The Hollow Pillar: Selective Implementation, Constitutional Illiteracy, and Interpretation by Connection

Thesis

The Ethiopian constitution is, in its 1995 text, an impressive document. It contains the most elaborate rights chapter in any African constitution. It establishes a federation with substantial regional autonomy. It enumerates structural protections — judicial independence, parliamentary supremacy in lawmaking, electoral mandates — that, on their face, would constrain executive power as comprehensively as any constitutional design in the world. Reading the text in 2026, more than three decades after its ratification, what is striking is how complete it is. Almost nothing that a well-designed modern constitution should contain is missing from it.

What is missing is the connection between the text and the citizens it claims to govern. Implementation of the constitution has been, since 1995, selective: present where the regime permitted it, absent where it would have constrained the regime. Knowledge of the constitution among the Ethiopian population has been, throughout, shallow: most Ethiopians cannot name the rights the document confers on them, much less the procedures for vindicating them against the state. And interpretation of the constitution has been, in operation, conditional on political relationship: what the constitution means in a particular case depends on the constitutional subject’s standing with the ruling party, with the local administration, with the security services. The text is the same for everyone. The constitution is not.

This is the hollow pillar. A pillar of state is structurally meaningful only if it bears load. The Ethiopian constitutional pillar, looked at architecturally, appears load-bearing: it has the dimensions, the materials, the placement. Looked at structurally, it bears no load. The state is held up by other forces — the army, the ruling party, the security apparatus, the international donor consensus on Ethiopia as developmental partner. The constitution is decorative.

The article’s argument is that this hollow-pillar pattern is the operational mechanism by which a textually impressive constitution has failed, for thirty years, to produce constitutional substance. The three mechanisms — selective implementation, constitutional illiteracy, interpretation by connection — are interlocking, mutually reinforcing, and each has been allowed to develop because the constitutional architecture did not include the institutional features that would have addressed it. This is not a critique that ethnic federalism is the wrong design. It is a critique that operates regardless of design choice. Any constitution that is selectively implemented, illegible to its citizens, and interpreted by political affiliation rather than by rule, is a hollow pillar — regardless of whether it organises the federation around nationality, region, or individual citizenship. The constitutional pathology of the 1995 settlement is therefore not entirely the pathology of its design. Most of it is the pathology of its operation.

Selective implementation

The 1995 Constitution is implemented selectively. The selection is not random; it is patterned. The provisions that have been implemented systematically are those that serve the ruling party’s political logic at any given moment. The provisions that have been implemented inconsistently, or not at all, are those that would constrain the ruling party’s political logic.

Concrete examples populate the constitutional record.

Article 9 (constitutional supremacy) declares the constitution the supreme law of the land and provides that “any law, customary practice or a decision of an organ of state or a public official which contravenes this Constitution shall be of no effect.” In the constitution’s thirty-one years of operation, no federal statute, executive proclamation, or administrative regulation has been struck down by any institution on the ground of contravention of this Article.¹ The Council of Constitutional Inquiry, which screens constitutional questions before the House of Federation, has never recommended striking down a federal statute. The House of Federation has never struck one down. The supremacy clause is in force on paper. In operation, it has produced no observable constraint on federal legislation.

Article 18 (freedom from inhuman treatment) prohibits torture, cruel, inhuman, or degrading treatment. Documented violations of this provision by Ethiopian security services have been continuous since 1995. Human Rights Watch, Amnesty International, and the Ethiopian Human Rights Commission have all documented cases of torture in federal and regional detention facilities through 2025.² The provision has been implemented when it suited regime convenience (the closure of Maekelawi detention centre in 2018, briefly celebrated as a constitutional moment) and not implemented when it did not (the wave of detentions and reported abuses during the Tigray war, the Amhara state of emergency, the OLA-Oromia counterinsurgency).

Article 29 (right to hold opinions and freedom of expression) has been honoured in some periods (briefly, 2018–2019) and violated in others. Proclamation No. 590/2008 (Press) and Proclamation No. 1208/2020 (Hate Speech and Disinformation) have introduced restrictions on press freedom that the constitution’s text would suggest are unconstitutional but that the House of Federation has not addressed.³

Articles 38–40 (rights of nations/nationalities/peoples, including land rights) have been implemented vigorously when the implementation served the regime’s political needs (Sidama 2020, SNNPR breakup 2021–2023) and not implemented when it would have constrained the regime (TPLF’s claim to constitutional standing for its 2020 election, Tigray’s claims to disputed territories).

The pattern is not that the constitution is sometimes followed and sometimes ignored at random. The pattern is that the constitution is followed when it produces outcomes the regime prefers and not followed when it would produce outcomes the regime opposes. This is not constitutional implementation; it is executive discretion exercised under constitutional cover. The constitution is, in operational terms, the executive’s tool rather than the executive’s constraint.

Constitutional illiteracy

The second mechanism is the absence of broad popular knowledge of the constitution. Most Ethiopians do not know the rights they hold under the 1995 Constitution. They do not know the procedures for vindicating those rights. They do not know which institution they can approach when a right has been violated. They do not know which provisions apply to their situation. The constitution, in its full text, has not been incorporated into Ethiopian civic education at any sustained scale, has not been the subject of significant public-information campaigns, and has not been internalised by the population as the framework that defines the relationship between citizen and state.

This is empirically documented in modest ways. The Afrobarometer surveys have, in various rounds, asked Ethiopians questions about their rights, the role of institutions, and their understanding of constitutional structure. The results consistently show low awareness of constitutional content beyond very general features (Ethiopia is federal, the country is multi-ethnic, there is a parliament).⁴ The depth of awareness about specific rights, specific procedures, and specific institutions is much shallower.

The reasons for this constitutional illiteracy are several. The constitution is written in formal Amharic that is not easily accessible to non-elite readers. Civic education in Ethiopian schools has been thin throughout the 1995–2026 period; the curriculum has covered the constitution in summary but not in operational depth.⁵ Public legal aid is limited; most Ethiopians have no recourse to professional legal advice when their rights are at stake. The Ethiopian Human Rights Commission, established under Article 55(14) of the constitution to monitor rights compliance, has been chronically under-resourced and, until significant reforms in 2018–19, was not perceived as independent of the executive.⁶ The civil society sector, which in other constitutional democracies plays a major role in disseminating constitutional knowledge, has operated under severe restrictions for most of the post-2005 period.

The consequence is that the constitution has not become the framework within which Ethiopian citizens understand their relationship to the state. The state is not understood, by most Ethiopians, as bound by a set of rules that the citizen can invoke; it is understood as an authority whose decisions can be appealed only through the political and administrative hierarchy. The constitutional pillar, having no recognition from the citizens it is supposed to support, cannot bear their weight.

Interpretation by connection

The third mechanism is the most operationally consequential. The constitution, where it is implemented, is implemented differently depending on who is being asked about it. The same constitutional provision, applied to a person with strong political connection to the ruling party, produces a different outcome than the same provision applied to a person without such connection. This is not a description of corruption in the narrow sense (although corruption is one of its forms). It is a description of how constitutional interpretation works in practice in a system where the institutions of interpretation are themselves politically dependent.

The mechanism operates through every level of administrative and judicial decision-making.

At the kebele and woreda level, a citizen seeking to register land use, contest a tax assessment, obtain a business permit, or appeal an administrative decision will receive different treatment depending on her or his relationship with the local officials and with the local ruling-party structure. The treatment is not necessarily corrupt in the bribe-paying sense; it can be perfectly procedurally regular while producing entirely different outcomes depending on the applicant’s political standing.

At the regional and federal level, the same pattern operates. A businessperson aligned with regional ruling-party structures will receive different treatment from the regional government than a businessperson not so aligned. Access to credit through the state-owned Commercial Bank of Ethiopia, allocation of land leases, awarding of government contracts, regulatory enforcement — all of these operate through administrative discretion that is, in practice, exercised differently depending on political relationship.⁷

At the judicial level, the documented record shows that politically sensitive cases are decided differently from non-politically-sensitive cases. The 2007 trials of CUD leaders for treason, the post-2018 federal cases against Eskinder Nega and other journalists, the 2009–10 trials of OFDM and UDJ figures, the post-Tigray-war prosecutions, the Amhara state-of-emergency detentions — these are not the cases the official Ethiopian judicial statistics report on, but they are the cases that determine what the constitution means for citizens who challenge the regime. The judicial system as it operates in non-political cases (commercial disputes, civil matters, ordinary criminal cases) may operate with reasonable professionalism; the same system, applied to political cases, operates differently.⁸

The pattern is Schmittian in the same sense the broader constitutional pattern is Schmittian. The constitution, in operation, means whatever the executive needs it to mean in the particular case. The standards by which constitutional interpretation should operate — predictability, equal application, judicial independence — are formally affirmed by the text and operationally violated by the practice.

Why the hollow pillar persists

The three mechanisms — selective implementation, constitutional illiteracy, interpretation by connection — are mutually reinforcing.

Selective implementation by the executive is enabled by constitutional illiteracy among the citizenry, because citizens who do not know what the constitution requires cannot demand its enforcement. Constitutional illiteracy is reinforced by selective implementation, because citizens who observe that the constitution is selectively applied draw the rational inference that learning its provisions is not worth the effort. Interpretation by connection is enabled by both — by the executive’s freedom to interpret selectively, and by the citizens’ inability to challenge interpretation through informed claim. Each mechanism deepens the others.

What is missing from the constitutional architecture, and would have addressed the hollow-pillar pattern, is what Madison identified: institutional ambition counteracting institutional ambition. A constitutional court with the standing to strike down legislation would address selective implementation. A serious civic education curriculum, supported by independent civil society and media, would address constitutional illiteracy. An independent judiciary with security of tenure and budget independent of executive control would address interpretation by connection. These features are not unknown — they are the standard kit of working constitutional democracies. They were not built into the 1995 Constitution, and they have not been built since.

This is not because the framers did not know about them. The Constitutional Commission consulted comparators. The decision not to have a constitutional court was deliberate (taken up in Article 13 of this series). The decision not to invest in civic education was a budget decision made repeatedly across administrations. The decision not to constitute a judiciary independent of executive control was a decision about appointments and resources made across thirty years of governance. The hollow pillar persists because the design and the operational choices that would have made it solid were not made, and the political-economy conditions for making them have not arisen.

Madison, Hayek, Acemoglu-Robinson

Each of the three framework anchors of this series sees something specific about the hollow-pillar pattern.

Madison sees the absence of ambition counteracting ambition. The text contains the formal separations; the operation does not contain the institutional motives. Each branch, each level of government, each office is dependent on the executive in ways that make resistance professionally costly. The Madisonian remedy — institutional design that makes the branches’ ambitions diverge — has not been implemented. The federal-state ambition divergence, which the 1995 design was supposed to provide through ethnic federalism, has not produced the Madisonian effect because state-level institutions have been politically aligned with the federal ruling party throughout most of the period.

Hayek sees the absence of the rule of law. Constitutional provisions that mean different things to different citizens depending on political affiliation are not law in the Hayekian sense; they are discretion exercised under legal vocabulary. The Ethiopian state’s interaction with its citizens is not rule-governed in the way a constitutional democracy’s interaction with its citizens should be.⁹ The Hayekian remedy — predictable general rules equally applied — requires institutional infrastructure that has not been built.

Acemoglu and Robinson see the absence of inclusive institutions. The constitutional pattern is extractive in the sense that institutions distribute power narrowly (to executive, ruling party, security services) rather than broadly, and entrench incumbents rather than constraining them. The pattern reproduces because the political coalition that would benefit from a switch to inclusive institutions has not formed.¹⁰ The Acemoglu-Robinson remedy — sustained pressure from a broad social coalition demanding inclusion — has not materialised, partly because Article 40 has foreclosed the formation of one of its most important social bases, partly because the constitutional architecture has incentivised political organisation along ethnic lines that cut against cross-ethnic coalition.

Steelman of the opposing reading

Two opposing readings deserve engagement.

The first is that the description of selective implementation, constitutional illiteracy, and interpretation by connection is true of most constitutions in most countries — that the hollow-pillar pattern is not specific to Ethiopia but is the universal condition of constitutional law everywhere, and that Ethiopia is being held to a standard no constitution actually meets. This is partly right and partly wrong. It is partly right that every constitution is selectively implemented to some degree, that constitutional illiteracy is widespread even in mature democracies, and that political connection affects legal outcomes everywhere. It is partly wrong because the degree matters. The U.S. Supreme Court strikes down acts of Congress on constitutional grounds with some regularity. The Indian Supreme Court has, since 1950, developed a robust jurisprudence of fundamental rights. The South African Constitutional Court has invalidated significant legislation. These institutions operate imperfectly, but they operate. Ethiopia’s House of Federation has not, in three decades, operated comparably. The difference of degree is, at some point, a difference of kind.

The second is that the diagnosis is correct but the remedy is overstated. To make a constitution function as a hollow pillar fully addressed would require institutional transformations that take generations, not decades. The 1995 Constitution has, the argument runs, made progress in some domains (language rights, electoral conduct in some elections, federalism’s basic operation) and should be judged against a realistic timeline of institutional development, not against an idealised template.¹¹ This is partly right. Institutional development is slow. The reply is that other countries have made comparable transitions faster (South Africa from 1996, Indonesia from 1998, India from 1950) and that the slowness of Ethiopian institutional development is not a fact of nature but a political-economy fact about which actors have chosen what to invest in. The slow timeline is not exogenous; it is a consequence of choices.

Multi-causal accounting

The hollow-pillar pattern is not solely a consequence of constitutional design. It reflects the inherited weakness of independent institutions from the imperial and Derg periods, the structural problems of the agrarian economy that limit the social bases for institutional demand, the regional security environment that has produced repeated crises absorbing state attention, the international donor consensus that has rewarded developmental outcomes over institutional ones, and the personal choices of successive leaders. Each is a contributing cause. The constitutional choice — to build a text without building the institutional infrastructure that would have made the text effective — is one cause among several. Naming it precisely is what allows the others to be addressed without it absorbing all the explanatory weight.

What the article establishes

The Ethiopian constitution functions as a hollow pillar — architecturally complete, structurally non-load-bearing — because three mechanisms operate: selective implementation by the executive, constitutional illiteracy among the citizenry, and interpretation conditional on political connection. These mechanisms are interlocking and mutually reinforcing. The constitutional architecture did not include the institutional features that would have addressed them. The architecture has not been retrofitted to include those features. Until the architecture is retrofitted, or rebuilt, the pillar will remain hollow, and the country will continue to be held up by the forces that have actually held it up for thirty-one years — the army, the ruling party, the security apparatus, and the international donor consensus. The constitution will be a description of the state. The state will not be bound by it.

The next article takes up the constitutional institution that, in any working constitutional democracy, would address the hollow-pillar pattern: the constitutional court. Ethiopia does not have one. The House of Federation has the constitutional-review authority instead, and that choice — examined as a choice with consequences — is the subject of Article 13.

---

References

Footnotes

1. This factual claim is supported by reviews of House of Federation and Council of Constitutional Inquiry decisions in the literature, e.g., Getachew Assefa, “All About Words: Discovering the Intention of the Makers of the Ethiopian Constitution on the Scope and Meaning of Constitutional Interpretation,” Journal of Ethiopian Law 24, no. 2 (2010); Adem Kassie Abebe, “From the ‘TPLF Constitution’ to the ‘Constitution of the People of Ethiopia’: Constitutionalism and Proposals for Constitutional Reform,” International Journal of Constitutional Law 21, no. 3 (2023). ↩

2. Human Rights Watch, World Report 2026: Ethiopia; Amnesty International, Ethiopia 2025: Annual Report; Ethiopian Human Rights Commission annual reports. ↩

3. Proclamation No. 590/2008 (Press); Proclamation No. 1208/2020 (Hate Speech and Disinformation). See “Ethiopia: Proposed Legal Changes Threaten Civil Society,” Human Rights Watch, July 2025, https://www.hrw.org/news/2025/07/29/ethiopia-proposed-legal-changes-threaten-civil-society. ↩

4. Afrobarometer Round 9 (2023), Ethiopia: constitutional knowledge questions show limited depth. ↩

5. On civic education and constitutional literacy in Ethiopia, see Tsegaye Regassa’s writings on constitutional culture; Berihun Adugna Gebeye, Constitutionalism in Ethiopia (Edward Elgar, 2021). ↩

6. Ethiopian Human Rights Commission, see annual reports; on the 2018–19 reforms, see Daniel Bekele, “Ethiopia: Civil Society and Rights Reforms,” speeches 2019–2020. ↩

7. For documented patterns of administrative discretion exercised politically, see Sarah Vaughan and Mesfin Gebremichael, Rethinking Business and Politics in Ethiopia: The Role of EFFORT, the Endowment Fund for the Rehabilitation of Tigray (London: Africa Power and Politics Programme, 2011); Berihun Adugna Gebeye, Constitutionalism in Ethiopia. ↩

8. On the political-judicial differential, see Getachew Assefa, “All About Words”; Adem Kassie Abebe, various writings; “Ethiopia’s endless constitutional chasm and authoritarian abyss,” Ethiopia Insight, September 2024. ↩

9. F.A. Hayek, The Constitution of Liberty, chs. 10–14; Law, Legislation, and Liberty, vol. 1. ↩

10. Daron Acemoglu and James A. Robinson, Why Nations Fail, chs. 3, 4, 12. ↩

11. This is broadly the position of Tsegaye Regassa and the moderate constitutionalists; see Tsegaye Regassa, “Sub-National Constitutions in Ethiopia,” Mizan Law Review 3, no. 1 (2009). ↩