An overview of some international organisations, with their voting structures

Summary 

• Membership: size, heterogeneity
• Scope
• Vote distribution: unweighted / weighted
• Voting rule:

  • Majority / supermajority / unanimity
  • Normal majority rules / special majority rules
• Selective representation: cameralism, permanent seats, etc

• Early period: unanimity voting
• 19th/early 20th century: international technical unions began to use majority voting
• After WW1: majority voting spread, international commodity organisations began to use weighted voting
• End of WW2: IMF and the World Bank set up with weighted voting, rest of the UN uses majority voting (with vetoes for the permanent seats on the Security Council)
• 1960s/70s: developing countries push against weighted voting and for bloc voting
• [I’m not sure if there are important trends after 1980]

• Level of democracy among founding members doesn’t correlate with voting rules
• Perceived legitimacy and problem-solving increase when orgs:

  • Are regional rather than global
  • Use deliberative diplomatic practices rather than bargaining
• Weighted voting seems pretty controversial 
• Rules are often hacky and strange and seem the upshot of particular negotiations
• The strictest voting rules usually relate to hard power (money, security) or amendments

• Intelsat: the US could have set up a communications satellite system unilaterally, but wanted the PR benefits of an international org, and was willing to trade some control for that.
• International Seabed Authority: set up to pre-empt conflict over deep sea mining once it became possible, but failed to get the US on board without weighted voting.
• ITER: recent organisation including the US, China, Russia, India, and EU to develop fusion technology. Uses weighted voting, and the EU has substantially more votes than any other individual member (45% to 10%). Has a unanimity rule for many issues.

General analysis

Who knows about this?

• International law

  • Zamora (1980)
  • Posner and Sykes (2014) - law and economics
• Political science

  • Blake and Payton (2014) - institutional design
• IR

  • Koremos, Lipson and Snidal (2001) - rational design
  • Panke, Polat and Hohlstein (2022) - comparative politics, legitimacy studies

What are the main structural options?

• Membership: size, heterogeneity
• Scope
• Vote distribution: unweighted / weighted
• Voting rule:

  • Majority / supermajority / unanimity
  • Normal majority rules / special majority rules
• Selective representation: cameralism, permanent seats, etc

1. One-vote-per-state versus weighted voting where different states have a different number of votes. 
2. The strength of the voting rule, ranging from majority rule, through various supermajority rules, to consensus. 
3. Cameralism, or the clustering of states with similar interests into different bodies that must separately approve a resolution. 
4. Variation in the scope of the authority of the voting body—regarding whether it can make a legally binding decision or not, and the importance of the decision that it is permitted to make. 
5. Different voting rules or procedures for different types of decisions—procedural versus substantive, for example.”1

• Membership (exclusive or inclusive, regional or universal, states only or NGOs too)
• Scope of the issues covered
• Centralization, including to disseminate information, to reduce bargaining and transaction costs, and to enhance enforcement
• Control, of which voting structure is one important aspect
• Flexibility, including adaptive and transformative

What patterns do we see?

Historically

• It was believed that international decisions couldn’t be imposed on states against their will without that violating their sovereignty

• Unanimity voting wasn’t practical
• The scope and authority was limited, so states were willing to cede veto powers
• They tended to be manned by technical experts, who were more willing to accept majority decisions than diplomats

• The League had unanimity but not for all issues; the ILO had majority rule
• International commodity organisations were suited to weighted voting because they had:

  • Narrow functions
  • Easily defined criteria for weighting (exports and imports)

• Most international lending institutions outside the UN also use weighted voting.11

• UN agencies tended to favour special majorities over weighted voting as a safeguard, for political reasons.12

Image

• (Note that this hides lots of complexity: organisations often have different rules for different bodies and different issues.)14

• UNCTAD was set up in 1964 as an alternative to GATT, and is the first organisation to use formalised political (rather than functional) bloc voting.
• Some developing countries still supported weighted voting in the international development banks and international commodity agreements though.

Today

• Unanimity voting: smaller orgs and security-related orgs are more likely to use unanimity voting.
• Democracy: The level of democracy among founding members does not have a significant impact on the choice of voting rules.

• Weighted voting: international commodity orgs,17

  development banks18

   
• Majority voting: most UN agencies,19

  international courts,20

  most human rights orgs21

• Distribution problems
• Enforcement problems
• Number of actors involved
• Uncertainty about behaviour, state of the world, and preferences

1. decision costs; 
2. exploitation costs; 
3. heterogeneity (meaning that some states gain more from collective decisions than others or lose more from adverse decisions than others); and
4. discount factors (which of course can be heterogeneous as well).”22

• More homogeneous groups will tend to have weaker voting rules.
• States with higher discount factors will have weaker voting rules.

  • These states will tend to be richer states.
• So “organizations involving only rich states will have weaker voting rules than organizations involving poor states or a mix of rich and poor states.”

• Recommendatory economic organisations tend to have simple voting procedures and limited decision making outside formal votes.24

• Task-oriented economic organisations have more formal voting safeguards (to ensure the support of influential members), and use informal routes more (for efficiency).25

  • But this requires high goal consensus.26

• “Most organizations require a simple majority of the membership to establish a quorum and a simple majority of the votes cast to make most decisions. However, the more important the purpose of the organization, the higher the normal majority requirement, with political and economic organizations generally having the highest normal majority requirements.”27

• Weighted voting orgs tend to have more and higher special majority decisions.28

• “The most efficient agencies are the least democratic [i.e. use weighted voting], yet they are also the organizations that operate with the highest degree of consensus.”29

  • They have well-defined functions which don’t threaten individual members.
  • There are objective criteria for decisions which reduces conflict.
  • They have highly competent secretariats.

What do we know about the strengths and weaknesses of different structures?

From theory

• It’s easy to get people to agree to unanimity voting, but hard to do anything with it
• It’s easy to do things with majority voting, but hard to get (powerful) people to agree to it
• Powerful states like weighted voting

From empirical studies

• Perceived legitimacy increases with: 

  • Regional (over global) membership
  • Non-state actor access
  • Deliberative diplomatic practices
• Perceived problem-solving increases with:

  • Regional (over global) membership
  • Autonomous secretariats
  • Deliberative diplomatic practices
• Perceived problem-solving decreases with:

  • Consensus decision-making rules

Particular organisations

Summary table

• Organisations with weighted voting may have either majority or supermajority rules. I’ve put notable rules in brackets
• Many organisations have different rules for different bodies/kinds of vote. Where I haven’t specified, this is the ‘normal’ rule in the main body of the organisation
• I’ve marked organisations with permanent seats with an ‘✢’ and those with cameralism with an ‘⁑’
• There’s a summary table by organisation here
• I’m not using precise definitions or selection criteria, and instead am pragmatically looking at examples which seemed relevant

Points of interest

• Weighted voting seems pretty controversial

  • Tentatively I do think weighted voting would be a feasible mechanism between ~Western allies

    • Mostly seems to be developing nations opposing weighted voting

      • Controversy around World Bank, IMF
      • It sounds as though in 1994 the International Seabed Authority (which includes developing nations) a) wouldn’t agree to weighted voting, b) consequently lost the US as a member 
    • The UK and the Netherlands have both joined weighted voting orgs in the recent past, including ones where they don’t have tonnes of voting power

      • Asian Infrastructure Investment Bank (AIIB), founded in 2016, which uses weighted voting based on contributions. 100 members including China but not the US, Netherlands has around 1% of the votes, UK around 3%
      • Netherlands also a member of the European Stability Mechanism (2012), which uses weighted voting based on contributions
  • Most likely ways I could be wrong about this:

    • Weighted voting is acceptable specifically in the context of lending institutions
    • Weighted voting is acceptable specifically in the context of orgs where members make substantial financial contributions

      • An AI joint project would probably meet this condition, but might not
• Rules are often hacky and strange and seem the upshot of particular negotiations

  • Council of the EU (Treaty of Nice): weighted, and the weights were just negotiated rather than based on objective criteria
  • International Seabed Authority: complicated system with four different blocs that need to be represented in the Council and which must all vote in favour for something to pass. Landlocked nations must be represented
  • Security Council: veto at the insistence of permanent members, supermajority at insistence of everyone else
  • Intelsat: important matters require 12.5% of votes in addition to US votes (requiring at least some European backing), but this drops to 8.5% after 60 days (not requiring any European backing)
• The strictest voting rules usually relate to hard power (money, security) or amendments

  • Money and security:

    • International Seabed Authority: unanimity for distributing benefits
    • IMF: 70% for investment decisions, 85% for quota changes
    • NATO: unanimity
    • Council of the EU (Treaty of Lisbon): unanimity for tax and security
    • UN Security Council
  • Amendments: IMF 85%, WTO 75%, WHO ⅔
• Intelsat is pretty interesting

  • The US could have set up a communications satellite system unilaterally, but wanted the PR benefits of an international org, and was willing to trade some control for that.
• The International Seabed Authority is pretty interesting

  • Set up to pre-empt conflict over deep sea mining once it became possible, but failed to get the US on board without weighted voting.
• ITER is pretty interesting

  • Recent organisations including the US, China, Russia, India, and EU to develop a new technology. Uses weighted voting, and the EU has substantially more votes than any other individual member (45% to 10%). Has a unanimity rule for many issues.

Organisations 

• These are arranged in rough chronological order.
• With the exception of GAVI, they are all regular international organisations

  • I didn’t include the Manhattan Project, as understanding its decision making seemed like a large undertaking
• I cover basic voting rules for all orgs. I also include (sometimes very) brief notes on:

  • Strengths and weaknesses, for the World Bank, IMF, UN General Assembly, UN Security Council, CERN, IAEA, OECD, IPCC, WTO
  • Why those rules were chosen, for UN General Assembly, UN Security Council, GATT, CERN, Intelsat, UNCTAD, International Seabed Authority, WTO

World Bank55

• International Bank for Reconstruction and Development (IBRD) - 1944
• International Finance Corporation (IFC) - 1956
• International Development Association (IDA) - 1960
• Multilateral Investment Guarantee Agency (MIGA) - 1988

• IBRD: Provides loans and financial advice to middle-income countries for development projects.
• IFC: Invests in private enterprises in developing countries to stimulate economic growth.
• IDA: Offers interest-free loans and grants to the poorest countries, supporting social and economic programs.
• MIGA: Issues guarantees against non-commercial risks to encourage foreign investment in developing countries.

• IRBD: 85% for amendments, 80% for increasing the number of Directors, 75% for increasing capital stock, unanimity for withdrawal57

• IFC: 85% for increases in capital stock, 75% for issuing shares, ⅗ and 85% of voting power for amendments, unanimity for withdrawal58

• IDA: ⅔ for increasing subscriptions, ⅗ of members holding ⅘ of voting power for amendments, unanimity for withdrawal59

• MIGA: ⅔ and 55% of shares for amendments and various other things60

• Directors get certain share of the vote regardless of what share they were elected by61

• Directors cannot split their votes62

• Bretton Woods. Very similar to the IMF. actually made up of many bodies
• Developing countries complain they are underrepresented and that lending favours nations who are friendly with creditors and comes with counterproductive conditions63

• “[I]n the World Bank, voting power is aligned to ownership shares. At its founding, when ownership shares were closely related to financial contributions, this made sense. Today, however some major shareholders, especially the United States exercise influence that is out of proportion to their current costs”64

IMF65

Image

• 70% for investment decisions
• 85% for amendments and quota changes

  • Note the US has an effective veto here, and is the only country to do so

• Formal votes are rarely taken on the Executive Board, which strives for consensus

• Large members have their own executive director. Small members band together to elect directors. Each director has votes equal to the votes of the countries they represent.

• Developing countries complain they are underrepresented and so conditionality is too restrictive and borrowers don’t get as much benefit as they should
• (Here is an interesting blog on this; changes are currently blocked because of the 85% requirement. Currently, China and Europe have too few votes and the US has too many.)

UN General Assembly67

• Decision-making is easy
• The majority often outvote the minority of powerful states
• Anticipating this, powerful states made sure resolutions would be non-binding

UN Security Council

• Powerful countries insisted on a veto
• Small countries insisted on a high supermajority, so that multiple rotating votes would be required
• The result is a very strict voting rule, and gridlock by default

GATT68

• Unanimity for basic rules
• ⅔ to amend other articles

• ITO never gets off the ground as votes aren’t weighted in favour of Western powers
• GATT is a part of the ITO charter and does survive
• Western powers joined even though it had an unweighted voting structure, because:69

  • It was mostly a negotiating forum
  • Withdrawal was easy

WHO

NATO

• Member countries voluntarily provide military forces and capabilities, and cover the costs of deploying their own troops.
• Financial contributions to the common budget are made according to a cost-sharing formula based on Gross National Income.

CERN

• External observers pushed for weighted voting, and using unweighted was controversial71

• See [Final] CERN Case Study, Maas and Villalobos (2023) and Hausenloy and Dennis (2023) for strengths and weaknesses.

IAEA72

• General Conference of all member states
• 35 member Board of Governors

• General Conference: ⅔ for finances, amendments, suspension73

• Board: ⅔ for budget, election of Director General, reconsidering rejected amendments74

• Description

  • “The International Atomic Energy Agency (IAEA) was founded in 1957 to promote the peaceful use of nuclear energy and limit its use for military purposes. The IAEA exists as an autonomous organization within the United Nations. In practice, one of its main purposes is to verify that states do not build nuclear weapons. 
  • The IAEA can conduct inspections to ensure that states are not secretly building nuclear facilities. Findings from inspections are reported to the IAEA Board of Directors. If the Board of Directors believes that a state is not complying with international agreements, the Board can escalate the issue to the UN Security Council. 
  • Ultimately, the IAEA does not have the authority to take action directly– it simply provides information to the UN Security Council and member states.”75

• Strengths:76

  • Proven success in governing nuclear technology for over 50 years
  • Established verification mechanisms for compliance
  • Some analogies to AI, such as monitoring hardware (like uranium in nuclear)
  • Success in limiting proliferation while enabling beneficial development
  • Ability to inspect systems, require audits, test for compliance with safety standards, and place restrictions on deployment and security levels77

• Weaknesses as an analogy for AI:

  • AI is harder to verify, audit and safeguard than nuclear materials

    • AI systems cannot be safeguarded in the same way as nuclear materials.78

    • Challenges in verifying opaque AI systems that are "black boxes"79

    • Lack of clear standards for AI systems to audit against80

  • The IAEA moved more slowly than would be necessary for AI 

    • The IAEA's development was slow (established in 1957, but only took a leading role in nonproliferation in 1968), which might not be adequate given the rapid pace of AI development.81

    • Unable to match the rapid pace of AI progress (IAEA projects take 1-2 years)82

  • The actors involved are different for AI

    • Focused on state actors, not private companies who are leading AI development83

    • The effectiveness of an IAEA-like agency for AI would depend on the existence of supportive international treaties and specific incentives for participation.84

      • UN’s limited technical expertise in AI safety85

  • The intense level of oversight provided by the IAEA might be prohibitively difficult to negotiate for AI.86

OECD

• Able to move quickly on policy-setting because its membership is mostly aligned88

Intelsat

• Interim agreements (1964-1973)

  • ISCS (executive)
  • Comsat (manager)
• Definitive agreements (1973-2001)

  • Assembly of parties (treaty changes)
  • Meeting of signatories (approves board)
  • Board of governors (manager)

• Interim agreements: international telephone traffic and domestic traffic above 2000 km (Lipscy)
• Definitive agreements: usage of the Intelsat system

• Interim agreements:

  • Unanimity, or failing that, US+12.5% for 14 key decisions

    • Decisions: types of space segment, standards for earth stations, budget, adjusting accounts, establishing rates, additional contributions, placing contracts, satellite launches, quotas, withdrawal, amendments, adopting rules of procedure, compensating Comsat89

    • Meant some European votes would be required
    • Though note that the UK had successfully pushed for freedom for European countries to not vote in a bloc90

  • After 60 days, US+8.5%, for a subset of key decisions

    • Decisions: space segment, budget, contracts and launches91

    • Intended to prevent the Europeans blocking things indefinitely92

  • Majority for procedural issues (Codding, p. 27)
• Definitive agreements:

  • Assembly: 2/3 of states plus 2/3 investment or 85% of membership (Codding)
  • Meeting of Signatories: majority for procedural, 2/3 for substantive (Codding)
  • Board of Governors:93

    • Simple majority for procedural matters
    • 2/3 majority of voting participation or total Board minus 3 for substantive matters
    • 2/3 majority to overrule the Chairman on procedural vs substantive determinations

• Interim agreements: most decisions decided without a vote (Codding, p. 27)

• Interim agreements: new members max 17%, with all other members' shares decreasing proportionally (which meant the US would always have a minimum of 50.5%) (Lipscy)
• Definitive agreements: US limited to 40% regardless of usage (Levy)

• Intelsat is an interesting model for AGI governance because:

  • It was a successful multilateral project to build a primarily commercial technology of geopolitical importance.
  • It was US-led, with the US retaining control, but with meaningful oversight from allies.
  • It provided major benefits to both the US and its allies, compared to the US going alone.
  • It was set up quickly, under interim agreements with an initially small number of close allies, with scope for flexibility long-term.
  • It is a successful example of differential technological progress, and led to a system which was strategically more beneficial to the US than would otherwise have arisen.
• Disanalogies with AGI don’t seem to undermine this:

  • The US needs allies even more for the development of AGI than it did for satellite communications.
  • The USG was more involved in the development of satellite communications technology than it has been so far in AGI development. But if scaling continues, this might need to change in order to reach AGI.

UNCTAD94

• Formal conciliation procedure to arrive at common position before a vote
• In fact never used, and informal consensus reached between blocs

• Set up by developing countries as a GATT alternative
• First formalised political bloc voting96

IPCC

• See Hausenloy and Dennis (2023) for strengths and weaknesses.

International Seabed Authority98

• Council issues regulations and licences and proposes the budget
• Assembly sets policy, approves the budget and distributes benefits

• Assembly: majority for procedural, ⅔ for substantive
• Council: majority for procedural, ⅔ and the majority of each chamber for substantive, unanimity for distribution of benefits101

  • “Effectively, this means that all decisions require a majority or tie vote in each chamber, plus two-thirds of all members in the aggregate.”102

• Council: 4 chambers:

  • 4x largest consumers of minerals like those in the seabed
  • 4x largest investors in deepsea mining
  • 4x exporters of minerals like those in the seabed (min 2x developing)
  • 24x developing states (inc 6x with special interests like being large or land-locked)

• Set up by the UN Convention on the Law of the Sea
• Deep sea mining wasn’t yet possible, but they wanted to prevent conflict and over extraction
• The plan was to generate earn royalties by licensing companies, and redistribute that to members
• 1994 agreement introduces cameralism, but the US wanted weighted voting and refused join103

WTO104

• ¾ for waivers and interpretations of obligations
• ⅔ for new members, but no member needs to extend its obligations to a new member without consent
• Any amendment to obligations is non-binding on members
• In arbitration (which are settled by majority rule on a panel), final decisions are automatically adopted by all members unless there is reverse consensus against the decision

• “Holdout issues were significant, and some GATT members balked at some of the proposed new commitments. In response, the major players agreed on a novel strategy – they would formally withdraw from the GATT, and enter a new treaty creating the WTO. Any GATT member who wished to retain the benefits of GATT membership in relation to the major players had to do the same even if they did not like aspects of the new WTO regime. Some members complained that the process was coercive, but they had little choice but to capitulate.”105

• It’s very hard to add or modify commitments
• There are side agreements which only bind the members who sign them, and separate free trade agreements (e.g. NAFTA, TPP)

ITER106

• Europe: 45.6%
• China, India, Japan, South Korea, Russia, US: 9.1% each

• Outlined in the Council’s rules of procedure, which unfortunately I can’t find
• Unanimity required for adopting the weighted voting system and the rules of procedure, electing the Director-General, budget, changes to cost sharing, new members, amendments, and various other things

GAVI

Council of the EU (Treaty of Lisbon)108

• 55% of members if proposed by Commission, or 72% otherwise
• 65% of EU population

  • Note that proposals that fail to meet this can still go through unless at least four members vote against, to prevent the three largest members from having a veto

• “[T]o a significant degree, unanimity requirements in the Council are being replaced with a co-decision procedure that requires non-unanimous approval in both the Council and the Parliament. The movement away from unanimity has its roots in a desire to promote greater policy flexibility. But the enhanced role of Parliament limits the importance of that change.”109

Appendices

Voting share in major banks

Image

Voting and design objectives

Image

Size and issue of IGOs

Image

Image

IO effectiveness and legitimacy

Image

Selected bibliography

The structure of institutions

Institutions relevant to AI