The industrial-policy test: infrastructure, skills, finance and buyers

Industrial policy often arrives as a list of preferred sectors. Batteries, pharmaceuticals, fertiliser, food processing and digital services appear in a speech, followed by a tax concession and an investment target. The list is the easy part.

The World Bank's April 2026 Africa Economic Update argues for an ecosystem approach that matches policy tools to state capability. The International Energy Agency makes a related point in its work on African clean-energy value chains: mineral deposits create an opportunity, but poor power, transport, skills and financing can stop local processing. Put the two reports together and a policy has to pass four gates before its production target deserves confidence.

Gate one: can the plant run?

Start at the proposed site. Record grid connection capacity, outage history, water availability, road or rail access, port route, telecommunications and the time needed to obtain each permit. Use measured service and signed commitments, not national potential.

A mineral-processing plant may sit near ore and far from reliable power. A food processor may have electricity but lose raw material on rural roads before it reaches the line. A digital-service exporter can avoid physical freight while depending on power, international bandwidth and secure payment access. “Infrastructure” needs a bill of quantities specific to the production process.

The IEA estimates that about 600 million people in Africa lack reliable electricity access. That continental figure cannot decide one factory location, though it warns against treating power as a footnote. A project model should include connection cost, tariff, backup generation, fuel, maintenance, downtime and the value of lost output.

Public infrastructure also has a delivery date. A factory completed before its substation or access road may carry interest during months of idle capacity. Policy documents should identify which party funds each shared asset, procurement status and the remedy if delivery slips.

Gate two: who can perform the work?

A promise to create jobs is not a workforce plan. Break the production process into occupations, required certifications, shift count and experience. Compare that demand with workers who can reach the site and training that has a funded curriculum, equipment and instructors.

The World Bank reports that most Sub-Saharan African countries spend between 0.1% and 0.4% of GDP on research and development, below the African Union's 1% benchmark. R&D spending does not map neatly to factory skills, but the gap shows why technology transfer cannot be assumed. Firms hesitate to invest in processes that require capabilities the labour market cannot supply.

Training should attach to equipment and a hiring date. A generic technical programme can produce graduates without the safety, maintenance or quality-control skills a plant needs. Employers need a role in curriculum design and assessment; trainees need a credible route to paid work. Publish completion and placement, not enrolment alone.

Management capacity belongs in the gate. A new plant needs procurement, maintenance planning, environmental compliance, finance and sales. Importing specialists may close an early gap, but visas, cost and a transfer plan must appear in the operating model.

Gate three: does the capital fit the asset?

Tax relief cannot finance construction by itself. A project needs equity for risk, long-term debt matched to asset life, working capital for inputs and foreign-exchange capacity for imported machinery. Lenders will ask who buys the output, who supplies power, what security is enforceable and which risks the state has accepted.

Currency mismatch can sink a project that works in engineering terms. Equipment and debt may be priced in dollars or euros while domestic sales arrive in local currency. An export contract can help, subject to price, volume and collection risk. A government guarantee can move risk to the public balance sheet rather than remove it.

Compare concessional finance with the obligations attached to it. A lower coupon may require procurement conditions, reporting or a long approval period. Commercial capital may move faster at a price the project cannot service. A blended structure needs one cash-flow model showing seniority, repayment dates and the point at which each provider can stop disbursing.

Policy success should not be measured by capital “mobilised” without defining whether the figure is signed, disbursed or an expression of interest.

Gate four: who has agreed to buy?

Production creates inventory until a customer accepts the price and specification. Name the domestic or export market, required standard, logistics route and competing suppliers. An offtake agreement deserves inspection: volume, price formula, duration, quality tests, termination rights and credit support.

Governments can create demand through procurement or standards, but that demand has to be funded and lawful. A local-content rule can support a new supplier and raise cost for downstream firms. The trade-off belongs in the policy assessment.

Export opportunity is not the same as market access. Tariffs, rules of origin, product certification and carbon requirements can change the delivered economics. The IEA's high-potential cases depend on strong market signals and successful investment; the agency labels them as scenarios. Repeating the output number without the conditions turns analysis into promotion.

A composite project that fails all four

Consider a hypothetical battery-material plant announced near a mineral deposit. The scenario is illustrative, not a report of an actual project. Its policy package offers a tax holiday and land. The grid connection remains a future public project. The training college has no production equipment. A development lender has issued a term sheet conditional on an offtake contract. The prospective buyer will not sign until the plant proves quality at scale.

Each party waits for another. The tax holiday has little value while the plant makes no taxable profit. Calling the delay an investor-confidence problem hides the circular dependency.

A workable policy would assign the grid project and deadline, fund a pilot line for training and product qualification, use milestone finance for that pilot, then convert a buyer's conditional interest into staged offtake. The sequence may still fail. At least the dependencies have owners.

Publish the failed gate

An honest industrial dashboard should show projects stopped at each gate. Failure can reveal that a sector choice was wrong, an agency lacked capacity or a global price changed. Hiding stalled projects encourages another announcement with the same missing inputs.

The World Bank's approach asks governments to fit ambition to capability and correct course. For investors and citizens, the shortest audit is four questions: can it run, can people operate it, can the capital survive the build, and will someone buy the output? A sector name cannot answer any of them.

Sources

• World Bank, 8 April 2026: Africa Economic Update April 2026
• International Energy Agency, 6 October 2025: Stepping Up the Value Chain in Africa