Content uploaded by Detlef F. Sprinz
Author content
All content in this area was uploaded by Detlef F. Sprinz on Nov 11, 2015
Content may be subject to copyright.
10 Nov. 2015
Predicting Paris:
Forecasting the Outcomes of UNFCCC COP-21
With the Predictioneer’s Game
Detlef F. Sprinz
Bruce Bueno de Mesquita
PIK – Potsdam Institute for Climate
Impact Research
Department of Politics
New York University
dsp@pik-potsdam.de
bruce.buenodemesquita@nyu.edu
Abstract
An international agreement at Paris in the end of 2015 or (shortly) thereafter
may indeed be feasible, yet some issues are predicted to take considerably
longer than the end of 2015. More specifically, our forecasts suggest:
• The general principle of CBDR will prevail.
• Compared to the Kyoto Protocol, a weakened compliance system will
be agreed upon.
• A legally binding agreement on mitigation is likely to come forward,
yet country-specific mitigations targets are likely not to be legally
binding.
• More than a collective non-binding commitment to adaptation is likely
to come forward plus a strengthening of present institutions for
adaptation.
• Financial commitments will be upgraded beyond $ 100b p.a. after
2020, and emerging economies will be urged to contribute.
• Loss & damage will likely take much longer to negotiate than the end
of 2015, and some separate agreement with new institutional
arrangements is likely.
• The “no backsliding” principle will be applied to future negotiations on
mitigation ambitions.
• The Ex-Ante Assessments will concentrate on aggregate ambitions,
but not a full assessment of individual ambitions.
Predicting Paris Page | 2
High expectations were raised for the international climate negotiations in
2009 – with the expectation that a successor agreement to the 1997 Kyoto
Protocol under the 1992 UN Framework Convention on Climate Change
(UNFCCC) would be finalized. These expectations were disappointed and led
to a second preparatory phase for an international climate agreement with
universal participation. These negotiations are expected to lead to an
international agreement for the period beginning 2020 by the end of 2015 at
Paris in order to guide national, international, and transnational policies.
Are such expectations for an international climate agreement by mid-
December 2015 at UNFCCC Conference of the Parties 21 (COP-21) well
founded? More specifically, what is the likely contents of such an
agreement as negotiators around the world face choices on a wide range of
issues and major decisions still have to be taken?
Together with our colleagues from CICERO – Center for International
Climate and Environmental Research Oslo and the University of Oslo,1 the
University of Groningen and the University of Strathclyde,2 we derived the
most pertinent issues under debate from the first negotiating text shared by
the UNFCCC Secretariat on 25 February 2015 (United Nations Framework
Convention on Climate Change 2015). Subsequently, we embarked on two
research tracks. First, the CICERO team undertook an online expert
elicitation of the likely outcomes at Paris (Kallbekken and Sælen 2015).
Second, two different sets of predictions were undertaken on comparable
issues, namely with the DECIDE model (Stokman and Thomson 2015) and
the Predictioneer’s Game. In this report, we document the forecasts on 11
issues undertaken with the Predictioneer’s Game (Bueno de Mesquita 2009,
Bueno de Mesquita 2011) with input data based on expert knowledge.
In the Predictioneer’s Game context, an “issue” is any specific policy
question for which Stakeholders have differing preferences regarding the
outcome. It is important that issues are defined such that the range of
1 We thank Guri Bang, Jon Hovi, Steffen Kallbekken, Håkon Sælen, and Arild Underdal.
2 We thank Frans N. Stokman and Robert Thomson.
Predicting Paris Page | 3
potential outcomes forms a single, continuous scale for all stakeholders. The
farther an option is from a stakeholder's stated position (in either direction),
the less preferred the option is. If, on an issue scale, position “C” is to the
right of position “B,” which is in turn to the right of position “A,” then all
stakeholders agree with this linear order and a stakeholder who prefers "A"
to "B" will also prefer "A" to "C" and "B" to "C."
The 11 issues assessed can be grouped into five clusters (see Appendix 1 for
details):
• overall differentiation principle,
• mitigation,
• adaptation,
• climate finance and loss & damage, as well as
• ambitions and review.
As stakeholders (actors), we chose the major countries and country groups
participating in the global climate negotiations within the UNFCCC. The
expert input data for the policy predictions (not to be confused with the
expert elicitation of the likely outcomes of the Paris negotiations,
documented, e.g., in Kallbekken and Sælen (2015) focuses on 16 groups of
countries, including major countries by themselves. The EU28 was treated
as one group. In particular, we included
• the African Group,
• AILAC - Association of Independent Latin American and Caribbean
States,
• ALBA - Bolivarian Alliance for the Peoples of Our America,
• AOSIS – Alliance of Small Islands States,
• the Arab Countries,
• Bangladesh (if no data were available, data were replaced with sources
from the group of LDCs – Least Developed Countries),
• Brazil,
Predicting Paris Page | 4
• China,
• EIG – Environmental Integrity Group,
• the EU28,
• India,
• Indonesia,
• Umbrella Group (minus Japan, Russia, and the USA),
• Japan,
• Russia, and the
• USA.
As the Predictioneer’s Game is normally not used for country groups but for
individual actors, we treat the prediction to follow as a test/feasibility study
of the Predictioneer’s Game in the context of a mixture of major countries
and groups of countries.
For each of the actors, the following inputs were generated (see Appendix 2
for details):
• stated position (outcome currently advocated by a stakeholder),
• salience (commitment the stakeholder has in pursuing this issue over
all other issues and topics competing for attention),
• potential influence (bargaining clout of each stakeholder, relative to
each other), and
• flexibility (stakeholder’s preference for reaching an agreement as
compared to sticking to his or her preferred position even if it means
failing to reach an agreement).
The forecasts to follow assume that no pertinent actor has been omitted and
that domestic and transnational actors exert their influence by way of the
stakeholders listed above. Furthermore, it is assumed that no trade-offs are
made across issues (see above), i.e., each issue is negotiated on its own
merits. In most cases, data were available for most actors. In case data
were missing for a range of actors, forecasts were not undertaken (see
Predicting Paris Page | 5
below). If only one actor had missing data – which was often the case for
Indonesia – predictions were undertaken while omitting this stakeholder.
1. Overall Differentiating Principle
Perhaps the most challenging question in a future international climate
agreement pertains to the equity principle used. Until recently, the division
of countries into developing and developed countries (as enshrined in the
Kyoto Protocol) held sway, yet China has been for nearly a decade the
largest CO2 emitter worldwide, and India has taken the third position in
yearly emissions. Thus, the question arose whether a division in Annex 1
(developed) and non-Annex 1 (developing) countries remains the overarching
principle in a future agreement.
Issue: What will be the dominant basis for effort-sharing in the new
agreement?
Scale range:
0: no explicit differentiation (self-differentiation)
100: Annexes I and II of the Convention (i.e., dichotomization into developing/ developed countries)
Predicting Paris Page | 6
Our simulations suggest that the matter will be resolved after 4 rounds of
negotiations with a predicted outcome of 58 (on a scale ranging from 0 to
100) and a 95% confidence interval of [50, 66].3 The predicted outcome is
clearly different from a dichotomous division of the negotiating parties into
developed and developing countries. Our results suggest that the outcome
will be close to the principle of Common but Differentiated Responsibilities
and Respective Capabilities (CBDR) in light of national circumstances, yet
not purely self-selection of ambitions (scale point 0). Focusing on the
relative narrow 95% confidence interval of [50, 66], the CBDR principle is
very likely to be mentioned, combined with some augmentation. The Arab
countries and Russia are expected to be particularly hesitant about the
outcome (as they prefer more individual differentiation), whereas ALBA and
India would be substantially happier with results closer to maintaining the
division of obligations according to developing vs. developed country status.
Many countries are expected to make compromises rather quickly, and the
equilibrium result may conceivably be reached by December 2015 or shortly
thereafter in 2016.
In case we optimize for the position of the EU28 and include countries
within ±10 points on the scale of the EU28’s position, only minor gains are
achievable for the EU28 and similarly-positioned countries (results not
shown here).
2. Mitigation
2.1 Mitigation: MRV and Compliance
Issue: What should be the minimum MRV (measuring, reporting, and
verification) and compliance provisions in the agreement on mitigation?
3 We will hence report the smoothed mean (which is the average of weighted means from the
previous, current, and future round) as well as the 95% confidence interval, e.g., 58 [50,
66].
Predicting Paris Page | 7
Scale range:
0 International Consultation and Analysis (ICA)
100 Kyoto complianc e regime
The Predictioneer’s Game suggests that the game ends in round 1 or 3 with
a score of 50 [30, 68], (i.e., an International Consultation and Analysis
combined with a multilateral consultative process). An International
Assessment and Review with a committee on implementation and/or
compliance lies outside the 95% confidence interval. If the equilibrium is
reached in round 3, many countries favoring adoption of the Kyoto
compliance regime make considerable concessions during these three
rounds.
2.2 Mitigation: Legal Form
The legal form of an international agreement has been a major issue during
the negotiations, in particular whether an internationally binding agreement
with strict emission reduction goals is feasible. Witnessing the hurdles of
ratifying the 1997 Kyoto Protocol in the USA as well as Chinese positioning
with respect to taking over internationally binding obligations, this point is
contentious. The Predictioneer’s Game indicates high levels of tension in
Predicting Paris Page | 8
stakeholder interactions through the first four or five rounds, contributing
to the pressure to break-off the discussion.
Issue: To what extent should the agreement and its components relating to
mitigation targets be (internationally) legally binding?
Scale range:
0 No binding agreement or binding country-specific targets
100 Binding agreement plus binding, country-specific targets.
The Predictioneer’s Game predicts that the game ends in round 4 at scale
point 53 [44, 63]. This equilibrium solution implies an internationally
binding agreement on mitigation, yet internationally non-binding country-
specific targets. This leaves enforcement of national obligations to national
institutions, not international ones for specific mitigation obligations. The
95% confidence interval does not include a binding agreement with non-
binding country-specific targets, plus MRV obligations. The Arab Countries,
Bangladesh, and the EU28 will be among the least satisfied countries.
Predicting Paris Page | 9
3. Adaptation
3.1 Adaptation: Legal Framework
Adaptation has not played a major role in the UNFCCC and its 1997 Kyoto
protocol, yet the issue of adaptation is at the heart of many developing
countries.
Issue: To what extent should countries’ commitments to adaptation targets be
country-specific and legally binding?
Scale range:
0 No new commitments to adaptation
100 Legally binding country-specific commitments
For the legal framework of adaptation, the Predictioneer’s Game was run for
24 negotiation rounds and predicts that the game will either end in round 1
at scale point 60 [48, 73] or negotiators will push through initial resistance
and carry on discussions. In the latter case, the trend is downward on the
scale and resolution is not reached. If the initial pressure to stop discussing
the issue succeeds, then this implies that we should expect that more than
a collective, non-binding provision is forthcoming; most likely some
countries will offer non-binding commitments. Given the predicted results,
Predicting Paris Page | 10
Russia and the USA should be particularly displeased which also holds for
most developing country groups, excl. the governments of Brazil, China, and
India. If, however, negotiations continue beyond round 1, there will be
prolonged, indecisive debates and discussion.
3.2 Adaptation Institutions
Issue: To what extent should the institutional framework for adaptation be
strengthened?
Scale range:
0 No strengthening
100 Esta blish subsidiary body on adaptation
For the creation of institutions for adaptation, the Predictioneer’s Game
suggests rather fast negotiations ending in round 1 at position scale 67 [54,
80]. This point prediction implies a strengthening of present institutions
(stronger mandate, funding, and knowledge platform), with at least some
strengthening and a maximum of creating new institutions. Should,
however, negotiations drag on until round 10 (substantially later than
December 2015), some of the fervor in strengthening institutions for
adaptation will be lost.
Predicting Paris Page | 11
3. Climate Finance and Loss & Damage
Financial issues are potentially some of the most challenging ones to resolve
as they may involve either quantifiable monetary contributions for
international redistribution or compensation for damages originating from
climate change.
3.1 Climate Finance - Volume
Developed countries have promised to make available $100 b per annum
effective 2020 to enable developing countries to contribute to international
climate policies. Given present challenges of substantiating commitments of
private and public funds on the order of magnitude envisioned, will the
amount be upgraded in the future or remain at $ 100b p.a.?
Issue: What will the size of agreed volume to be mobilized (private and public)
beginning 2020 (p.a.)?
Scale range:
0 no new target (i.e. $100b p.a.)
100 ≥$500 b (in excess of 1% of present OECD GDP p.a.)
Predicting Paris Page | 12
The Predictioneer’s Game predicts that the game will end in round 3 at scale
point 55 [35, 75]. The smoothed mean implies a considerable upgrading
beyond the $ 100 b p.a. from public and private sources offered so far.4 The
95% confidence interval is comparatively wide (see above), including
amounts in the range of ca. $ 175b - $ 375b p.a.
All industrialized countries stay below the 95% confidence interval. As the
amounts implied pose a major challenge, it would not be far-fetched that a
range of those countries which are expected to provide major contributions
may ultimately use a veto. Indeed, if the discussions are pushed past the
third round, the trend becomes sharply negative. That, in fact, is why there
is strong pressure to end debate in the third round, avoiding the cascade
away from commitment.
3.2 Climate Finance - Who Pays?
Given the hopes by some to mobilize major financial contributions in the
context of a global climate agreement, the question arises, who shall make
these financial contributions.5 Our experts suggest that all stakeholders are
either positioned at 0 (only developed countries are obliged to contribute) or
60 (developed countries and certain other countries required to contribute
(e.g. “countries in a position to do so” or emerging economies)). Will
emerging economies be asked to contribute to financing the global response
to climate change at the international level?
4 We predict promises to be made, not the degree of their implementation.
5 The predictions undertaken under 3.2 here are assumed to be unrelated to the outcomes
predicted for 3.1.
Predicting Paris Page | 13
Issue: Who will be requested to pay for climate finance?
Scale range:
0: developed countries only required to contribute
100: all countries required to contribute
On the question of “who pays,” the Predictioneer’s Game suggests that the
negotiations may stop in rounds 1 or 5. For round 1, the prediction is scale
point 27 [12, 43], i.e., developed countries are required to contribute, while
developing countries are demanded a bit more than voluntary contributions.
If the equilibrium is reached in round 1, this will be in time for December
2015, otherwise, negotiations may well run into 2016. Focusing on the 95%
confidence interval, at maximum, select developing countries will be strongly
invited to make contributions. Many countries make only relatively small
concessions - which explains the wide confidence interval if negotiations end
in round 1. If they do end in round 1, the substantive findings are not very
different with a smoothed mean at 25 [13, 38], however, the 95% confidence
interval begins to narrow. ALBA, the Arab countries, Brazil, China, India,
and Russia are forecast to remain particularly unhappy with the outcome.
What will happen if we optimize for the negotiation position for the EU 28
only as well as everyone else positioned at scale point 60 (i.e., AILAC, AOSIS,
Bangladesh, EIG, EU28, and the Umbrella Group - except for Russia)?
Predicting Paris Page | 14
Scale range:
0: developed countries only required to contribute
100: all countries required to contribute
Even with optimization for the EU28 only or for all countries positioned at
scale point 60, the game’s length and substantive results are essentially the
same as for the base run without optimization.
Predicting Paris Page | 15
3.3 Adaptation Reserved Financing
Issue: To what extent should funds be reserved for adaptation?
Scale:
0: no earmarking for adaptation
50: approximately 50% earmarked for adaptation
100: dedicate d levy for adaptation
Until now, adaptation is without a dedicated finance stream for itself. The
Predictioneer’s Game suggests that the negotiations will end in round 2 at
scale point 66 [48, 83] in an effort to avoid subsequent erosion of an
agreement. This smoothed mean implies that somewhat more than 50% of
finances be dedicated to adaptation. The Umbrella Group, esp. Japan,
Russia, and the USA, will be particularly displeased as they show little
flexibility on the issue. Negotiations continue beyond round 2, the
Predictioneers’ Game indicates a downward pressure on financial
commitments for adaptation.
3.4 Loss & Damage
While adaptation is a well-established aspect of climate policy, the issue of
loss & damage (L&D) is comparatively new and unchartered. In the very
Predicting Paris Page | 16
long run, it is likely to deal with issues surrounding compensation for
climate impacts not avoided or reduced by either mitigation or adaptation.
Issue: To which degree will loss & damage (L&D) be included in an
agreement?
Scale range:
0: no mention/omission of L&D
100: separate chapter on L&D and new non-financial and financial elements, including a compensation regime
The Predictioneer’s Game predicts that the negotiations end in round 9 at
scale point 45 [37,53], i.e., the equilibrium is reached well after December
2015, perhaps running into 2016/17. Compared to the status quo (20), the
establishment of the Warsaw International Mechanism, the smoothed mean
at 45 indicates that we should expect at least a separate chapter on L&D
with new institutional arrangements as part of an international agreement,
yet with little additional obligations. Should a global climate agreement
without L&D inclusion be finalized by the end of 2015, our forecasts could
translate into a separate international agreement with new institutional
arrangements that elevates the issue of L&D, yet does not add more than a
few non-financial elements (such as coordination and capacity building).
The 95% confidence interval clearly indicates departure from the status quo
(20) and some probability that non-financial elements might be included.
Predicting Paris Page | 17
Many developing countries (ALBA, AILAC, AOSIS, Bangladesh and India)
quickly make concessions and move to the predicted outcome, with the
exception of China and Brazil (which are originally positioned reasonably
close to the ultimate outcome). The EU moves into the direction of the
consensus, yet Japan, Russia, and the USA hold positions in round 9
substantially different from the consensus. A comprehensive compensation
regime does not appear to be within the scope of the present negotiations.
How do the results change if we optimize for the EU28 (opening position: 20)
and countries within a ±20 scale point interval around the EU28’s position?
Scale range:
0: no mention/omission of L&D
100: separate chapter on L&D and new non-financial and financial elements, including a compensation regime
Optimizing for the EU28 leads, inter alias, to two end points in round 3 and
round 10, both of which are similar to the results of the base run – except
for the length of negotiations. If the negotiations are optimized for the EU28
and actors ±20 scale point interval (essentially all industrialized countries
plus the Arab countries), a substantial improvement of 10 scale points over
the base run is reached if negotiations stop in Round 10 as compared to the
Base Run or a deal is accepted in Round 3. Thus, optimization for a
coalition surrounding the EU28’s position on L&D yields global results
Predicting Paris Page | 18
substantially closer to the ideal point of the EU28 and roughly within the
scope of COP-21.
Conversely, what would happen if we optimized for the position of India and
±10 scale point interval around India’s position?
Scale range:
0: no mention/omission of L&D
100: separate chapter on L&D and new non-financial and financial elements, including a compensation regime
As in the case for optimization for the EU28 only, an optimization for India
(opening position: 90) ends in round 3 with barely a difference in the
equilibrium outcome as compared to the base run. If optimized for India ±
10 scale point interval (comprising ALBA, AOSIS, India, and Indonesia), this
results in an equilibrium in round 3 that is three scale points closer to the
position of India, yet in the opposite direction as compared to the EU28
optimization.
Whenever optimization for one of the actors or groups of actors is
undertaken, the models equilibrate substantially faster as compared to the
base run and lead to agreement until Dec. 2015 or shortly thereafter. In a
non-optimized setting, agreement will take place considerably later.
Predicting Paris Page | 19
4. Ambition and Review
Ambition levels come in three flavors, namely (i) an overarching principle, (ii)
specific collective ambitions for 2050 and 2100, and (iii) the ex-ante
assessment (EAA) of future (Independent) Nationally Determined
Contributions ((I)NDCs). As we did not have sufficient input data for
collective ambitions for issue (ii) across country groups, we abstained from
predictions on these issues and concentrate on the first and third issue.
4.1 Ambition Level – Mitigation Mechanism
Global climate negotiations will not stop at Paris in the end of 2015, even if
an agreement is reached. Thus, the question arises how the global climate
regime is supposed to develop in the future. In particular, shall the parties
of the UNFCCC enshrine a uni-directional principle as a yardstick for
directions beyond an initial post-2020 agreement?
Issue: What should be the mechanism for strengthening commitments over
time?
Scale range:
0: no ambition mechanism
100: binding commitment to strengthen targets in line with the 2 degrees goal
Predicting Paris Page | 20
The Predictioneer’s Game suggests that negotiations end with a smoothed
mean of 35 [25, 45] in round 12, i.e., most likely well into 2017.
Substantively, the equilibrium implies adoption of the “no backsliding”
principle, i.e., actors cannot adopt positions in the future that are less
onerous than those taken over now. At the upper end of the 95% confidence
interval, a non-binding progression principle may be adopted.
Many smaller developing countries are predicted to make considerable
concessions to their high ambitions. Many industrialized countries would
be willing to go further in strengthening progression, yet are held back by
ALBA, China, and India. While the 2°C target has been accepted at and
after Copenhagen (2009), there is little chance of adopting a legally binding
commitment to strengthen mitigation targets in line with the 2°C goal.
4.2 Ex-Ante Assessment (EAA) of future (I)NDCs
Besides a general principle on how to develop future country positions, these
positions may (or not) merit an ex-ante assessment system.
Predicting Paris Page | 21
Issue: Which provisions for assessment and review of the nationally
determined contributions will be included in an agreement?
Scale range:
0: no EAA
100: EAA of aggregate ambition and technical EAA of individual INDCs (transparency, clarity, comparability,
etc.), plus a political assessment of individual INDCs (ambition and equity/fairness), plus a formal
mechanism for involving inputs from civil society
On this issue, the Predictioneer’s Game predicts a smoothed mean outcome
at scale point 47 [24, 70], i.e., beyond an ex-ante assessment (EEA) of
aggregate ambitions, but not yet a full technical EEA of individual positions.
The (wide) 95% confidence interval includes departure beyond the EEA of
aggregate ambitions and includes slightly higher ambitions than a technical
EEA of individual INDCs, on the one side, and essentially an EEA of
aggregate ambitions only, on the other side of the interval. Many countries
at the upper and lower ends of the scale (0, 100) are likely to be displeased.
5. Summary
An international agreement at Paris in the end of 2015 or (shortly) thereafter
may indeed be feasible, yet some issues are predicted to take considerably
longer than the end of 2015. More specifically, our forecasts suggest:
Predicting Paris Page | 22
• The general principle of CBDR will prevail.
• Compared to the Kyoto Protocol, a weakened compliance system will
be agreed upon.
• A legally binding agreement on mitigation is likely to come forward,
yet country-specific mitigations targets are likely not to be legally
binding.
• More than a collective non-binding commitment to adaptation is likely
to come forward plus a strengthening of present institutions for
adaptation.
• Financial commitments will be upgraded beyond $ 100b p.a. after
2020, and emerging economies will be urged to contribute.
• Loss & damage will likely take much longer to negotiate than the end
of 2015, and some separate agreement with new institutional
arrangements is likely.
• The “no backsliding” principle will be applied to future negotiations on
mitigation ambitions.
• The Ex-Ante Assessments will concentrate on aggregate ambitions,
but not a full assessment of individual ambitions.
Predicting Paris Page | 23
Appendix 1: Issues and Scales
1. Differentiation
What will be the dominant basis for effort-sharing in the new agreement?
0: No explicit differentiation (self-differentiation)
25: National circumstances
50: CBDR-Respective Capabilities in light of national circumstances
75: CBDR-Respective Capabilities (with no direct reference to the
Convention’s Annexes or Articles referring to those Annexes)
100: Annexes I and II of the Convention
2. Mitigation MRV and compliance
What should be the minimum MRV and compliance provisions in the
agreement on mitigation?
0: International Consultation and Analysis (ICA)
45: ICA plus multilateral consultative process
65: International Assessment and Review (IAR)
75: IAR plus committee on implementation and/or compliance
100: Kyoto compliance regime
3. Mitigation legal form
To what extent should the agreement and its components relating to mitigation
targets be(internationally) legally binding?
0: No binding agreement or binding country-specific targets
30: Binding agreement without country-specific targets
50: Binding agreement plus obligation to have a (nonbinding) country-
specific target (NDC)
70: The above plus obligation on measuring, reporting and verification
100: Binding agreement plus binding, country-specific targets
4. Adaptation legal framework
To what extent should countries’ commitments to adaptation targets be
country-specific and legally binding?
0: No new commitments to adaptation
40: Collective, non-binding provisions. E.g. “all parties are encouraged to
integrate adaptation into their national plans”
80: Non-binding country-specific commitments
100: Legally binding country-specific commitments
Predicting Paris Page | 24
5. Adaptation institutions
To what extent should the institutional framework for adaptation be
strengthened?
0: No strengthening
60: Strengthen present institutions (stronger mandate, funding and
knowledge platform)
80: Establish new institutions stronger than present ones
100: Establish subsidiary body on adaptation
6. Climate Finance (volume)
What will the size of agreed volume to be mobilized (private and public)
beginning 2020 (p.a.)?
0: no new target (i.e. $100b p.a.)
20: start at $100b p.a. and be regularly scaled up
40: $ 200b p.a.
60: $ 300b p.a.
80: $ 400b p.a.
100: ≥$500 b (in excess of 1% of present OECD GDP p.a.)
6 bis. Climate Finance (who pays?)
Who will be requested to pay for climate finance?
0: Developed countries only required to contribute
20: Developed countries required to contribute, and other countries invited
to contribute voluntarily
60: Developed countries and certain other countries required to contribute
(e.g. “countries in a position to do so” or emerging economies)
80: All countries minus LDCs and SIDS required to contribute
100: All countries required to contribute
7. Adaptation reserved financing
To what extent should funds be reserved for adaptation?
0: No earmarking for adaptation
50: Approximately 50% earmarked for adaptation
100: Dedicated levy for adaptation
8. Loss and Damage
To which degree will loss & damage (L&D) be included in an agreement?
0: No mention/omission of L&D
10: Preambular reference only
20: Reference to Warsaw International Mechanism (WIM) (under adaptation)
30: Separate chapter on L&D with little substance
40: Separate chapter on L&D and new institutional arrangements with little
substance
50: Separate chapter on L&D and new institutional arrangements with new
non-financial elements (such as coordination and capacity-building)
70: Separate chapter on L&D and new mechanism with new non-financial
and financial elements (such as insurance) but no compensation regime
100: Separate chapter on L&D and new non-financial and financial
elements, including a compensation regime
Predicting Paris Page | 25
9. Ambition level – mitigation mechanism
What should be the mechanism for strengthening commitments over time?
0: No ambition mechanism
30: No backsliding principle
40: A non-binding progression principle
65: A binding progression principle
100: A binding commitment to strengthen targets in line with the 2 degrees
goal
10. Ambition 2050
What should be the strength of the mitigation goal set for 2050?
0: No 2050 goal
20: Qualitative goal
30: Qualitative goal with a roadmap
50: Goal of 40% GHG reduction relative to 2010
70: Goal of 70% GHG reduction relative to 2010
100: Goal of zero net emissions
11. Ambition 2100
How ambitious should the mitigation goal for 2100 be?
0: No 2100 goal
20: Qualitative goal
30: qualitative goal with a roadmap
80: Goal of zero net emissions
100: Goal of negative net emissions
12. Ex-ante assessment (EAA) of future (I)NDCs
Which provisions for assessment and review of the nationally determined
contributions will be included in an agreement?
0: Option 1: No EAA
20: Option 2: EAA of aggregate ambition
60: Option 3: EAA of aggregate ambition and technical EAA of individual
INDCs (transparency, clarity, comparability, etc.)
90: Option 4: Option 3 plus a political assessment of individual INDCs
(ambition and equity/fairness)
100: Option 5: Option 4 and a formal mechanism for involving inputs from
civil society
Predicting Paris Page | 26
Appendix 2: Core Concepts of the Predictioneer’s Game
Stated Position of Stakeholders
The stated position is the outcome currently advocated by a stakeholder.
It is important to note that the stated position is not necessarily any of the
following positions, which stakeholders will often not reveal:
• the outcome that the stakeholder would truly prefer above all
others,
• the outcome that the stakeholder anticipates at the end of the
negotiations, or
• the outcome that the stakeholder is prepared to accept.
It is not uncommon for a stakeholder to have a “public” position on an issue
that is different than the position that he advocates to other stakeholders.
One example of this is when elected officials have a public position geared to
their electorate, whereas they indicate their actual bargaining position on an
off-the-record basis to other stakeholders. Where available, it is the
indicated bargaining position that should be recorded as the stated position.
Salience
Salience is the commitment the stakeholder has in pursuing this issue over
all other issues and topics competing for attention. Salience is not a
measure of the percentage of time or effort that a stakeholder will devote to
an issue. Rather, salience is a measure of the stakeholder's preparedness to
focus on the issue when it comes up, even if it means putting aside some
other issue. Salience is not primarily judged in relation to other
stakeholder's time and attention.
Numerical Definitions:
90-100: This is the most important issue to the stakeholder. The
stakeholder would drop whatever they are doing and turn to
this issue whenever asked.
70-80: This issue is very important to the stakeholder. It is certainly
one of the most important issues. The stakeholder would try
very hard to reschedule to handle this issue when it arises.
50-60: This is one of several important issues. Others are more
important. The stakeholder would have to drop this if one of
those other issues arose, but otherwise would try to focus on
this issue.
30-40: This is an issue the stakeholder cares about, but it is not that
important to the stakeholder. The stakeholder has many more
important issues to deal with and so generally would not drop
what they are doing to deal with this and generally would focus
on something else.
Predicting Paris Page | 27
10-20: This is a minor issue to the stakeholder. The stakeholder rarely
pays attention or makes much effort.
<10: The stakeholder really doesn’t care about this issue.
Potential Influence
Potential influence, often referred to as “resources,” is simply the bargaining
clout of each stakeholder, relative to each other. It is a measure of the
amount of influence a stakeholder could have on determining the outcome
relative to other stakeholders if all stakeholders were fully motivated.
It is important to distinguish potential power from salience. Just because a
stakeholder is not involved actively in an issue does not mean that the
stakeholder could not exert great influence. The president of the United
States, for instance, has the potential to influence many issues. The
president controls a lot of resources but is also very busy, so few issues can
have high salience. Advisers to the president, by comparison, may have
fewer resources but exert more effective influence because they exert more
effort (the issues have higher salience for them). Their influence scores
should be lower than the president’s and their salience scores should be
higher in this case.
When comparing one stakeholder’s resources with another’s, one measures
only those resources controlled directly by the stakeholder and not
resources controlled by associates or allies of the stakeholder who are also
listed as stakeholders. This is important to avoid double counting.
Numerical Definitions:
100: The most powerful stakeholder on this issue. There can be
more than one group at this score or at any other score.
All other: A stakeholder’s value must be positive and must be
Values evaluated relative to 100 (or the maximum score assigned)
and relative to the values other stakeholders. So, two
stakeholders with 40 and 60 would equal the one
stakeholder at 100 in a head to head contest with no one else
involved if everyone tried as hard as they could. Two groups
at 15 and 30 would, if they shared a common position, be
very close in potential influence to a group at 40 and
probably would just barely persuade the 40 to accept their
point of view if there were no other players involved. The
resource scores should not be thought of as percentages. A
decision-maker with a score of 100 does not have 100
percent of the resources and may, in fact, have only a small
percentage of the total. The total, of course, is the sum of all
of the resources across all of the groups or decision-makers.
Predicting Paris Page | 28
Flexibility/Resolve
Every stakeholder is assumed to care about two dimensions when
addressing an issue. The “position” variable assesses the outcome the player
currently advocates. Flexibility/Resolve evaluates the stakeholder’s
preference for reaching an agreement as compared to sticking to his or her
preferred position even if it means failing to reach an agreement.
The variable ranges between 0 and 100. Higher values reflect greater
flexibility; lower values greater resolve.
• A convenient rule of thumb is to think of the value as revealing the
percentage magnitude of proposed position shifts the player would
pay any attention to. For example if the scale is 100 points, a player
with flexibility of 20 would listen to arguments for positions within 20
points higher and 20 points lower than his/her current position.
• Players generally have a value on this variable that is 35 or lower.
• A value of 0 means the player declares that s/he is immovable and
resolved to stick to the current position. The model recognizes that
this may be a bluff or a sincere position and calculates the odds of
each as part of its updating process.
Optimization
Nash equilibrium requires that each player adopts the best strategy
available to them given the constraints of the game. That, of course, is true
for the Predictioneer's Game in which players pursue Perfect Bayesian
Equilibrium strategies. However, the model also assumes that players
cannot look ahead more than one period and that they act rationally on
their beliefs about each other player's type.
The Predictioneer’s Game assumes that when there are many players
involved people have a harder time sifting through all interactions to fully
realize the information available to them about other players. That is,
players observe outcomes in each round but not the details of interactions
between player pairs that do not involve them. The game's software exploits
this asymmetry between what the computer "knows" by keeping track of all
pairwise interactions (and third party contributions to them) in each period
while players only know this information for the pairs for which they are
members. This means that the output allows us to exploit information more
fully than players can (by assumption) and so the output identifies "optimal"
proposals that any selected player or group of players could make in each
round, taking advantage of fuller information about player types computed
by the model but assumed not to be accessible to the players who only
update on their pairwise interactions.
The Predictioneer’s Game can be set to use these "optimal" proposals rather
than the proposals players otherwise are expected to make in the game. The
"optimal" proposals correct both for errors of commission -- asking for less
Predicting Paris Page | 29
than one could have gotten or asking for too much to be taken seriously --
and errors of omission -- failing to make a proposal when a credible proposal
was available but outside the belief structure of the player involved. It is, of
course, difficult to implement "optimized" actions but their effects can
generally be approached. Hence, they provide information on the maximum
deviation a player can achieve relative to the un-optimized "base case."
References
Bueno de Mesquita, B. (2009). The Predictioneer's Game: Using the Logic of
Brazen Self-Interest to See and Shape the Future. New York, Random
House.
Bueno de Mesquita, B. (2011). "A New Model for Predicting Policy Choices."
Conflict Management and Peace Science 28(1): 65-87.
Kallbekken, S. and H. Sælen (2015). Predicting Paris: Forecasting Key
Outcomes from COP 21 Using an Expert Survey. Oslo, CICERO<
http://www.researchgate.net/publication/282974577_Predicting_Paris_-
_Forecasting_key_outcomes_from_COP_21_using_an_expert_survey .
Stokman, F. N. and R. Thomson (2015). Forecasting the Paris 2015
UNFCCC Negotiations. The Exchange Model's Analysis of
Developments and Potential Obstacles to Reaching an Agreement,
http://www.researchgate.net/publication/282974338_Forecasting_the_Paris_2015_UNFCCC
_Negotiations._The_Exchange_Model's_Analysis_of_Developments_and_Potential_Obstacles_to
_Reaching_an_Agreement .
United Nations Framework Convention on Climate Change. (2015). "Ad Hoc
Working Group on the Durban Platform for Enhanced Action:
Negotiating Text."
https://unfccc.int/resource/docs/2015/adp2/eng/01.pdf.
