The Necessary Orientation
France has one of the most capable states in the world. Mexico’s finance ministry and South Africa’s Reserve Bank each dwarf anything Rwanda possessed when Kagame’s government took office in 1994. None of them has transformed their productive structure. Rwanda has. Twenty Years ago, surveying fifty years of development history, the Commission on Growth and Development found only thirteen countries that did. That question — why the number is so small, and what is missing before the capability question is even asked — is where this analysis begins.
Economic transformation is not more of the same. A generational quintupling of GDP, as experienced by those thirteen countries, involves people doing genuinely different things. Whole industries displace others; occupational structures shift; the productive base changes in kind rather than in scale alone — and so, too, does the distribution of power within it. That is what makes transformation rare: not the difficulty of executing familiar policies better, but the difficulty of committing to a future that is structurally unlike the present.
Transformation requires a governing coalition committed to a future structurally unlike the present — if only implicitly. When that commitment is absent, the policies and institutions that transformation requires will not be built, not because capability is missing, but because what the destination entails is not wanted. WGI Government Effectiveness data makes the hierarchy plain: France near the top of the global distribution; South Africa and Mexico in the middle, at substantially higher scores than Rwanda, even at lower income levels. Rwanda was still below both as recently as the 2010s, rebuilding from near zero after 1994. That capability hierarchy predicts the wrong direction. Capability is not irrelevant. It matters as transformation proceeds. But it is neither a necessary nor a sufficient starting condition. The objection that capability was built in Rwanda during the transformation, not before, doesn't change this: where orientation is present and sustained, the strategic demands of the commitment build the capability it requires.
Stefan Dercon named it in Gambling on Development (2022). He argued that the binding variable is not capability or policy quality, but a governing elite’s willingness to prioritise productive capacity over short-term extraction. What that bargain establishes is a governing orientation: a system pointed toward structural transformation as its organising purpose. When it is not — as Tanzania’s successive donor-funded civil service reform programmes, running from the late 1990s through the mid-2010s, demonstrate — the downstream investments enter a system whose own dynamics absorb them and restore the same equilibrium. Not because implementation failed. Because the governing purpose was never transformation. Four observable characteristics in the final section of this post build on Dercon’s indicators and systematise them, enabling assessment of orientation before outcomes are known.
That orientation can sit anywhere on a spectrum from deep state direction to radical market liberalisation. Four cases show what it looks like in practice: two where the bargain was made, in contrasting modes, and two where it was not.
The Bargain Made: Direction Varies, Commitment Doesn’t
Rwanda is transformation-oriented in the directive mode; Estonia, in the liberalising mode. The most telling evidence for Rwanda is not the budget figures but what happens when members of the ruling coalition fail to perform.
Since 2006, district mayors and ministers have publicly signed imihigo performance contracts with the president, with results ranked and published annually, with consequences. In 2014, Gasabo's Willy Ndizeye and Gatsibo's Ambroise Ruboneza resigned, and Kirehe's Protais Murayire was relieved of his duties, all following poor imihigo rankings; Ruboneza named his district's last-place finish as the reason. This is genuinely unusual. Most systems that claim to condition support on performance don’t, when the moment is real. Rwanda has done it repeatedly, on the record, in public. Its reporting system is not always reliable (the same incentive that conditions support on performance creates pressure to over-report achievement) but the dismissal events are datable and on the record, not derived from the administrative data. Rwanda’s governing coalition withdrew support from members who failed to deliver. That is what makes Rwanda’s configuration distinctive, not the ambition of the plans but the governing environment they produced: one where underperformance was eliminated rather than absorbed.
Its appointment record reflects the same logic (technically competent leadership at the revenue authority, maintained, the same pattern extending across the Rwanda Development Board and second-tier operating agencies), and the budget confirms it. Public investment has averaged about 11 per cent of GDP, above the sub-Saharan Africa norm, sustained with unusual consistency. Rwanda’s capital programmes have been maintained through external shocks, aid suspensions, and periods of regional tension, and committed expenditure has been maintained when it would have been easier to raid it. Announced and actual capital expenditure remain consistently close by regional standards, setting Rwanda apart from peers including Kenya and Tanzania, where governments routinely sacrifice capital budgets under fiscal stress.
Tim Kelsall names this: neopatrimonial governance — systems in which personal authority and formal institutions coexist, with the former typically dominating — organised to produce developmental rather than predatory outcomes. The orientation assessment applies to Rwanda’s productive governing logic; the political character of that system is a distinct question addressed in the Technical Notes. This rarity is not confined to any one governing mode.
Estonia under Laar was transformation-oriented in the liberalising mode.
Mart Laar took office in 1992 at 32. Laar’s governing purpose was stated plainly: the Soviet-era productive structure was not to be reformed but dismantled. A currency board pegging the kroon to the Deutsche Mark removed monetary discretion entirely. A flat tax introduced in January 1994 — 26 per cent — eliminated the complexity that had served as both a barrier to entry and a source of bureaucratic discretion. Laar has said he had read almost no economics before taking office beyond Friedman’s Free to Choose. Estonia’s direction under Laar was simple, explicit, and oriented toward a future the existing system could not produce. Not building state capacity but dismantling state capture. That architecture held: the flat tax survived Laar’s own defeat in 1994 and remained in place through every subsequent government. Rwanda’s governing orientation and Estonia’s are the same. Only the instruments differ.
Rwanda has been building what doesn’t yet exist; Estonia removed what blocked it. These are not opposing positions in the state-versus-market debate. Both instruments point toward the same destination: the governing coalition's acceptance that the productive structure will look fundamentally different in a generation, and that its purpose is to make that possible. It is that commitment, not the instrument, that distinguishes transformation-oriented governing from the performance of it.
The Bargain Absent: Kuwait Opted Out, Tanzania Performed It
Transformation orientation can be absent in more than one way: coherent governing toward a different purpose, or performing the signals while the orientation points elsewhere. Kuwait is straightforward. Tanzania under Magufuli is not.
Kuwait is not transformation-oriented, and the explanation is structural rather than a failure of will or capability. The Kuwait Investment Authority manages the sovereign wealth fund with genuine professional competence, which is precisely the point: that capacity concentrates in the institution that stewards the endowment, not in ministries charged with economic diversification.
Kuwait’s budget allocates about 80 per cent of expenditure to wages, subsidies, transfers, and the mandatory annual allocation to the Future Generations Reserve Fund. Capital investment in productive diversification is modest relative to GDP and tightly correlated with oil revenue fluctuations. Its National Assembly primarily negotiates the allocation of oil rents to constituencies. Oil income removes the fiscal pressure that elsewhere compels governments to ask whether the productive structure is inadequate. The 2014–16 oil price collapse applied the one pressure that could have forced a response, and revealed which adaptation the system had actually made. When Brent fell from approximately US$110 to US$30 — stripping roughly US$15 billion of annual revenue — the government's expenditure response was a 1.5 per cent pullback. No wage-bill reform, no VAT, no subsidy restructuring for nationals. Coalition preservation was the response; the productive structure question was not asked.
Tanzania under Magufuli was not transformation-oriented. It is the hardest of the four to call, because the outward signals seem to point the other way.
Magufuli governed with obvious energy. Capital expenditure accelerated; fiscal discipline tightened measurably in the first years; executive attention concentrated on infrastructure delivery with a directness uncommon in the region. His Julius Nyerere hydropower project (2,115 MW) and the Standard Gauge Railway extension were announced with deadlines and tracked publicly. Public dismissals of underperforming civil servants were broadcast on state television. A diagnostic framework applied in 2017 would have had to look past what was visible to see what was absent, which is precisely what makes it a harder case.
What proved decisive was whether professional appointees were trusted to act on their judgment. This failure mode is more precise than loyalty screening. Some key economic positions retained career technocrats throughout the Magufuli years. Decisive evidence lies not in who was appointed but in what happened when professional capacity was exercised.
In April 2020, Deputy Health Minister Faustine Ndugulile, a microbiologist, publicly warned that the steam-inhalation treatment the president was promoting as a COVID remedy could obstruct the respiratory system. He was sacked the following month. The national laboratory director was suspended the same month. A ten-member expert committee formed to assess the evidence never published its findings. Tanzania submitted no official COVID data to the WHO for the following fourteen months. No professional resistance emerged: not because the layer was absent, but because exemplary dismissals had converted it to silence. The system had found its equilibrium, and professional judgment was not part of it.
A professional apparatus converted to compliance cannot deliver sustained reform, whatever the energy at the top. Confirmation came from the other direction. Under President Hassan, dismissed officials were reinstated, and an IMF credit facility was approved; the professional layer was reactivated. A professional layer that had never existed could not have been reactivated. What Magufuli had suppressed was functional. What he had built, the performance of transformation-oriented governing, was not. The political trajectory under Hassan is a separate matter, covered in the Technical Notes .
Assessing Orientation Before Outcomes Are Known
Whether governing orientation can be recognised before outcomes arrive is the question that determines whether the concept earns its keep. Looking back at successful cases, the answer seems straightforward: Rwanda transformed, so it was transformation-oriented; Tanzania did not, so it was not. But a concept that sorts cases only after their trajectories are known does little analytical work; it merely redescribes what outcomes already tell us. What gives it genuine force is whether orientation can be recognised at the time, during a government’s early years, before outcomes have confirmed or denied it. It becomes testable in the stronger sense: assessed independently of the outcome it is supposed to explain. And it becomes something development partners can account for before committing to engagement, rather than discovering the mismatch after fifteen years of programming pointed at the wrong variable.
Dercon points to several intuitive signals: whether technocrats have genuine influence and whether the governing coalition takes economic performance seriously. Both signals are sound, but neither is specified precisely enough to be assessed independently before trajectories are established. Standard governance indices measure effectiveness and quality; none distinguishes the productive direction of that commitment. The four characteristics below address that gap, drawn from administrative records that exist in real time, making the bargain assessable in year two, not year fifteen.
Four characteristics reveal the governing orientation: what the system actually rewards and what it eliminates. The first is what public budgets actually fund across election cycles, not what is announced but what is committed and executed, and whether it holds when fiscal pressure makes raiding the easier political move. Budget allocation is the most directly appraisable of the four: the data exists in executed versus announced expenditure figures, and the comparison holds across cycles without requiring interpretation of intent. The second is who holds actual governing authority over the productive agenda: whether technical competence concentrates in the institutions that matter, or whether professional appointments are nominal while real authority sits elsewhere. Of the four, this dimension is the hardest to fake: a government seeking to perform the signal without the substance would simply avoid technical appointments, but that choice is itself readable from the appointment record. The third is where executive attention demonstrably goes, not stated priorities, but measurable concentration of political capital and personal engagement. Executive attention is also the least systematically observable of the characteristics: the traces (delegation patterns, where political capital is spent) resist the comparability that budget lines permit. The fourth is whether the orientation is being translated into an active structural reform programme: concrete initiatives that condition private allocation on performance or open competitive access to new actors, not announced ambitions.
None of these is a clean solo indicator. Rwanda’s budget held through the 2012 aid suspension; the same orientation was readable in the appointment and dismissal records from 2006. What matters is the pattern across all four, and whether it holds under fiscal stress, when abandoning it is the path of least resistance.
Political settlement type constrains the available range of productive choices — it is the underlying configuration of power among elites that determines what governing coalitions can credibly commit to. A fragmented settlement makes long-horizon commitment harder to sustain; a dominant-party settlement can sustain it but can just as readily entrench extraction. The constraint matters. But it is not determination. Settlements of identical structural type can orient toward different productive destinations, because within any structural configuration, there exists a choice set. Rwanda and Tanzania, under Magufuli, both operated within dominant-party settlement structures; one oriented toward structural transformation, the other toward performing it. Settlement type tells you what a governing coalition can sustain; the four characteristics tell you what it is actually trying to do.
Rwanda maintained its capital programme through the 2012 aid suspension; Tanzania raided comparable allocations under fiscal stress. Rwanda, Estonia, Kuwait, and Tanzania under Magufuli are not points on a quality spectrum. The first two are transformation-oriented through opposite instruments. Kuwait is a coherent distribution logic aimed at a different objective. Tanzania demonstrated the performance of transformation without the governing orientation that would give it substance.
The reason only thirteen countries passed the Commission for Growth’s test for delivering transformation is not that the capability was absent; it is that the orientation that made transformation a governing purpose was. Fifty years of development programming that misses this engages with an ecosystem that is not seeking to use it. The programme succeeds at what it set out to do and leaves the underlying equilibrium untouched.
In the current era, how many countries actually show such a transformative orientation?
That is the question the next post addresses.
Further Reading
Stefan Dercon, Gambling on Development (2022) — the development bargain as the binding variable in whether transformation-oriented governing forms and holds; Dercon’s core argument that such bargains are often fragile, and that it is the rare ones that survive political transitions that prove truly transformative
Mushtaq Khan, “Political Settlements and the Governance of Growth-Enhancing Institutions” (2010) — the political settlements framework from which this post’s orientation concept partly derives; Khan’s rent-management typology provides theoretical grounding for distinguishing developmental from predatory governing configurations
Mick Moore, “Political Underdevelopment” (2001) — why accountability structures track how the state extracts revenue from its own citizens, not rhetorical commitment to accountability; rent-sustained states face structurally weaker pressure to build the productive base that generates taxable income
Tim Kelsall, Business, Politics and the State in Africa (2013) — developmental patrimonialism: the conditions under which neopatrimonial governance produces developmental rather than predatory outcomes; the comparative basis for understanding why Rwanda’s configuration is the exception rather than the pattern
Brian Levy, Working with the Grain (2014) — a complementary diagnostic framework; Levy’s configuration typology (competitive-clientelism, dominant-party, developmental) classifies governing arrangements by political structure, where this post’s observable dimensions read orientation from behaviour; the two approaches are complementary rather than duplicative
Sam Hickey, Badru Bukenya et al., ESID (Effective States and Inclusive Development) research programme (2015–2023) — the current empirical literature on political settlements and development outcomes across Africa; ESID’s cross-country findings on Rwanda, Tanzania, Uganda, and Ghana situate the cases in this post within the broader political settlements evidence base
For more details see the Technical Notes for this post.