Article 39: The Secession Clause and What It Did to the Federation

Thesis

Article 39 of the 1995 Constitution is the single most consequential provision of any Ethiopian constitution in modern history, measured by its political weight rather than by its frequency of formal invocation. It has been invoked successfully only once — by the Sidama people in 2019, and only to leave the SNNPR rather than the Ethiopian state. It has never been invoked successfully against the Ethiopian state, although the 2020 TPLF unilateral election was an attempted invocation in spirit if not in form.¹ Yet Article 39 has shaped Ethiopian constitutional politics more than any other provision because it is the textual confirmation of the foundational theoretical commitment of the 1995 Constitution: that the constituent sovereigns of Ethiopia are the nations, nationalities, and peoples, and that their right to leave is the ultimate guarantee of their constitutional standing.

The article’s argument is that Article 39 has done three things to Ethiopian constitutional politics: it has made the federation perpetually contingent (because dissolution is always constitutionally available to any constituent unit); it has made cross-ethnic political coalition-building structurally harder (because the constitutional framework rewards the consolidation of ethnic political identity rather than its dilution); and it has made the federal government’s responses to centrifugal pressure more securitised than they would otherwise be (because every regional challenge can be framed by the federal government, and is increasingly framed by opposition critics, as an existential question about state survival). These three consequences are not failures of implementation. They are direct consequences of the constitutional choice. They would follow from Article 39 even if Article 39 were never invoked.

The text

Article 39 of the 1995 Constitution reads in full:

(1) Every Nation, Nationality and People in Ethiopia has an unconditional right to self-determination, including the right to secession.

(2) Every Nation, Nationality and People in Ethiopia has the right to speak, to write and to develop its own language; to express, to develop and to promote its culture; and to preserve its history.

(3) Every Nation, Nationality and People in Ethiopia has the right to a full measure of self-government which includes the right to establish institutions of government in the territory that it inhabits and equitable representation in state and Federal governments.

(4) The right to self-determination, including secession, of every Nation, Nationality and People shall come into effect:

(a) When a demand for secession has been approved by a two-thirds majority of the members of the Legislative Council of the Nation, Nationality or People concerned;

(b) When the Federal Government has organized a referendum which must take place within three years from the time it received the concerned council’s decision for secession;

(c) When the demand for secession is supported by majority vote in the referendum;

(d) When the Federal Government will have transferred its powers to the council of the Nation, Nationality or People who has voted to secede; and

(e) When the division of assets is effected in a manner prescribed by law.

(5) A “Nation, Nationality or People” for the purposes of this Constitution, is a group of people who have or share a large measure of a common culture or similar customs, mutual intelligibility of language, belief in a common or related identities, a common psychological make-up, and who inhabit an identifiable, predominantly contiguous territory.²

Article 39 is one of the longest substantive provisions in the constitution. Its procedural specificity is notable: secession is not, in operational terms, “unconditional” in the sense of “without procedure.” There are five conditions (a through e in clause 4), and at every stage the federal government retains significant procedural control. A unit cannot, under the constitutional text, simply declare independence; it must follow the five-step process.

The word “unconditional” in clause 1 refers to the right itself: there are no substantive grounds the federal government can invoke to deny secession to a unit that follows the procedure. Other constitutions with secession provisions — Canada under the Clarity Act, after the Supreme Court of Canada’s 1998 Quebec reference, the Constitution of St. Kitts and Nevis — require substantive justifications and federal participation in negotiating terms. Article 39 does not. If the procedure is followed, the secession proceeds.

The intellectual origins

Article 39’s provisions originated in the TPLF’s reading of Lenin’s writings on the national question, particularly Lenin’s defence of the right of nations to self-determination including separation as a precondition for genuine voluntary union. The argument, in TPLF formulation, was that a federation can only be stable if its constituent units are free to leave, because only freely chosen union is stable union.³ The argument has theoretical weight. The constitutional commitment is the operationalisation of Leninist national-question theory in a way that other ostensibly Marxist-Leninist constitutions (Soviet, Yugoslav, Czechoslovak) had also attempted, with varying outcomes.

The Soviet precedent is instructive in both directions. The Soviet Constitution of 1936 and 1977 both contained an unconditional right of republics to leave the union (Article 17, then Article 72). For the seventy years between 1922 and 1991, the right was understood as constitutionally available but politically unusable. In December 1991, it was used. The republics left. The union dissolved. The TPLF leadership in 1991–95 knew this history; the inclusion of Article 39 was not naive about its implications. The drafters believed the political conditions for invocation would not arise if the federation was managed correctly. Whether that belief was correct is one of the questions this article addresses.

What it has produced

Three operational consequences of Article 39, all of which would have followed from the textual commitment regardless of frequency of formal invocation.

First, perpetual federation contingency. The fact that any nation/nationality/people can, by following the procedure, leave the federation means that the federation is constitutionally a voluntary association rather than a permanent state. This has consequences for political behaviour that do not require any party to actually invoke Article 39. Federal-state relations are negotiated against the implicit background of the secession option. Federal demands on regional governments are weighed by regional governments against the cost of intensifying the centrifugal pressure that might lead to invocation. Federal investments in regional infrastructure, federal budget transfers, federal protection of minority populations within regions — all of these operate in the shadow of Article 39, even when no one is talking about it.

The TPLF’s 2020 unilateral regional election — held in defiance of the federal postponement under House of Federation interpretation — was the closest any Ethiopian regional government has come to invoking the constitutional right of self-determination against the federal state.⁴ The TPLF’s framing was not formally an Article 39 invocation, because no two-thirds vote of the regional council for secession was passed and no referendum was sought. But the political logic was that the federal government’s actions had constitutionally invalidated its authority, and that the Tigray regional government’s continued authority was constitutionally grounded in the people of Tigray as a sovereign nation. This is the conceptual vocabulary Article 39 made available. It would not have been available in the absence of Article 39.

Second, structurally harder cross-ethnic political organisation. A constitution that vests sovereignty in nations/nationalities/peoples and grants each of them a right to secession creates incentives for political organisation along ethnic lines. The constituent unit of constitutional politics is the nationality. Parties organised by nationality have a constitutional standing — they can claim to speak for “their” people in a way the constitutional architecture validates. Parties organised across nationalities have no equivalent constitutional standing; they are competing in a field structured against them.

The empirical record reflects this. The major Ethiopian political parties since 1995 have, with one significant exception, been organised by nationality. The TPLF was Tigrayan. The OLF and the later OFC are Oromo. The AAPO, later the NaMA (National Movement of Amhara), is Amhara. The ONLF is Somali. The SEPDM (the EPRDF’s southern affiliate) was a coalition of southern nationality parties. The exception — the Coalition for Unity and Democracy (CUD/Kinijit) of 2005 — attempted to organise across nationality on a pan-Ethiopian liberal-democratic platform. It received massive electoral support, was repressed, and was destroyed. The lesson the political class drew from 2005 was that cross-ethnic organisation against the constitutional grain was not viable. The pattern has held: the Prosperity Party (2019–present) is, despite its formal pan-Ethiopian branding, an Oromo-led coalition with regional affiliates whose internal politics are organised by ethnicity.⁵

Third, securitised federal responses to regional challenge. When secession is constitutionally available, every regional challenge can be reframed by the federal government — and often by opposition critics — as a question about state survival. The framing changes the political logic of federal response. A challenge that, under a unitary constitution, would be treated as administrative noncompliance becomes, under Article 39, a potential prelude to constitutional dissolution. The federal response escalates accordingly.

This is visible in the federal government’s posture toward Tigray from 2018 onward. The TPLF’s resistance to the formation of the Prosperity Party in late 2019 was framed by Abiy Ahmed’s government as a threat to the federal order. The TPLF’s holding of regional elections in September 2020 was treated as constitutionally fatal — an act that, in Meles’s earlier framing, would have voided the constitutional status of the regional government. The November 2020 outbreak of military operations was justified by the federal government partly on this constitutional ground.⁶ The Abiy government’s framing has continued: the February 2026 House of Federation removal of five constituencies from Tigrayan oversight, the deregistration of the TPLF as a party, the post-April 2026 buildup along Tigray’s borders — each of these is presented as a constitutional necessity to prevent the dissolution that Article 39 makes constitutionally available.

The pattern is also visible in the Abiy government’s posture toward the Sidama referendum process in 2019, which produced the only successful Article 39 invocation in the constitution’s history. The federal government delayed the referendum process for over a year past the constitutional deadline, producing violent protests in Sidama in July 2019 that killed an estimated 25 people, before finally conceding the vote.⁷ The federal hesitation is most plausibly explained by concern about precedent: a successful Sidama invocation would, under the constitutional logic, make subsequent invocations easier. The Wolayta, Gurage, Hadiya, and other SNNPR groups have continued to press for separate statehood since.

The federal-centralist critique

The contemporary critique of Article 39 — coming from the Amhara nationalist tradition, from significant parts of the urban Ethiopian liberal intelligentsia, and from elements of the Prosperity Party leadership — holds that the constitutional right of secession is, in the words of one recent critic, “a poison pill for a multi-ethnic state” that “no stable nation can be built on a foundation that explicitly permits its own dissolution.”⁸ The argument is that Article 39 must be removed by constitutional amendment, that secession should be made either constitutionally impossible or substantively conditioned on demonstrated injustice by the federal state, and that the federation should be re-founded on the principle of individual citizenship rather than corporate nationality sovereignty.

This argument has force. The comparative evidence is on its side: most stable multi-ethnic federations do not contain secession clauses, and the federations that have contained them (Soviet, Yugoslav, Czechoslovak) all dissolved. Article 39’s defenders have to explain why the comparative pattern would not apply to Ethiopia.

The defenders’ reply is twofold. First, that the comparison is misleading because the Soviet, Yugoslav, and Czechoslovak federations contained secession provisions in constitutions that were not actually constraining — they dissolved when the central authority lost the will and capacity to suppress, not because of the secession clauses. The clauses were not the cause; they were available text. Second, that the Ethiopian federation’s stability depends, in part, on the constitutional reassurance to regional/nationality elites that their participation in the federation is voluntary. Remove that reassurance, and the centrifugal pressure that the federation has so far contained becomes much harder to contain. The defenders’ worry is that Article 39 repeal would, in operational terms, produce secession movements where the constitutional availability of secession had previously deferred them.⁹

Both arguments have weight. The question is empirical, and the empirical record is mixed. Article 39 has produced the SNNPR subdivisions (which are not, formally, secessions from Ethiopia, but follow the logic of corporate self-determination). It has not produced state dissolution. It has been adjacent to the Tigray crisis in ways that suggest it is part of the constitutional architecture that makes the crisis hard to resolve.

Steelman of opposing readings

The TPLF/federalist defence: that Article 39 was the constitutional bargain that enabled the post-Derg federation to form peacefully, that it has not produced state dissolution in thirty years, that the recent Tigray crisis was caused by Abiy’s violation of the constitutional federation rather than by the secession clause, and that removing Article 39 would betray the bargain on which the federation was constituted. This is partly right. Article 39 was, in 1991–95, the price of TPLF, OLF, and southern nationality agreement to participate in the federation. The bargain was real. The reply is that the bargain was made by a small coalition of armed movements without democratic mandate from the populations affected, and that the bargain’s costs have grown over thirty years while its benefits have been intermittent. Bargains can be renegotiated; the question is on what terms.

The Amhara/unionist critique: that Article 39 is the source of Ethiopia’s centrifugal pressure, that it has institutionalised the ethnic political logic that produces violence, and that removing it would not only be constitutionally correct but would substantively reduce the federation’s instability. This is partly right and partly wrong. It is partly right that Article 39 is part of the constitutional architecture that has produced the current crisis. It is partly wrong because the centrifugal pressures Article 39 expresses are not created by the article; they pre-existed it, and removing the article without addressing the underlying ethnic-political fragmentation would not, by itself, produce stability. The constitutional question and the political question are related but not identical.

The Abiy-Prosperity hybrid position: that Article 39 should not be repealed but should be made effectively unusable through procedural elaboration, federal control of the referendum process, and the development of a constitutional jurisprudence that requires substantive grounds for secession.¹⁰ This is, in operational terms, what the post-2020 federal government has done. The TPLF deregistration, the constituency removals, the budget pressures, the military buildup — these are all federal actions that, by making any invocation of Article 39 practically impossible, transform the constitutional text into a dead letter without formal repeal. This is constitutionally Schmittian: the executive decides what the constitutional text means in practice, regardless of what it says.

Multi-causal accounting

The political pathologies that Article 39 is associated with are not solely caused by Article 39. Ethiopia’s centrifugal pressures predate the 1995 Constitution; they were the political fact that produced the constitution rather than its consequence. The structural weakness of cross-ethnic political organisation reflects, in part, the very low level of mass-organised political life in Ethiopia generally, not just the constitutional incentives. The securitisation of federal responses reflects the institutional weakness of the federal civilian state and the strength of the federal military, not just the constitutional logic. Article 39 is one factor among several. Naming it precisely is what allows the others to be addressed in turn.

What the article establishes

Article 39 is the single constitutional provision whose theoretical and political weight has shaped Ethiopian constitutional politics most heavily. It has produced the federation’s perpetual contingency, has made cross-ethnic political organisation structurally harder, and has produced the securitised federal responses to regional challenge that have culminated in the Tigray war and the current crisis. These consequences are downstream of the constitutional choice. They would not have followed from a constitution that recognised diversity without elevating corporate ethnic identity to constitutional sovereignty. The constitutional choice was a choice. Its consequences are still being absorbed.

The next article takes up the other foundational provision of the 1995 Constitution: Article 40 on land, which has done for Ethiopian economic constitutional structure what Article 39 has done for political constitutional structure.

---

References

Footnotes

1. 1995 Constitution, Article 39; on the Sidama invocation, see 2019 Sidama Region referendum, https://en.wikipedia.org/wiki/2019_Sidama_Region_referendum. On the 2020 Tigray election, see “Ethiopian Politics - Election 2020,” Global Security, https://www.globalsecurity.org/military/world/ethiopia/politics-2020.htm. ↩

2. 1995 Constitution, Article 39, full text. ↩

3. On the TPLF’s reading of Leninist national-question theory, see Aregawi Berhe, A Political History of the Tigray People’s Liberation Front (Los Angeles: Tsehai, 2009); John Young, Peasant Revolution in Ethiopia, ch. 5. ↩

4. On the September 2020 Tigray election, see Freedom House Ethiopia 2021 report, https://freedomhouse.org/country/ethiopia/freedom-world/2021. ↩

5. For the Prosperity Party’s effective ethnic structure, see “Abiy Ahmed’s Hegemonistic Prosperity Party,” Borkena, June 2026. ↩

6. For Abiy’s framing of the Tigray crisis as constitutional, see his addresses to the House of Peoples’ Representatives, November 2020 and subsequently. ↩

7. On the Sidama referendum delay and 2019 violence, IFES, “Ethiopia Holds Referendum to Determine Statehood for Sidama Zone,” https://www.ifes.org/news/ethiopia-holds-referendum-determine-statehood-sidama-zone. ↩

8. “The Faultlines Of A Nation: Ethiopia’s Ethnic Federalism And The Need For Constitutional Reform,” The Reporter Ethiopia, November 2025, https://www.thereporterethiopia.com/47667. ↩

9. This is broadly the position of EPRDF/TPLF-aligned constitutional scholars, e.g., Andreas Eshete and Tsegaye Regassa. ↩

10. On the contemporary Abiy government strategy toward Article 39 and the federation, see “Ethiopia’s endless constitutional chasm and authoritarian abyss,” Ethiopia Insight, September 2024, https://www.ethiopia-insight.com/2024/09/13/ethiopias-endless-constitutional-chasm-and-authoritarian-abyss/. ↩