Federal and confederal designs compared: from the Articles of Confederation to Rojava

Federations and confederations are the main ways of combining self-rule and shared rule among territorial units. They range from loose alliances of sovereign members to integrated states with constitutionally protected regions. This study compares six designs: Switzerland, the United States under the Articles of Confederation and under the Constitution, the Haudenosaunee Confederacy, the European Union, the democratic confederalism of north-eastern Syria, and Ethiopia's ethnic federalism. It sets them against the main theories, William Riker's bargain theory, Alfred Stepan's distinction between "coming-together" and "holding-together" federations, and Daniel Elazar's account of federalism as self-rule plus shared rule, and asks what failure modes recur. We find that confederal designs that require unanimity and lack their own revenue tend to be replaced or to stagnate, that federations survive when their safeguards are multiple and mutually reinforcing, and that designs built on ethnic units are fragile when one unit dominates, when the governing party substitutes for federal institutions, or when the formal right to secede is combined with central control in practice.

Key findings

• The United States moved from a confederation to a federation within a decade: the Articles of Confederation, in force from 1 March 1781, gave Congress no power to tax and required the consent of every state legislature for amendment; the Constitution replaced them in 1789.
• Switzerland's path ran from a confederation of sovereign cantons to a federal state after a short civil war (3–29 November 1847, about 100 dead) and the Federal Constitution of 1848. Its federal design is embedded in direct democracy and a double majority of people and cantons for constitutional change.
• The Haudenosaunee Confederacy has kept a confederal, consensus-based Grand Council of 50 chiefs for centuries, but its founding date is disputed (estimates range from about 1142 to about 1600) and its practical jurisdiction is limited by the United States and Canada.
• The EU combines confederal and federal features: member states remain the "masters of the treaties", but the Union legislates by qualified majority in many fields and its law has primacy. Conflicts over the final arbiter, such as the German Constitutional Court's 2020 PSPP judgment, are a recurring tension. The UK's departure in 2020 showed that exit is a real option.
• Rojava's democratic confederalism built communes and councils across north-eastern Syria from 2012 but was never recognised by Damascus. After an agreement of 29 January 2026, its commander announced the dissolution of the SDF and the Autonomous Administration on 25 August 2026.
• Ethiopia's 1995 constitution grants every "nation, nationality and people" an unconditional right to secession, yet the federation was long run through a single ruling coalition. The breakdown of party-mediated relations contributed to the Tigray war (November 2020 to November 2022), with death toll estimates ranging from about 100,000 to 600,000 or more.
• Comparative research identifies "core ethnic regions", single-party dominance and weak judicial or political safeguards as risk factors for federal collapse.

Research question and scope

What makes federal and confederal designs work, and how do they fail? We define:

• Confederation: a union of states that retain sovereignty, with common institutions that act mainly through the member governments and decide important matters by unanimity or supermajority.
• Federation: a state in which authority is constitutionally divided between a central government and constituent units, each acting directly on citizens within its sphere, and in which neither can unilaterally abolish the other.

We compare cases chosen to span the spectrum and to include both long-lived and failed designs. We do not attempt a census of the world's two dozen or so federations, and we treat devolved unitary states such as the UK separately, in the companion study on territorial autonomy.

Background

Three theories

Riker's bargain. William Riker's Federalism: Origin, Operation, Significance (1964) argued that federations are bargains struck by politicians who want to expand territory or meet a military threat, and who must concede autonomy to the constituent units to win their agreement. On this view, federations are born of external threats and kept together largely by political parties.

Stepan's correction. Alfred Stepan, writing in the Journal of Democracy in 1999, argued that the US model of "coming-together" federalism, in which sovereign units pool their sovereignty, is only one type. Many federations, including India, Belgium and Spain in its quasi-federal form, are "holding-together" federations, in which a unitary state devolves power to keep a diverse society together. Others, such as the Soviet Union, were "putting-together" federations imposed by force. Stepan also noted that most people in long-standing democracies live in federal systems, and that the territorial break-ups of post-communist Europe all occurred in formally federal states.

Elazar's definition. Daniel Elazar, in Exploring Federalism (1987), defined federalism as the combination of self-rule and shared rule. He treated it as a family of arrangements, from confederations through federations to federacies and associated states, rather than a single model.

Later work on survival

Later research turned to the question of why some federations last. Mikhail Filippov, Peter Ordeshook and Olga Shvetsova argued in Designing Federalism (2004) that because members always have incentives to renegotiate the rules, a federation needs actors "naturally committed" to the status quo; integrated national political parties play that role. Jenna Bednar's The Robust Federation (2009) argued that federations need multiple, complementary safeguards, including courts, parties and the people, each covering transgressions the others miss. Henry Hale (2004) found that all ethnofederal states that collapsed had a "core ethnic region" with a pronounced population superiority. Valerie Bunce's Subversive Institutions (1999) explained how the institutional design of socialist federations helped bring about the dissolution of the Soviet Union, Yugoslavia and Czechoslovakia.

Method and sources

We compare the six cases on five dimensions: founding logic (coming together, holding together or putting together), decision rules, fiscal powers, safeguards against central or member overreach, and exit rules. We use constitutional texts where available (the Swiss Federal Constitution, the Articles of Confederation through the US National Archives, the Ethiopian Constitution through the Constitute Project, the Treaty on European Union, the 2014 Charter of the Social Contract of north-eastern Syria), court decisions and standard secondary works. For contested points, such as the dating of the Haudenosaunee League or casualty estimates for the Tigray war, we report the range of estimates.

Finding 1: confederations without revenue and with unanimity rules tend to be replaced

The Articles of Confederation

The Continental Congress adopted the Articles of Confederation on 15 November 1777; they entered into force when Maryland ratified them on 1 March 1781. Each state had one vote in Congress regardless of population. Major decisions, such as declaring war, making treaties and coining money, required nine of the thirteen states. Amendments required the agreement of Congress and confirmation by every state legislature. Congress could not levy taxes, had little authority over commerce and limited means of settling disputes between states.

These weaknesses became acute in the 1780s, as Congress struggled to pay war debts and the states competed over trade. The Philadelphia Convention of 1787 did not amend the Articles under their own unanimity rule. It drafted a new Constitution to take effect once nine states had ratified, which replaced the Articles in 1789. James Madison described the result in Federalist No. 39 as "neither a national nor a federal Constitution, but a composition of both": the new government acted directly on citizens and could tax, while the states retained their own governments and equal representation in the Senate.

What the American case teaches

The Articles show two classic confederal failure modes: dependence on member governments for money and a unanimity rule that makes reform impossible. They also show a less discussed lesson: the shift to a federation was achieved by stepping outside the confederation's own amendment rule, which raises questions of legitimacy that any reform of a confederation must face.

Finding 2: Switzerland combined federation with a dense web of safeguards

From confederation to federal state

The Old Swiss Confederacy was a network of alliances among sovereign cantons. After the French-imposed Helvetic Republic (1798–1803) and the restoration period, conflict between liberal Protestant cantons and conservative Catholic cantons came to a head when seven Catholic cantons formed a separate alliance, the Sonderbund. The resulting civil war lasted from 3 to 29 November 1847 and cost about 100 lives. The victorious side drafted the Federal Constitution of 1848, which created a federal state with a bicameral parliament: the National Council representing the people and the Council of States representing the cantons. The constitution was totally revised in 1874 and again in 1999; the current Federal Constitution of 18 April 1999 entered into force on 1 January 2000.

Why it has lasted

Several features reinforce each other in the Swiss design. The cantons are sovereign insofar as their sovereignty is not limited by the federal constitution, and they hold residual powers. Constitutional amendments require a double majority of the people and the cantons in a mandatory referendum, so neither the large cantons nor the small ones can impose changes alone. Optional referendums and popular initiatives give citizens a direct veto and agenda-setting power. Cross-cutting cleavages of language and religion mean that no single canton or group dominates. The creation of the canton of Jura, which separated from Bern by a series of votes and became a canton in 1979, shows that even boundaries can be changed through the federal system's own procedures.

The Swiss case fits Riker poorly. It was born of civil war more than external threat, and it is held together less by integrated parties than by direct democracy, consensual executive government and a strong cantonal identity.

Finding 3: the Haudenosaunee Confederacy shows durable confederal consensus, with limits

In Haudenosaunee tradition, the Peacemaker and Hiawatha brought the Great Law of Peace to warring nations and united the Mohawk, Oneida, Onondaga, Cayuga and Seneca; the Tuscarora joined as a sixth nation in 1722. A Grand Council of 50 chiefs meets on Onondaga territory. Clan mothers select chiefs and can remove them. Questions pass between groups of nations, each seeking agreement, and the Onondaga, as keepers of the council fire, confirm or resolve disagreement. Each nation keeps its own council for internal matters.

The founding date is contested. Mann and Fields (1997) proposed 1142, linked to a solar eclipse in oral tradition; Henige (1999) questioned whether a myth can be astronomically dated; archaeological work by Kuhn and Sempowski (2001) points to coalescence around 1590–1605. The Confederacy itself treats the date as unknown. Claims that the Confederacy directly shaped the US Constitution are also disputed.

As a design, the Confederacy shows that confederal consensus can last for centuries in a polity with shared ceremonies, kinship institutions that cut across nations and a strong norm of deliberation. It also shows the cost of unanimity under external pressure: during the American Revolution the member nations divided, and the Confederacy's practical jurisdiction today is constrained by the United States and Canada, within whose borders its communities live.

Finding 4: the European Union is a hybrid with an unresolved question of final authority

The EU began as a set of economic communities among sovereign states and has acquired federal features without becoming a federal state. Its law has primacy over national law in its fields; many decisions in the Council are taken by qualified majority rather than unanimity; the European Parliament is directly elected; and the Court of Justice has developed the doctrines of direct effect and primacy. But the member states remain the "masters of the treaties": they alone can amend them, unanimity remains the rule in sensitive fields such as foreign policy and taxation, and the Union has limited own resources compared with federal budgets.

Two features illustrate its hybrid character. First, Article 50 of the Treaty on European Union, introduced by the Treaty of Lisbon, gives every member state a right to withdraw; the United Kingdom left on 31 January 2020. A federation rarely provides such a right; a confederation usually does. Second, the question of who has the final say over the limits of EU competence is unresolved. On 5 May 2020 Germany's Federal Constitutional Court ruled that the European Central Bank's Public Sector Purchase Programme decisions were ultra vires, and that the Court of Justice's 2018 judgment upholding them was itself ultra vires in its proportionality review. The Court ordered the Bundesbank to stop participating after a transitional period unless the ECB produced a substantiated proportionality assessment.

The EU therefore displays both confederal risks (unanimity in key fields, limited fiscal capacity, a right of exit) and federal tensions (conflicts of jurisdiction between the centre and constitutional courts of the members). Its survival so far rests on dense legal integration, economic interdependence and repeated renegotiation, in line with Bednar's emphasis on multiple safeguards.

Finding 5: democratic confederalism in north-eastern Syria was never recognised and has been absorbed

From 2012 Kurdish-majority areas of northern Syria came under a movement led by the Democratic Union Party (PYD), inspired by Abdullah Öcalan's reading of Murray Bookchin's libertarian municipalism. The Charter of the Social Contract of 29 January 2014 proclaimed democratic autonomy for three cantons, Afrin, Jazira and Kobane, as part of Syria. The system expanded and was reorganised in 2016 and again in 2018 as the Autonomous Administration of North and East Syria (AANES). Its design built from communes through elected councils to regional bodies, with co-chairs of both sexes and gender quotas. Critics, including Human Rights Watch in 2014, documented abuses in PYD-run areas and pointed to the leading role of a single party and its armed forces.

As a confederal design, the AANES faced an unusual problem: it was a confederation of communes and councils inside a state that never recognised it. After the fall of the Assad government in December 2024, the Syrian Democratic Forces (SDF) and the new authorities in Damascus agreed in March 2025 to integrate, but the agreement made little progress. After weeks of fighting in January 2026, a further agreement provided for the phased integration of SDF forces into the Syrian army and of civil institutions into state structures. On 25 August 2026 the SDF commander, Mazloum Abdi, announced the dissolution of the SDF and the Autonomous Administration, saying that the president had pledged to protect the right to study in Kurdish.

Whatever one's view of the system's internal democracy, its fate shows that bottom-up confederal institutions without recognition, fiscal base or external guarantee are highly vulnerable when the surrounding state reconstitutes itself.

Finding 6: Ethiopia's ethnic federalism combined a formal right to secession with party centralisation

Ethiopia's constitution, adopted in December 1994 and in force from 1995, created a federation of nine member states, largely defined by ethnicity and language, and two chartered cities. Article 39 grants every "nation, nationality and people" an "unconditional right to self-determination, including the right to secession", subject to a procedure: a two-thirds vote of the group's legislative council, a federal referendum within three years and a majority vote in that referendum. The federation has since grown to twelve regional states as new regions were created from the Southern Nations, Nationalities, and Peoples' Region: Sidama (2020), South West Ethiopia (2021), and South Ethiopia and Central Ethiopia (2023).

For most of its life the federation was run through a single coalition, the Ethiopian People's Revolutionary Democratic Front (EPRDF), whose discipline substituted for intergovernmental institutions. When the coalition dissolved in 2019 and the governing party was reorganised, relations between Tigray's ruling party and the federal government broke down. The federal government postponed the 2020 elections because of COVID-19; Tigray held its own regional election on 9 September 2020, which the federal government considered illegal. War began on 3–4 November 2020 and ended with a cessation of hostilities agreed on 2 November 2022. Estimates of the death toll range from about 100,000 (Uppsala Conflict Data Program) to 600,000 or more.

The Ethiopian case illustrates several failure modes that the theories predict: a federation held together mainly by a dominant party, as Filippov, Ordeshook and Shvetsova would expect, becomes brittle when the party fragments; an ethnic design creates incentives to mobilise along ethnic lines; and a generous formal right to secession offers little when central institutions do not mediate disputes. Fessha and Gebreyesus (2026) describe the result as "constitutional infidelity" leading to civil war.

Comparison of designs

Recurring failure modes

1. Fiscal dependence of the centre. Confederations that rely on contributions from members, as under the Articles, cannot sustain common goods.
2. Unanimity traps. Unanimity for reform protects members but blocks adaptation. The American founders escaped by stepping outside the rule; the EU manages by keeping unanimity for treaty change while using majorities for legislation.
3. Single-party substitution. When a dominant party holds a federation together, its collapse can bring down federal relations, as in Ethiopia and, on Bunce's account, in the socialist federations of Europe.
4. Core regions. Hale's finding that every collapsed ethnofederal state had a dominant core region suggests that dividing a large group among several units may be safer.
5. Unresolved final authority. Federations need a credible umpire. Where courts at different levels claim the final word, as in the EU's jurisdictional disputes, conflicts can recur.
6. Lack of recognition and guarantees. Confederal experiments that grow up without recognition by the surrounding state, like the AANES, depend on military balances that can change abruptly.

What the evidence does not show

• It does not show that federations are generally more stable than unitary states. The cases were chosen to illustrate designs, not to estimate average effects.
• It does not establish that ethnic federalism causes conflict. Ethiopia's war had many causes, including the dissolution of its ruling coalition and disputes over elections. Other federations with ethnic or linguistic units, such as Switzerland and India, have survived for long periods.
• It does not resolve the historical debates. The dating of the Haudenosaunee League and its influence on American constitutional thought remain contested.
• It does not judge the AANES's internal democracy. Assessments by supporters and by human rights organisations differ sharply; this study treats only its federal and confederal design.
• Casualty figures for the Tigray war are estimates. They differ by a factor of six or more depending on methods and sources.

Implications for institution design

1. Give the common level its own revenue. Shared rule without shared money is fragile.
2. Avoid unanimity for ordinary decisions; keep it, if at all, for foundational change. The Swiss double majority offers a middle path: a demanding rule that does not give any single member a veto.
3. Build several safeguards, not one. Courts, parties, second chambers and direct democracy each catch different transgressions. A federation relying on one of them, especially a dominant party, is vulnerable.
4. Design units to cut across dominant groups. Where possible, avoid a single region holding a majority of the population, and draw boundaries that reduce, rather than entrench, ethnic mobilisation.
5. Specify exit and boundary-change procedures. Clear, demanding procedures, such as Switzerland's for the Jura or the EU's Article 50, can channel conflicts that would otherwise become violent.
6. Plan for recognition. Bottom-up confederal experiments need a route to legal recognition and fiscal security if they are to outlast the circumstances that created them.

Open questions

• Can the EU resolve its final-authority conflicts without either becoming a federal state or weakening the primacy of its law?
• What happens to the institutions of communal self-government in north-eastern Syria after formal integration into the Syrian state?
• Can Ethiopia reform its federation, for example by strengthening intergovernmental institutions or redrawing units, without reopening conflict?
• Do digital tools for deliberation and voting make large-scale confederal consensus more feasible, or do they reproduce the unanimity trap?
• How far do Bednar's and Hale's findings, drawn largely from states, apply to supranational unions?

Sources

1. Riker, W. H. (1964). Federalism: Origin, Operation, Significance. Boston: Little, Brown.
2. Stepan, A. (1999). Federalism and democracy: beyond the U.S. model. Journal of Democracy 10(4), 19–34.
3. Elazar, D. J. (1987). Exploring Federalism. Tuscaloosa: University of Alabama Press.
4. Filippov, M., Ordeshook, P. C. and Shvetsova, O. (2004). Designing Federalism: A Theory of Self-Sustainable Federal Institutions. Cambridge University Press.
5. Bednar, J. (2009). The Robust Federation: Principles of Design. Cambridge University Press.
6. Hale, H. E. (2004). Divided we stand: institutional sources of ethnofederal state survival and collapse. World Politics 56(2), 165–193.
7. Bunce, V. (1999). Subversive Institutions: The Design and the Destruction of Socialism and the State. Cambridge University Press.
8. US National Archives. Articles of Confederation (1777).
9. Madison, J. (1788). The Federalist No. 39. Library of Congress, Federalist Papers full text.
10. Swiss Confederation (1999). Federal Constitution of the Swiss Confederation of 18 April 1999. Fedlex.
11. Wikipedia. Sonderbund War (dates and casualty estimates).
12. Mann, B. A. and Fields, J. L. (1997). A sign in the sky: dating the League of the Haudenosaunee. American Indian Culture and Research Journal 21(2).
13. Kuhn, R. D. and Sempowski, M. L. (2001). A new approach to dating the League of the Iroquois. American Antiquity 66(2), 301–314.
14. Haudenosaunee Confederacy. Government.
15. European Union (2016). Consolidated version of the Treaty on European Union. EUR-Lex.
16. Federal Constitutional Court of Germany (2020). ECB decisions on the Public Sector Purchase Programme exceed EU competences. Press release No. 32/2020, 5 May 2020.
17. Charter of the Social Contract (2014). Charter of the Social Contract, 29 January 2014.
18. Human Rights Watch (2014). Under Kurdish Rule: Abuses in PYD-run Enclaves of Syria.
19. Al Jazeera (2026). Kurdish-led SDF agrees integration with Syrian government forces.
20. Shafaq News (2026). Mazloum Abdi announces dissolution of SDF, AANES, 25 August 2026.
21. Federal Democratic Republic of Ethiopia (1994). Constitution of the Federal Democratic Republic of Ethiopia. Constitute Project.
22. Wikipedia. Regions of Ethiopia and Tigray war.
23. Fessha, Y. T. and Gebreyesus, T. (2026). Ethiopian federalism: ethnicity, constitutional infidelity, and civil war. In Federalism in a Turbulent Era. Edward Elgar.