tuvalu than to us.”34

Indeed, Tuvalu’s voice and that of AOSIS are listened to in the climate change negotia-

tions. In Durban, the Alliance joined forces with other vulnerable and progressive countries,

and was able to obtain many of its goals, especially regarding adaptation, finance, technology

transfer and capacity building. On mitigation, however, the so-called Durban Package “falls

well short of what these countries wanted - and need to avoid catastrophic climate change

32 Interview with delegate from an AOSIS member country, 4th April 2010, Bonn.33 Interview with delegate from an AOSIS member country, 4th April 2010, Bonn; Interview with delegate

from an AOSIS member country, 4th June 2010, Bonn.34 Interview with delegate from an AOSIS member country, 4th June 2010, Bonn.

139

impacts” (Wold, 2011).

AOSIS remains a key player in global climate policy and one of the most active proponents

of deep cuts in global greenhouse gas emissions. In spite of a proliferation of issues in the

UNFCCC process, the core of AOSIS negotiating position is strong and urgent enough to

keep the Alliance together. Unfortunately, however, emissions cuts need to come from larger

countries that are reluctant to pay heed to the warnings of AOSIS. As cohesive as the Alliance

thus may be, at the end of the day, action must come from other countries.

140

References

Ashe, J., R. van Lierop, and A. Cherian (1999). The role of the alliance of small island states(AOSIS) in the negotiation of the united nations framework convention on climate change(UNFCCC). Natural Resource Forum 23 (3), 209–220.

Axelrod, R. (1970). Conflict of Interest. A Theory of Divergent Goals with Applications toPolitics. Chicago: Markham Publishing Company.

Betzold, C. (2009). The Use of Power Borrowing by Small Island Developing States inNegotiations under the Climate Change Regime. Master thesis.

Betzold, C. (2010). ‘Borrowing’ power to influence international negotiations: AOSIS in theclimate change regime, 19901997. Politics 30 (3), 131–148.

Chasek, P. (2005). Margins of power: Coalition building and coalition maintenance of theSouth Pacific Island States and the Alliance of Small Island States. Review of EuropeanCommunity & International Environmental Law 14 (2), 125–137.

Costantini, V., R. Crescenzi, F. De Filippis, and L. Salvatici (2007). Bargaining coalitionsin the WTO agricultural negotiations. The World Economy 30 (5), 863–891.

Davis, W. (1996). The Alliance of Small Island States (AOSIS): The international conscience.Asia-Pacific Magazine 2, 1722.

Dupont, C. (1994). Coalition theory: Using power to build cooperation. In I. W. Zartman(Ed.), International Multilateral Negotiations. Approaches to the Management of Com-plexity, pp. 148–177. San Francisco: Jossey-Bass Publishers.

Dupont, C. (1996). Negotiation as coalition building. International Negotiation 1 (1), 47–64.

Fry, I. (2005). Small Island Developing States: Becalmed in a sea of soft law. Review ofEuropean Community & International Environmental Law 14 (2), 89–99.

Fry, I. (2008). Reducing emissions from deforestation and forest degradation: Opportuni-ties and pitfalls in developing a new legal regime. Review of European Community &International Environmental Law 17 (2), 166–182.

Garrett, G. and G. Tsebelis (1996). An institutional critique of intergovernmentalism. In-ternational Organization 50 (2), 269–299.

Heileman, L. (1993). The Alliance of Small Island States(AOSIS): A mechanism for coor-dinated representation of small island states on issues of common concern. Ambio 22 (1),55–56.

Honore, S. (2004). Processus d’evaluation generale du Programme d’action de la Barbade10 ans apres [BPoA]. Objectif Terre 6 6 (1), 6–10.

IISD (2007-2009). Earth negotiation bulletin. http://www.iisd.ca/vol12.

141

Kelman, I. and J. West (2009). Climate change and Small Island Developing States: Acritical review. Ecological and Environmental Anthropology 5 (1), 1–16.

Larson, M. J. (2003). Low power contributions in multilateral negotiations: A frameworkanalysis. Negotiation Journal 19 (2), 133–149.

Larson, M. J. (2005). Low-power Pacific contributions to multilateral diplomacy in climatechange negotiations. Paper presented at the Peace, Justice and Reconciliation Asia PacificRegion Conference. March 31 - April 3, Brisbane, Australia.

Martinet, A. and M. Christovam (2009). The Countries Positions on REDD MechanismNegotiations and Determinants of These. Paris: ONF International.

McMahon, V. M. (1993). Environmental nongovernmental organizations at intergovern-mental negotiations. In L. E. Susskind, W. R. Moomaw, and A. Najam (Eds.), Paperson International Environmental Negotiation, Volume 3, pp. 1–21. Cambridge, MA, TheProgram on Negotiation at Harvard Law School: Harvard University Press.

McNamara, K. E. and C. Gibson (2009). ‘We do not want to leave our land’: Pacific ambas-sadors at the United Nations resist the category of ‘climate refugees’. Geoforum 40 (3),475–483.

Michaelowa, K., P. Castro, and L. Hornlein (2011). Path dependence of negotiation stucturesin international organizations: The impact of annex i membership on discussions withinthe UNFCCC. CIS Working Paper 67 .

Mimura, N., L. Nurse, R. McLean, J. Agard, L. Briguglio, P. Lefale, R. Payet, and G. Sem(2007). Small islands. In M. Parry, O. Canziani, J. Palutikof, P. van der Linden, andC. Hanson (Eds.), Climate Change 2007: Impacts, Adaptation and Vulnerability. Contri-bution of Working Group II to the Fourth Assessment Report of the IntergovernmentalPanel on Climate Change, pp. 687–716. Cambridge: Cambridge University Press.

Najam, A. (2004). Dynamics of the Southern collective: Developing countries in desertifica-tion negotiations. Global Environmental Politics 4 (3), 128–154.

Sanz-Sanchez, M. J. (2011). Current status and outcomes of REDD negotiations under UN-FCCC. http://www.cbd.int/doc/meetings/for/wscbredd-apac-01/other/wscbredd-apac-01-unfccc-en.pdf.

Sebenius, J. K. (1983). Negotiation arithmetic: Adding and subtracting issues and parties.International Organization 37 (2), 281–316.

Sebenius, J. K. (1984). Negotiating the Law of the Sea. Cambridge, MA and London:Harvard University Press.

Shibuya, E. (1996). ‘Roaring mice against the tide’: The south pacific islands and agenda-building on global warming. Pacific Affairs 69 (4), 541–555.

142

Slade, T. (2003). The making of international law: The role of small island states. TempleInternational & Comparative Law Jounal 17, 531–43.

Starkey, B., M. A. Boyer, and J. Wilkenfeld (2005). Negotiating a Complex World: AnIntroduction to International Negotiation. Plymouth: Rowman & Littlefield Publishers.

Taplin, R. (1994). International policy on the greenhouse effect and the Island South Pacific.The Pacific Review 7 (3), 271–281.

Verchot, L. V. and E. Petkova (2009). The State of REDD Negotiations: Consensus Points,Options for Moving Forward and Research Needs to Support the Process. Bogor, Indonesia:Center for International Forestry Research.

Vihma, A., Y. Mulugetta, and S. Karlsson-Vinkhuyzen (2011). Negotiating solidarity? TheG77 through the prism of climate change negotiations. Global Change, Peace & Secu-rity 23 (3), 315–334.

Weiler, F. (2012). Preference attainment: Determinants of bargaining success in the climatechange negotiations. Climate Policy 12 (5), 552–574.

Wold, C. (2011). The Durban package and the goals of Pacific Small Island DevelopingStates. American Society of International Law Insights 16 (1).

Wong, P. P. (2011). Small island developing states. Wiley Interdisciplinary Reviews: ClimateChange 2 (1), 1–6.

Yamamoto, L. and M. Esteban (2010). Vanishing island states and sovereignty. Ocean &Coastal Management 53 (1), 1–9.

Yamin, F. and J. Depledge (2004). The international climate change regime: A guide torules, institutions and procedures. Cambridge: Cambridge University Press.

143

Appendix: Coding rules for participant lists

Sector Coding rules for Figure 4

Environment/Foreign affairs Whenever ‘climate change’, ‘environment’, or ‘foreign’ isincluded in the name of the ministry, or for the ‘Ministryof Sustainable Development’. Also includes all heads ofstate, and whenever a diplomat (e.g. an ambassador) ora diplomatic mission (‘permanent mission’, ‘embassy’)is mentioned. Also, whenever a climate change council,office or agency, or an environmental or meteorologicalagency or service is mentioned without specifying an-other ministry.

Finance/Business Whenever ‘finance’, ‘economic’, ‘development’, or ‘plan-ning’ is included in name of the ministry, except if ‘envi-ronment’ is also there. Also Ministry of Infrastructure,Ministry of Home Affairs. Also includes utilities, carbonconsultancies (even international ones), business associ-ations, etc.

Energy Whenever ‘energy’ is included in name of the ministry,except if ‘economic’ or ‘finance’, or ‘environment’ is alsothere.

Agriculture Whenever ‘agriculture’ or ‘forest’ or similar is includedin the name of the ministry, except if ‘environment’ or‘economic’ is also there. Also includes national parks orother conservation agencies, or land management agen-cies, whenever the word ‘environment’ is not included.

Other government Whenever it is clear that the delegate is from the na-tional government (other ministries, parliament, localgovernments, various agencies) but not from any of theabove

International cooperation & NGOs Includes bilateral cooperation agencies or projectsthereof (e.g. Deutsche Gesellschaft fur InternationaleZusammenarbeit), UN or non-UN international agencies(e.g. African, Caribbean, and Pacific Group of Statessecretary, Coalition of Rainforest Nations, CaribbeanCommunity Climate Change Centre, United NationsDevelopment Programme (UNDP), United Nations En-vironment Programme (UNEP), national offices, etc.),domestic and international NGOs, also those that maybe acting as advisors to the government, if mentioningthe name of the NGO. Also includes youth representa-tives.

144

Additional categories used in Figure 5 (which may denote specific interests, but canoverlap with the previous ones)

Climate change, meteorol-ogy, vulnerability

Count of ‘national communication’, ‘snc’, ‘focal point’,‘point focal’, ‘punto ’, ‘clima’, ‘meteor’, ‘meteo’, ‘adapt’,‘vulnerab’, ‘disaster’, and ‘desastre’ within the dele-gates’ affiliations.

CDM carbon markets Count of ‘carbon’, ‘mechanism’, and ‘mecanismo’ withinthe delegates’ affiliations.

Energy Count of ‘energy’ within the delegates affiliations.Aviation, maritime, trans-port

Count of ‘avia’, ‘maritim’, and ‘transport’ within thedelegates’ affiliations.

Forestry Count of ‘forest’ and ‘bosque’ within the delegates’ af-filiations.

Note for both tables in the Appendix: Delegates serving security, protocol, or logistic purposes,

from media, university, research institutions, or without clear affiliation, were not included in

the analysis.

145

Conclusion

This dissertation set out to explore, inter alia, under which conditions a global treaty to tackle

climate change and to limit the consequences of global warming might be possible. When I

started my research about three years ago as a member of the research project ‘Negotiating

Climate Change’, funded by the Swiss Network for International Studies (SNIS), hopes that

the global warming problem could be solved were high. Just month before the Conference

of the Parties (COP) 15 in Copenhagen started, negotiators, pundits, researchers, as well

as the wider public had high expectations for what was then considered to be the biggest

diplomatic event in history (Dimitrov, 2010, p.795).1 In the aftermath of the disappointing

Copenhagen conference there was a widely shared feeling that the whole process of tackling

climate change via huge international conferences had failed, which induced some researchers

to question the benefits of future negotiations rounds (see e.g. Falkner et al., 2010), or

even to write a postmortem on the climate change negotiations (Bodansky, 2010). Global

leaders, most prominently the Obama administration, paid much less attention to the issue

(at least in public) than prior to the Copenhagen summit. Also the media and the public

turned their focus elsewhere, notably to the global financial and economic crisis, which has

captured headlines and the public imagination to a much larger degree than climate change

and climate policy over the past three years.

Yet the climate change negotiations have continued, are making progress (although some-

what slowly), and the outlook of finding an agreement look less grim now than after the

failure of Copenhagen. The two latest COPs in Cancun and Durban can both be labeled

as moderately successful, and just recently Christiana Figueres, Executive Secretary of the

United Nations Framework Convention on Climate Change (UNFCCC) stated that “there

are still some tough political decisions ahead, but we now have a positive momentum and

a greater sense of convergence that will stimulate higher-level political discussions ahead

of Doha and set a faster pace of work once this year’s conference begins” (United Nations

News Center, 2012). Climate change negotiations are thus still our biggest hope to avoid

catastrophic global warming.

Against this background, the focus of this dissertation to analyze the climate change

negotiations and to find patterns in the way negotiators and states act during international

conferences is still highly relevant. Not only can such an analysis help to advance our under-

standing of how the involved parties behave during the negotiations. A better comprehension

of what drives positioning behavior, or cooperation and coordination among parties might

1 The Copenhagen Summit has since been overtaken by the United Nations Conference on SustainableDevelopment, Rio+20, which was attended by more than 150 Heads of State and over 50,000 visitors (UnitedNations Conference on Sustainable Development, 2012).

146

also advance our knowledge of how to overcome the free-rider problem underlying the cli-

mate change problem (see Introduction for an overview). Thus, a close examination of the

negotiations, as carried out by numerous researchers around the world, might contribute to

finding solutions to a common problem and eventually give rise to a treaty able to tackle

climate change. In such a global endeavor, a dissertation like the present one can of course

only play the tiniest of parts. Yet it is my hope that the four papers presented in this dis-

sertation shed some light on a few so far not very well understood aspects of the UNFCCC

negotiations, and that the findings in these papers are of some value.

I start with an examination of the results of the first paper in this dissertation, co-

authored with Stefanie Bailer, which evaluates the negotiations positions of countries prior

to the climate change meeting in Cancun. As outlined by such notable negotiation experts

as Hopmann (1996) or Hinich and Munger (1997), knowing where negotiating parties are

positioned on a policy issue is crucial to determine the bargaining space. This bargaining

space is an important indicator of how likely finding an agreement on the issue in question

will eventually be. The dependent variables of the first paper of this dissertation, ‘Annex

1 reduction targets’ and ‘Mitigation finance’, are good illustrations for how countries adopt

positions on an issue and thus form the bargaining space. In general, Annex 1 countries tend

to adopt positions on the lower end of the scale, while non-Annex 1 countries on average

demand more emission reductions by developed countries and a higher level of mitigation

finance. This division is not surprising and reflects the ‘Firewall’ established between the

two groups through the Kyoto Protocol (Bodansky, 2010, p.234). However, a more in-

depth analysis reveals that the width of the bargaining space depends on certain country

characteristics. For example, democratic status or vulnerability to climate change influence

how a country positions itself on the two issue dimensions mentioned above. The more

democratic or the more vulnerable countries are, the closer they are (on average) to each

other, and the more narrow the gap caused by the Kyoto Firewall becomes.

What these results indicate is that there is a core group of countries among which find-

ing a negotiated agreement to tackle the issue of climate change might be comparatively

easy. Thus, the exclusion of undemocratic and/or less affected countries might narrow the

bargaining space substantially. Attaining a treaty involving all the world’s nations, on the

other hand, is much more difficult. The reason is that non-environmental concerns, such

as monetary compensation for possible losses, play a more central role in some country’s

bargaining strategies, while environmental matters only play a secondary role. Saudi Arabia

is a good example for such obstructionist behavior. The country mainly seeks to be com-

pensated for potential oil revenue loss, and otherwise mostly wishes to stall the negotiations

(see Depledge, 2008).

147

Hence, the question global leaders and policy makers have to answer is whether they

want to keep negotiating under the current rules at all costs. If the answer is yes, countries

such as Saudi Arabia, but also other states unwilling to accept binding obligations such as

for example China, will have to be compensated or exempt in order to achieve a globally

accepted treaty. Therefore, an alternative option might be a coalition of those states willing

to seal a deal, and negotiations outside the traditional UN framework.2 This might very

likely lead to accusations of the left out countries and environmentally minded NGOs of

obstructing rather than advancing the negotiations. An option might therefore be to carry

on the negotiations including all countries of the world under the auspices of the UNFCCC,

while opening an additional branch of the negotiations, including only those countries willing

to cooperate. In the smaller bargaining space thus formed an agreement might be found.

Whether such an option is politically feasible, i.e. whether the UNFCCC negotiations can

be side-lined, is another question and to a large degree depends on political will and global

leadership.

The second paper, focusing on bargaining success, demonstrates that preference attain-

ment in the climate change negotiations is not (or at least not only) randomly allocated and

thus not only determined by luck (as theorized by Barry, 1980a,b). Instead, factors such

as power, the amount of greenhouse gas emissions a country emits, or the extremity of the

bargaining positions play a crucial role. This last mentioned point, extremity of positions,

feeds into the previous discussion of the bargaining space. Often countries such as Saudi

Arabia, Bolivia or Venezuela deliberately adopt relatively extreme positions. Thus, they

mainly aim at advancing an idiosyncratic objective instead of preventing climate change.3

These countries are clearly the least successful in terms of reaching negotiation outcomes

close to their stated preferences. They nevertheless might reach their (not officially stated)

goals, for example by slowing the negotiations down considerably, or by attaining the atten-

tion of the world’s media and the NGO community.4 These countries are thus obstacles on

2 To some extend this is already happening, as leaders of the G20 nations during their annual meetingsregularly discuss the issue of climate change. In 2009, at the summit in Pittsburgh, they also agreed to reducefossil fuel subsidies (Keohane and Victor, 2010, p.6). This strategy might be one way forward, although itexcludes many nations willing to participate in a future treaty.

3 As stated earlier, Saudi Arabia wants compensation and slow progress of the negotiations. Boliviaand Venezuela, on the other hand, have an anti-capitalist agenda and mainly focus on such issues as socialjustice or the climate debt of the North, while bringing forward high demands with regards to compensationpayments from Annex 1 to non-Annex 1 countries. They even held their own World People’s Conference onClimate Change and the Rights of Mother Earth in Cochabamba, Bolivia (April 2010), which amongst othersended with a proposal that developed countries should contribute 6% of their GDP annually to compensatedeveloping countries for the damage they had caused (for a full version of the ‘People’s Agreement’ seehttp://pwccc.wordpress.com/support/).

4 I personally had the chance to attend a side event organized by Bolivia where Evo Morales held aspeech mainly for the press, and to address well-wishing supports from the big NGO community. Although

148

the road to a future treaty, rather than constructive negotiation partners. Finding a way to

circumvent them might therefore be necessary to conclude the negotiations successfully, and

would have the desirable side-effect of narrowing down the bargaining space.

Besides this additional argument for negotiations in smaller groups, the results of the

paper on bargaining success also suggest that more constructive parties are not worse off

in terms of achieving their goals. Although, for example, the European Union was severely

criticized for being side-lined in Copenhagen (see e.g. Traynor, 2010), a year later in Cancun

the EU countries were able to achieve their goals comparatively well (as the Appendix 3 of

Chapter 2 shows). The EU is among the most conciliatory negotiators during the UNFCCC

meetings. Europe has pledged relatively high amounts of mitigation and adaptation funding,

and is also committed to impose emission reduction targets on the member states even if

the negotiations for a second Kyoto commitment period should fail (Bohringer et al., 2009).

Such a conciliatory course also means adopting centrist positions, a key factor for bargaining

success as shown in the paper.

That the EU’s success in Cancun was not only a one-hit wonder was demonstrated a

year later in Durban, where the community was widely praised for winning over other par-

ties to their cause (Harvey, 2011; Harvey and Vidal, 2011b). The community even managed

to convince India and China to agree to a ‘roadmap’, which will eventually lead to bind-

ing emission reduction commitments for all countries (previously strongly rejected by those

parties, see Harvey and Vidal, 2011a). It follows that parties with less conflict-oriented

bargaining strategies and a higher willingness to compromise need not necessarily be less

successful. This is confirmed by the statistical models, which show that particularly coun-

tries with high salience values can employ soft bargaining strategies in order to convince

others of their positions and consequently to succeed in the negotiations.

Forming alliances with other negotiating parties, the topic of the third paper of this dis-

sertation, is essential for achieving bargaining success from a single party’s perspective, but

also to reduce the ‘noise’ of all the diverging opinions in the negotiations to a more man-

ageable number. Coalitions and alliances are thus a key feature in multiparty negotiations

to reduce complexity and to facilitate the bargaining process (Dupont, 1994, 1996). In the

part of the dissertation covering alliances and position coordination behavior I demonstrate

that there are a host of factors that help to align countries’ interests, and consequently

cause them to coordinate their positions (although sometimes only on single issues where

preferences overlap). For example power, the democratic status, or integration into the in-

such happenings produce little of substance, apart from the usual accusation, they cause a big stir duringthe negotiations and provide countries like Bolivia (or rather Evo Morales) with the international attentionthey seek.

149

ternational systems are characteristics that help countries to identify allies. In addition,

this part of my work shows that the already established official negotiation groups are good

predictors of who is susceptible to coordinate with whom. Countries who are members of

the same coalition show a more than 170% higher propensity to coordinate their positions

than two randomly chosen countries.5

The idea that the complexity of multiparty negotiations needs a way to be managed,

highlighted by Zartman (1994), was given credit by establishing formal negotiation groups.

The results of my research show that the countries pooled together in these groups indeed

to have common interests, and that formalizing a coalition framework has its merits. By

creating the coalition groups the UNFCCC has taken an important step to reduce the number

of players6, and as a consequence the complexity of the negotiations. Allowing countries to

declare common positions via joint statements goes one step further and helps parties to

unite on single issues, as my research shows. Not only does this reduce complexity, it also

might help to build trust among otherwise rather diverse parties. For example, Switzerland

and Tuvalu are two rather dissimilar countries in many aspects, yet they did issue joint

statements during the climate change negotiations twice. The results of the exponential

random graph models (ERGMs) suggest that such behavior is driven by, in this example,

similarities in vulnerability to climate change impacts (both countries are highly sensitive

to climate change, although the implications of a changing climate are very different for the

two countries). Similarly, fellow democracies find it easier to coordinate their positions, as

do countries who know each other well from cooperating repeatedly in different international

organizations. Such more informal bonding via joint statements is a relatively simple way

for countries to form alliances and further reduces complexity. Encouraging countries to

cooperate in this fashion might therefore be a good strategy for the UNFCCC secretariat to

further ease the bargaining process.

Finally, the paper on the bargaining behavior of the Alliance of Small Island States (AO-

SIS) negotiation group sheds light on the inner workings of the already mentioned coalitions.

AOSIS is a particularly interesting example, as this group of politically rather insignificant

countries achieved remarkable success especially in the earlier years of the negotiations. In

cooperation with my colleagues Paula Castro and Carola Betzold, we could demonstrate that

a high level of cohesion among the group members was responsible for this success story.

5 Random dyads have a probability to form ties of around 5%, hence two countries from the same coalitiongroup exhibit slightly less than 14% likelihood to coordinate their positions.

6 When countries decide to team up and adopt the same position they can essentially be seen as a singleparty (at least regarding the issue in question). The most prominent example is the EU, whose memberstates have decided to issue statements and positions only in accordance with each other, and have thereforebeen accepted as negotiating as a single entity by the UNFCCC.

150

In later years, when first cracks among member states appeared and solidarity started to

erode, the level of AOSIS’ success began to dwindle to some degree. However, as the group

remains highly uniform and still manages to attain a relatively high level of unity compared

to other groups, the alliance is likely to continue to be a key player in the climate change

negotiations. Therefore, AOSIS should serve as an example for other more divided groups of

how to settle internal disputes and to ensure unity, an in particular for other small countries

of how to maximize success.

In addition, the willingness to compromise of these highly climate sensitive island states

serves as an important model for many other more egoistic states, for whom much less is

at stake. This must also be the final lesson drawn from this dissertation. There are ways

to structure the negotiations in a way to facilitate finding common ground, or lessons can

be learned by individual countries how to behave in order to maximize success. However, if

global warming is to be avoided, a substantial number of countries (ideally all) must unite

with the common goal to solve the problem in mind, and (at least partly) brush aside their

individual interests and differences. Game theorist are often skeptical that such ‘benign’

behavior can be fostered (Cline, 1992; Nordhaus and Boyer, 2000). In contrast, I believe

cooperation is possible if mistakes of the past are avoided, and the political will exists.

What can be learned from my work is that some little adjustment to the structure of the

negotiations might help to pave the way to a negotiated agreement. The main conclusion

of my work is therefore that positioning behavior, striving for success in the negotiations,

and coalition building, are all crucial elements which structure the negotiations and help

countries to signal where they themselves stand, and how far apart they are from others. All

these elements, if properly understood, can help to foster a negotiated agreement after a long

(and often painful) bargaining process. Yet a willingness to cooperate and to compromise is

a pre-condition for ending the negotiations with a treaty.

This last point, the willingness to find a compromise, is a problematic issue for the

negotiations, but also from the perspective of my work. About three years ago, when I

started working on this dissertation, I took the willingness of the negotiating parties to find

solutions in the common interests for granted. A major outcome in Copenhagen was expected

against which I would compare positions of bargaining parties, and thus assess their success.

However, no major deal came forward in Copenhagen, and the whole endeavor of obtaining

success scores was in limbo. Also the Cancun agreements, which I eventually employed as

the benchmark for a country’s success, are only an intermediate outcome of the negotiation

process. An evaluation after Durban, or Doha, or maybe a future negotiation round which

delivers a breakthrough, might find entirely different success scores and regression results

than I did in this thesis. Investigating bargaining success in the climate change negotiations

151

(as well as in other negotiation settings) remains therefore a highly interesting topic for

future research.

The same is true for bargaining positions. There are a number of good reasons why to

rely on expert interviews, as our research team did. But there are also good reasons to

employ different techniques, as Genovese (2012) correctly pointed out in a paper discussing

our data generating effort in detail. On the one hand, we only were able to collect positions

for a limited number of countries, and it is unclear to what extend our data really represent

a random sample.7 On the other hand, and in my view more seriously, we faced the problem

of measurement error. During the many interviews I conducted personally, I repeatedly

had doubts that the delegate was able to provide well informed answers. Some interviewees

were excellent, e.g. we had the chance to interview the delegation leaders of Indonesia and

Bangladesh. Yet in some cases we had to content ourselves with interviewing much less

experienced negotiators, or delegates who were experts in one field of the negotiations only.

All this might distort the results presented in this thesis. Therefore, I welcome the effort of

Genovese and colleagues to measure the bargaining positions of states via content analysis

of relevant documents. More work in this direction is needed.

Personally, I also consider content analysis as carried out by Genovese problematic, as I

already made sufficiently clear in the introduction of this dissertation (crucially, the chosen

documents in my view are inadequate to capture negotiation positions). The pros and cons of

the data this dissertation is based upon are also discussed in more detail in the introduction.

Although new approaches such as advocated by Genovese are a welcome addition, I believe

that the preceding chapters have shown that more traditional methods are also able to

provide valuable insights into how negotiations work. I however highly recommend data

triangulation to enhance the reliability of the data, as carried out for the project ‘Negotiating

Climate Change’.

Having studied the negotiations for three years, I believe a good way forward to study

negotiation positions would be to investigate only a subset of countries, and only one or two

variables driving positioning behavior. This is what I plan for my future research. More

specifically, I intend to analyze the mechanisms by which the public opinion in democracies

effects a country’s bargaining positions. On the one hand my plan is to conduct surveys to

check to what degree positions are in line with public opinion, but also to investigate which

groups of voters have the greatest influence over positioning behavior. On the other hand

I intend to employ times series approaches to follow public opinion over time, and to check

whether and how positions react to changes in public preferences. To conclude, I believe

that I personally learned a lot during my work on this dissertation, and I hope to be able

7 Yet this shortcoming can relatively easily be adjusted for, employing Heckman Correction Models.

152

to make valuable contributions to the field of climate change and climate policy research in

the future.

153

References

Barry, B. (1980a). Is it better to be powerful or lucky? Part I. Political Studies 28 (2),183–194.

Barry, B. (1980b). Is it better to be powerful or lucky? Part II. Political Studies 28 (3),338–352.

Bodansky, D. (2010). The Copenhagen climate change conference - A postmortem. AmericanJournal of International Law 104 (2), 230–240.

Bohringer, C., A. Loschel, U. Moslener, and T. Rutherford (2009). EU climate policy up to2020: An economic impact assessment. Energy Economics 31 (Supplement 2), S295–S305.

Cline, W. (1992). The Economics of Global Warming. Washington, D.C.: Institute forInternational Economics.

Depledge, J. (2008). Striving for No: Saudi Arabia in the climate change regime. GlobalEnvironmental Politics 8 (4), 9–35.

Dimitrov, R. (2010). Inside UN climate change negotiations: The Copenhagen conference.Review of Policy Research 27 (6), 795–821.

Dupont, C. (1994). Coalition theory: Using power to build cooperation. In I. W. Zartman(Ed.), International Multilateral Negotiations. Approaches to the Management of Com-plexity, pp. 148–177. San Francisco: Jossey-Bass Publishers.

Dupont, C. (1996). Negotiation as coalition building. International Negotiation 1 (1), 47–64.

Falkner, R., H. Stephan, and J. Vogler (2010). International climate policy after Copenhagen:Towards a ’building blocks’ approach. Global policy 1 (3), 252–262.

Genovese, F. (2012). States’ interest and bargaining positions at the UN climate changenegotiations. Paper presented at the EPSA 2nd General Conference, Berlin, June 2012.

Harvey, F. (December 2011). Durban talks: How Connie Hede-gaard got countries to agree on climate deal. The Guardian.http://www.guardian.co.uk/environment/2011/dec/11/connie-hedegaard-durban-climate-talks.

Harvey, F. and J. Vidal (December 2011a). Climate deal salvaged after marathon talksin Durban. The Guardian. http://www.guardian.co.uk/environment/2011/dec/10/un-climate-change-summit-durban.

Harvey, F. and J. Vidal (December 2011b). Durban climate talks see US back EU pro-posal. The Guardian. http://www.guardian.co.uk/environment/2011/dec/08/durban-climate-talks-us-backs-europe.

Hinich, M. and M. Munger (1997). Analytical Politics. Cambridge, New York, Melbourne:Cambridge University Press.

154

Hopmann, P. T. (1996). The Negotiation Process and the Resolution of International Con-flicts. Columbia: University of South Carolina Press.

Keohane, R. and D. Victor (2010). The regime complex for climate change. The HarvardProject on International Climate Agreements. Discussion Paper 10-33 .

Nordhaus, W. and J. Boyer (2000). Warming the world: economic models of global warming.Cambridge, MA: MIT Press.

Traynor, I. (February 2010). Europe loses seat at top table. The Guardian.http://www.guardian.co.uk/world/2010/feb/08/european-parliament-crisis.

United Nations Conference on Sustainable Development (2012). Rio+20may be largest event in the history of the United Nations.http://www.uncsd2012.org/index.php?page=viewnr=723type=230menu=38.

United Nations News Center (2012). Ahead of Doha gath-ering, concrete progress made at Bangkok climate talks.http://www.un.org/apps/news/story.asp?newsid=42810cr=cr1=.uehzhjhyeyi.

Zartman, I. W. (1994). Introduction: Twos company and moresa crowd: the complexitiesof multilateral negotiation. In I. W. Zartman (Ed.), International Multilateral Negotia-tions. Approaches to the Management of Complexity, pp. 1–12. San Francisco: Jossey-BassPublishers.

155