Built Against the Grain

Developmental governing orientation has to be constructed

Apr 28, 2026

Many governing coalitions have developmental intent. Few build the architecture that converts it into orientation. This post explains why — and what the configurations that do manage it actually look like.

Kenya competes at the top of the global flower market, supplying roughly 38 percent of the European Union’s cut roses. Kenya’s agrochemicals are imported. Its packaging is imported. Its cold chain equipment is imported. A sector that is world-class in one domain has barely touched the surrounding economy.

That gap is not a Kenyan peculiarity. Ghana earns billions from cocoa exports; the domestic processing industry barely exists. M-Pesa handles tens of millions of daily transactions; it has compounded into consumption credit, not productive investment. In each case, a governing coalition delivered growth within one or two domains and little beyond. The question is why the developmental commitment stopped at the sector boundary, and why, in a small number of cases, it didn’t.

That answer lies in what holds governing coalitions together: not ideology, but the structured distribution of power and resources among those capable of threatening stability.1 No formal agreement is required: only the ongoing calculation that the existing arrangement beats the cost of disrupting it.2 Those who benefit from the existing distribution accumulate the political resources that make challenging it harder — a mutual accommodation becomes, across repeated rounds, a structure that organises itself against disruption.3 The question is whether a governing coalition can direct that equilibrium toward productive compounding, or whether the settlement’s logic will always contain growth within the domains it already rewards.

Developmental intent is not the scarce resource. What is scarce is the commitment that crosses domain boundaries and compounds, and the settlement architecture that carries it. Most governing coalitions hold neither.

Where Kenya’s Flowers Stopped

From the 1970s, Kenya’s flower sector attracted the political-business interests of Daniel Arap Moi’s coalition. Access to cheap land and capital made floriculture a natural vehicle for those accumulating political resources. Operational management was left to people who knew the business; European buyer relationships were maintained, production standards held, and the sector grew.

Kenya became the world’s fourth-largest exporter of cut flowers and the largest supplier of roses to the EU. What donors consistently narrated as a “private-sector success story” looks different through the lens of political settlement analysis: it shows why the growth took the particular shape it did.4

Kenya’s political economy was constituted around a specific bargain: state support (export licensing, land security, market-access infrastructure) delivered to large-scale producers, underwritten by the political-business interests of the Moi coalition. That bargain delivered for the export coalition. It did not extend to the adjacent domains where compounding could occur. Kenya’s floriculture settlement had no equivalent in those adjacent domains — not because incumbents in those domains blocked it, but because the distributional bargain had simply never reached them. The sector boundary was where the settlement’s rewards ended. The sector’s expansion compounded into import demand rather than a domestic supply chain.

Where the settlement blocks

Developmental transformation is structurally different: it requires labour and investment to move from low- to high-productivity activities in a way that compounds.5 Productive capabilities are combinatorial: each new domain that forms reduces the cost of adjacent ones, and the returns from earlier stages become the inputs that make later stages possible.6 A distributional settlement does not block all growth. It blocks at the points where growth in one sector should compound into the next — precisely where structural transformation happens.

Ghana’s cocoa sector applied the same logic to a different crop: an agricultural policy settlement that delivered for export commodity interests without spilling over into processing.7

M-Pesa processes tens of millions of daily transactions, yet a 2021 survey of twenty-five leading Kenyan fintech providers found that only five had significant focus on micro-enterprise lending, commanding a combined market share of roughly two percent in the MSE credit market.8 The capability was present. The demand was present. M-Pesa is a world-class capability deployed within a settlement that had no reason to point it toward productive lending. Not a failure of technocratic capacity, but a failure of governing orientation. And the settlement had no logic that would produce one.

Consolidated Authority: Accountability in Two Directions

Most coalitions that hold developmental intent never build the architecture to carry it. Two configurations do. The first is consolidated authority with accountability running in two directions. What separates Korea under Park from Mobutu’s Zaire? Both held consolidated authority. What Korea had, and Mobutu’s Zaire did not, was accountability running in two directions: the coalition holding the leader answerable for aggregate developmental outcomes, and the leader holding firms accountable for productive results.

Korea’s Park held monthly export promotion meetings, reviewed firm targets, and let Shinjin Motor Company fail when it couldn’t meet them.9 In the configurations that compound, survival is contingent on economic performance, not distributional loyalty to key supporters.10 Mobutu held Zaire’s copper revenues for three decades. One coalition’s accountability structure produced compounding; the other produced the systematic stripping of whatever productive capacity existed at independence. Architecture, not endowment.

Rwanda’s Crystal Ventures is the most structurally explicit contemporary instance of this architecture: centralised rent management through a party holding company, with performance discipline formally built in.11 Rwanda has produced two parallel tracks: health and education gains alongside the growth of modern services. Neither has yet coupled with the other — the domain junctions not yet activated. The orientation is real. The compounding it has not yet produced is what distinguishes Rwanda from the more complete transformation cases.

Dubai shows what consolidated authority with genuine developmental orientation can produce. In greenfield domains (Jebel Ali, Emirates, DIFC), the ruling coalition faced no entrenched distributional interests and built real cross-domain capacity. These are not enclaves in the Kenyan sense: each domain’s gains have actively enabled the next. The practice of kafala and the merchant compact are older than the free zones and survived them. They are part of the same settlement — the coalition that built Jebel Ali also depended on them. The developmental project could be built in new domains. It could not reach back into those that the settlement had already organised.

Democratic Orientation: Inherited Settlement or Built Under Pressure

The second developmental configuration does not require consolidated authority. Mauritius in 1968 had every characteristic associated with failure: a monocrop sugar economy, high unemployment, communal tensions across a multi-ethnic population with no shared civic tradition. Nobody expected it to succeed. What followed was one of the developing world’s few genuine transformations, and the founding conditions are what made it possible.

Two founding choices established the developmental orientation. The best loser electoral system rewarded cross-ethnic coalition-building over communal patronage: any governing majority required representation across communities. The Export Processing Zone, launched in 1971, created a domain of productive activity (textile and garment manufacturing, spanning ethnic lines) that delivered gains without requiring redistribution from established sugar interests. Both choices expressed a clear orientation: build productive capacity rather than distribute existing wealth. Blocking strategies were unrewarding before they could take hold.

What followed was compounding rather than enclave growth; each domain crossing required active governing commitment, not just economic spillover. EPZ manufacturing generated foreign exchange, fiscal revenues, and a disciplined industrial workforce; successive governments directed those revenues into education rather than patronage, progressively raising the workforce’s skill ceiling over two decades.

Tourism expanded alongside manufacturing, drawing on the same infrastructure and active investment promotion; financial services followed through deliberate regulatory choices that depended on the institutional stability successive governments had maintained across electoral cycles. Each crossing was a governing decision.

The democracy worked developmentally because the founding choices had removed the incentive to defect from the productive bargain before competitive politics could fragment it, not because democratic institutions inherently produce developmental orientation.

The founding moment is decisive because it initialises feedback dynamics before organised opposition can form: each domain crossed makes the next cheaper, and the productive pattern becomes what the political system organises itself to protect.

The distributional pattern initiates the same dynamic in reverse — each budget cycle routing resources through patronage networks creates constituencies with a stake in the arrangement, each round of coalition-building that rewards loyalty deepens the logic that made it stable. Like an invasive species altering habitat conditions to favour its own propagation, the distributional settlement cannot be reversed from within. A governing coalition facing that accumulated resistance is not choosing between two policy options. It is trying to move a system that has spent decades selecting for its own stability.

What breaks it, when anything does, is almost never internal. The force that reliably suspends that selection pressure — if only briefly — is external: a crisis severe enough to make the settlement’s accumulated costs undeniable to those with the power to change the arrangement.

When crisis creates the opening

The second democratic developmental route is built rather than inherited — and it generally requires a crisis to activate. Bodies spanning employers, labour, and the state can negotiate cross-domain commitments and make them stick across changes of government. Where Mauritius built that architecture into its founding settlement, most democracies have to construct it under pressure. Building these bodies takes decades; sustaining them against normal electoral pressures is harder still.

The crisis type that creates this opening is competitive inadequacy: the sense that the country’s productive model is failing on its own terms. The pressure must be severe enough to break distributional coalition resistance and unlock what Balcerowicz called the “extraordinary politics” window,12 briefly suspending the normal calculus before it reasserts itself. Ireland in 1958 and again in 1987 is a recurring example of that mechanism.

By 1958, the country was losing forty-five thousand people per year to emigration; the population had fallen to 2.8 million. T.K. Whitaker’s diagnosis was unsparing: decades of protection had delivered stagnation. The Programme for Economic Expansion that followed opened the economy to foreign investment and reoriented the IDA toward actively attracting FDI.13 A first domain emerged in light manufacturing and early electronics assembly, generating foreign exchange and an industrial base whose skill requirements, by the mid-1960s, were outrunning supply.

Donagh O’Malley announced free secondary education in 1967 (without Cabinet approval and against the Finance Secretary’s explicit opposition), because the growing industrial economy had made it both necessary and viable.14 The IDA’s 1974 electronics strategy targeted the higher-value domain that an increasingly educated workforce would make accessible.

That arc broke in the 1970s. The 1977 expansion programme, borrowing-funded, without structural reform, initiated a debt spiral that by 1985 had pushed exchequer debt to £20 billion, 134 percent of GNP, by far the highest in the OECD. The subsequent adjustment protected current expenditure: social welfare transfers grew at seventeen percent per year, while productive investment was cut. The IDA continued to operate, its statutory independence insulating it from political direction, developing the pharmaceutical and high-value technology strategy the electronics wave had seeded: a new domain prepared before the subsequent crisis that would make it executable.

Public debt above 120 percent of GDP. Unemployment at seventeen percent. The 1987 social partnership was not a policy choice so much as a forced negotiation, with government, employers, and unions accepting wage moderation and fiscal consolidation because no credible alternative existed. It restored macroeconomic stability and cleared the ground for execution.

But the crisis also removed the IDA’s monopoly on industrial transformation, opening institutional space for the science and technology agencies that would carry out the strategy the IDA had been preparing. Padraic White’s IDA landed Intel’s European facility at Leixlip as the electronics wave peaked; the pharmaceutical wave followed directly. By 2009, pharmaceuticals alone accounted for close to a quarter of Ireland’s total goods exports, a domain that had not existed as a significant export sector thirty years earlier.15

In 2009, the mechanism was tested directly. Ireland’s banking collapse ended social partnership at exactly that moment. What followed was stabilisation, not a renewal of the developmental mode. A banking collapse frames itself as a stabilisation problem, crowding out everything else. Crisis type, not severity, is the variable. While competitive inadequacy activates the cross-domain negotiation required for developmental orientation, a banking collapse activates fiscal retrenchment.

The Conjunction

Both conditions that made these examples work are structurally uncommon: a founding settlement that pre-empts blocking, or bodies spanning employers, labour, and the state capable of cross-domain commitment when the right crisis creates the opening. Competitive elections in most democracies reward patronage, fragmenting ruling coalitions and directly undermining both routes.16

The categories are less distinct than they appear. Consolidated authority with performance accountability need not precede a crisis, it can emerge from one. China and Vietnam arrived at it through the catastrophic failure of the prior doctrine, which created space for a pragmatist faction to prevail inside the same party structure. Post-conflict rupture works through the same mechanism: the existing settlement breaks down, a window opens, and the orientation established during it determines the trajectory for decades, as it did in Rwanda after 1994 and in Côte d’Ivoire after 2011. What varies across these cases is the triggering condition. What the configurations describe is the architecture that gets built during it.

That architecture is rare. What looks, in retrospect, like developmental success is almost always a narrow conjunction: founding conditions that set the structure before politics could fragment it, or a crisis that briefly suspended the old settlement and allowed a different one to take hold. Most governing coalitions never assemble either. When they do, it explains why some countries transformed, and most did not — more than geography, endowment, or colonial inheritance.

Hidden Rules of Development

Discussion about this post

Ready for more?