Why this work exists

The question this page answers

Every organisation can describe what it does. Fewer can say what would have to be true for the work to be unnecessary. This page states the conditions we think are actually producing the problem, in a form that can be argued with: each is written as a condition, because a condition can be tested and a complaint cannot.

Where the evidence differentiates by region, we say so. Where it does not, we do not invent a regional story to make the case feel sharper.

Seven conditions

1. Capability was substituted by provision

Conflict and enclave economics, followed by years of imported programme delivery, left these regions with services rather than with the ability to run them. When a funded programme ends, the capacity to run the next one does not stay behind.

The peacebuilding literature supports the structure of this claim from a different angle. In post-war Jaffna, formal schooling was found inadequate to the central work of reconciliation, while non-formal community education carried the potential but could not reach the structural inequalities underneath[67]. Across the wider case set, relationship- and trust-building were observed only in non-formal education led mostly by civil society[65]. And the argument in the Sri Lankan literature is explicitly for combining in-school and non-formal provision, because the formal system does not reach every child and civil-society organisations have little room to operate on their own[66].

2. The state’s delivery layer is thinnest where its mandate is largest

Sri Lanka’s curriculum reform makes ICT and Entrepreneurship & Financial Literacy compulsory essential subjects in Grades 6–9 for the first time, phasing in from 2026 with Grades 1 and 6 and reaching the full band by 2029[61, 62]. The delivery capacity for that mandate was never built out. About 19% of the least-resourced school types have an IT lab, against 72% of the top tier, and the technology-stream student-to-teacher ratio runs at 35.3 against a national average of 12.8[63]. On the household side, computer ownership in the estate sector is a little over one per cent[58].

We read this as a demand signal rather than a competitive threat. The state has committed to something it cannot yet deliver rurally, which creates a partner position — and decentralising support infrastructure is itself a stated national priority, so this is execution of a policy rather than a request for an exception to one.

3. The link from capability to income fails for want of coupling, not training

This is the hinge of the whole analysis, and it is set out on its own below.

4. There is no local node, so capability leaves

The innovation ecosystem is concentrated in Colombo, early-stage capital is structurally scarce, and no Sri Lankan cluster appears among the world’s ranked innovation clusters. Domestic venture capital is immature, so young companies run on bootstrapping, angel money or foreign grants. That is a property of the ecosystem before it is a property of any organisation working inside it.

Skilled emigration is scored nationally as a medium-term drag on growth[60]. Our own record shows movement in both directions — people who stayed and founded businesses locally, and people who left under household economic pressure and later came back. Talent circulates; it is not simply lost.

5. The formative window closes before most provision starts

Entrepreneurial capability is developmentally banded. Children cannot cleanly separate capital ownership from labour until roughly ages twelve to thirteen[44], profit understanding arrives in middle childhood but is gated on a prerequisite that has to be taught[45], while saving is accessible from about six in a concrete, playful setting[43]. The most on-point trial, with roughly 2,751 children aged eleven and twelve, found robust gains on non-cognitive entrepreneurial skills, none on business knowledge, and a negative effect on stated entrepreneurial intentions[39].

We reached this conclusion from our own record before we found the literature. Targeting eighteen-and-over proved late for lasting change, which is why the work moved down into schools.

6. Actual readiness sits well below what enrolment implies

The binding constraint on delivery is not equipment, connectivity or curriculum. In one selected group, five of about fourteen students could reproduce a demonstrated program and none built one independently; the fastest finishers could not explain why the skills mattered. A trainer generalising across many sessions put it plainly: fewer than one in ten students were fit for the planned content, and the biggest issue was not facilities, laptops, internet or tools but the readiness mismatch. Attendance decays at the point complexity arrives.

This is corroborated outside our own record. Dosage is decisive — a single workshop produced no significant gains where a multi-session programme did[40] — and financial education moves knowledge far more than it moves behaviour. Regionally it sharpens: estate-sector literacy is falling, and that sector is overwhelmingly smartphone-only for internet access. A cohort that has never used a desktop operating system is not a cohort that simply needs the advanced module.

This is why more programmes is the wrong answer. Adding delivery on top of an unmeasured baseline reproduces the same decay. The answer is a measured intake baseline, basics-first sequencing, and enough dose to matter.

7. Nobody is generating the evidence that would settle any of this

There is no clean result anywhere linking a rural makerspace to durable income. Sri Lanka does not appear in the maker-education literature at all. There is no Sri Lanka-specific impact evidence on early-age entrepreneurship education — only policy intent — and the nearest usable evidence comes from Rwanda, whose transfer to post-conflict, multilingual, rural Sri Lanka is untested. The only sub-national Sri Lankan study of start-ups covers the Northern Province; there is no Eastern or estate-belt equivalent.

That is a cause of the problem, not merely an inconvenience for our reporting. Funders allocate to what is measured, so an unmeasured region is an underfunded region — which reproduces the fourth condition above.

The hinge: training alone is not the mechanism

Of every condition on this page, the third is the one that changes what an organisation should actually build.

The evaluation literature reaches the same verdict from the other direction. Training on its own is null-to-modest and often fades: only about 29% of short-term earnings estimates are positive, a meta-analysis across thirty-seven evaluations found no immediate translation into business set-up or increased income, and the most rigorous skills trial shows effects fading by twelve to fifteen months[55, 56].

Training plus a complementary input is a categorically different thing. A job referral produced a large earnings gain where training alone had done nothing. One-off capital faded over years, while multi-component bundles — capital, training and sustained supervision together — produced durable gains. And making specifically has been found to foster only some entrepreneurial skills unless explicit programmes and incubation are layered on top of it[6, 32].

That converts our weakest claim from an article of faith into a design specification with a known failure mode: deliver training without the couplings and the gain decays on a schedule the literature can already predict. It is why the model is built as an ecosystem with finance, bridging and multi-year contact in it, rather than as a course.

What we are not saying

• Not that aid, or grant money, was the problem. Grant funding is one of the three sources that pay for this work, and we say so. The criticism is of imported solutions, and specifically of solutions designed without the people who knew the problem.
• Not that outside knowledge is the problem. Bringing the world in is half of what we commit to. The solver has to be local; the tools, capital and expertise need not be.
• Not that the North and East are the worst-off part of the country on unemployment. They are not, and the labour-force data says so plainly. The claim here is about how the war reshaped communities, economies and environments — which is a different statement, and a defensible one[57].
• Not that any of this is unique to Sri Lanka. Most of these conditions are recognisable across the global South. What is local is the particular combination, and the absence of anybody measuring it here.

What follows from it

Each condition points at a different part of the model, which is why the model has more than one stream in it. Capability substituted by provision is answered by building people rather than delivering to them. A thin state delivery layer is answered by working inside government schools and training the teachers, principals and preschool teachers who stay after we leave. The missing coupling is answered by finance, bridging and long contact. The missing local node is answered by building one. And the missing evidence is answered by treating research as a product rather than as overhead.

How those fit together, and how well each is evidenced, is set out in the theory of change.