Sector Selection in SEZs: How Governments Pick Winners (and Mostly Lose)

Walk through any four-star hotel conference room in the developing world and you will hear the exact same five target sectors pitched by every zone authority on earth.

By Yaschin Mohabir · 6 November 2026 · 16 min read

Every feasibility presentation for a new special economic zone ends with a slide titled 'Priority Target Clusters'. Without looking at the country's customs data, power grid, or engineering graduates, I can tell you what is on that slide: semiconductors, electric vehicles, biotechnology, aerospace, and artificial intelligence. Ninety percent of the zones chasing that aspirational list end up ten years later with empty cleanroom plots and a handful of bonded logistics warehouses. Governments lose at sector selection when they target industries by global prestige rather than by product-space adjacency to the capabilities their economy already holds.

Why sixty biotechnology parks across three continents are currently leasing space to call centers

Over the past fifteen years, I have visited more than two dozen state-funded "Biotechnology and Life Sciences Special Zones" across Latin America, Africa, and Southeast Asia. On opening day, the renderings always feature researchers in white coats peering into electron microscopes. Walk those same estates five years after ribbon-cutting, and the actual tenant roster consists of two generic pharmaceutical packaging warehouses, a customs brokerage, and three back-office call centers occupying the glass "Innovation Tower" at a sixty percent rent discount.1 The authorities running those zones did not fail because their marketing brochures were poorly printed. They failed because they confused a sector's global growth rate with their own jurisdiction's right to win. Designing novel biologics requires a pre-existing ecosystem of research hospitals, animal-testing vivariums, cold-chain reagent customs clearance under four hours, venture capital partners who understand Phase-II clinical trial risk, and hundreds of postdoctoral molecular biologists. When a zone authority in a jurisdiction with none of those prerequisites spends $180 million building wet-lab shells, it is building an airpor...

Countries and zones upgrade by monkey-jumping to nearby trees, not by flying across the forest

Ricardo Hausmann and César Hidalgo's work on the Product Space at Harvard's Growth Lab provides the exact analytical tool zone planners have been missing.2 Think of every traded industry as a tree in a forest, and a country's existing firms and skilled workers as monkeys sitting in the trees they already know how to harvest. Industrial upgrading happens when firms jump to a neighbouring tree whose branches nearly touch their own, because both trees require the same underlying soil of capabilities: similar cleanroom discipline, similar precision CNC machining tolerances, similar cold-chain logistics, or similar regulatory certification. When Costa Rica pivoted its free zones from apparel and electronics assembly in the late 1990s into medical devices, it did not ask its zones to invent cancer drugs from scratch. Its investment agency, CINDE, recognized that manufacturing disposable intravenous catheter sets and heart-valve housings required the exact same cleanroom operator discipline, plastic injection moulding, and ISO quality control that Costa Rican workers and local toolmakers had already mastered for Intel's semiconductor test plant.3 The label changed from electronics to med...

Costa Rica, Penang, and Ras Al-Khair show how disciplined climbing beats speculative leaping

Across the empirical record of specialized versus copy-paste zones over the past three decades. Every zone that built a durable global cluster started by solving the binding shared-infrastructure bottleneck of one tightly defined value-chain node, and then climbed the complexity ladder step by step. Consider three concrete capability staircases. In Costa Rica's Coyol Free Zone, the management did not market generic industrial plots to 'advanced manufacturing'. They recognised that sterile medical catheter, heart-valve, and endoscope assembly required one non-negotiable shared asset: onsite ethylene-oxide (EtO) and electron-beam sterilisation facilities coupled with FDA-grade microbiology labs.4 Once Coyol brought third-party sterilisation providers like Steris inside the perimeter fence, medical-device OEMs could manufacture, sterilise, and ship directly to US hospitals from a single customs enclosure, cutting lead times by fourteen days and propelling medical devices from zero to over forty percent of Costa Rica's goods exports. In Morocco's Tanger-Med and Nouaceur (Midparc) zones, policymakers resisted the temptation to subsidise final consumer electronics or wafer fabs. Instea...

Industrial subsidies in the US, Europe, and China have made 'me-too' sector targeting fatal

If generic five-sector masterplans were wasteful in the 2010s, two technological shifts in the late 2020s have made the 'generalist industrial park' physically unviable. The first is the extreme divergence of utility and regulatory envelopes across modern industrial sectors. A 100-megawatt AI data centre or silicon-carbide power-electronics plant requires 99.999 percent grid reliability, closed-loop liquid cooling, and 24/7 renewable Power Purchase Agreements (PPAs). A green-hydrogen ammonia export plant requires gigawatt-scale electrolysers, desalination brine outfalls, and dedicated cryogenic pipeline berths. A cGMP biologics fill-finish park requires ultra-pure water loops, backup bio-waste incineration, and sub-two-hour airport tarmac cold-chain transfers. No zone developer on earth can afford to over-engineer every hectare of a 1,000-hectare park to meet all three specifications simultaneously.8 At the same time, reshoring OEMs using digital supply-chain twins now screen out any industrial park that cannot prove immediate Tier-2 sub-supplier availability within a two-hour trucking radius. A generalist park with one lone auto-parts plant, one food processor, and one solar asse...

Why four actors keep signing off on fantasy sector lists

Wherever a zone authority adopts a copy-paste five-sector wish list instead of a capability-filtered specialisation, the misallocation cascades across all four actors in the zone ecosystem. On the balance sheet of master developers, targeting five incompatible sectors produces a fatal capital-allocation dilemma: either build cheap, generic warehouse shells that cannot host advanced tenants, or spend $150 million on specialised cleanrooms and heavy power substations that sit unleased for eight years while interest compounds. For multinational anchor investors, arriving as the lone pioneer in an aspirational sector means bearing 100 percent of the cost of training technicians, flying in foreign maintenance engineers, and holding ninety days of buffer inventory because no local Tier-2 ecosystem exists. For investment promotion agencies, a five-sector mandate destroys sales credibility: when an IPA pitch deck claims a single desert or coastal site is simultaneously the world's best home for AI cloud regions, EV battery gigafactories, pharmaceutical API synthesis, and luxury tourism, serious corporate site-selection consultants throw the deck in the bin. Meanwhile, for ministries of f...

Five questions that separate a real cluster strategy from a brochure

Before any zone board or cabinet signs off on a sector strategy, I would subject every proposed target industry to a five-gate Capability Distance and Shared-Commons Filter. Run the ten-year developer and public-capex math on a 300-hectare zone comparing a conventional 'Five-Sector Aspirational Masterplan' against a 'Single-Niche Capability Staircase' where seventy percent of specialist infrastructure capex is concentrated on one validated value-chain node. 10-year developer & sovereign return: Wish-List 'Biotech/Fab' Park vs. Adjacent Med-Tech/Precision Cluster (200 ha) Illustrative 200-hectare zone comparison. Targeting a sector one rung up the product-space ladder from existing regional capabilities lifts Year-7 occupancy from 22% to 84% and raises developer IRR from a distressed 2.8% to 16.2%. Translating this discipline into practice requires re-architecting how zone authorities select and sequence sectors across three operational horizons. In Horizon 1 (Years 1 to 4), the zone targets a single high-feasibility wedge that sits within one capability step of existing regional strengths—such as agricultural cold-chain processing, automotive wire-harness and metal stamping, or...

Earn the right to the next rung by mastering the one in front of you

Strategy in special economic zone design is defined by what you refuse to subsidise. A zone that claims to be open to every high-growth industry on the planet is, in operational reality, prepared for none of them. High-value industrial tenants do not buy land; they buy a pre-solved operating ecosystem that eliminates their specific technical, regulatory, and supply-chain bottlenecks. When a zone authority grounds its sector selection in honest Product Space diagnostics—starting where domestic capabilities already exist, building the one shared technical facility that private firms cannot justify alone, and climbing the complexity ladder one adjacent rung at a time—every new tenant makes the zone more valuable for the next tenant. So the next time a consultant hands you a masterplan recommending Pharmaceuticals, Semiconductors, EV Batteries, AI, and Aerospace for the same greenfield site, ask one question: which specific ISO-accredited testing laboratory, utility specification, and vocational faculty are we funding out of Phase 1 capex on Monday morning? If they cannot name the single shared machine that anchors the cluster, tear up the five-sector slide and start with what your e...