Builders versus Owners

The runway, read in both directions

One side of the departures board is read aloud in every boardroom on the continent. The other side is booked by no one, and it is the side this brief is about.

We count the builders at the gate. We rarely audit the owners.

For thirty years African policy has had one word for the executive who moves to Amsterdam: drain. A vivid word doing far too much work.

A drain runs one way; the evidence looks more like a tide. AnnaLee Saxenian documented the return leg across Taiwan, India, Israel and China, where engineers who supposedly drained into Silicon Valley came back to seed the industries their leaving was meant to kill.⁷ She called it brain circulation, real for some sectors and false for others.

The harder paper is the one most cited and least read right. Studying Indian patent citations, Agrawal, Kapur, McHale and Oettl find skilled emigration harms domestic knowledge access on average, while the diaspora's access matters most for a country's highest-impact inventions.⁸ The channel is an offset, not a reversal, which the "brain bank" slogan flattens into a reassurance its authors never offered.

Africa is now running that experiment at continental scale, almost by accident. Founders leave Lagos, Nairobi and Johannesburg in numbers you can count; nearly every serious African start-up is incorporated in Delaware before its founder books a flight;⁹ and the diaspora wired roughly $95 bn home in a year, more than most categories of foreign direct investment.¹³ The flows are enormous, run both ways, and no institution manages either.

The migration argument is really an ownership argument wearing a passport. A builder makes value early, visibly, often locally; an owner captures it late, wherever the paperwork points. So every economy keeps two sets of books, one for who makes the value and one for who keeps it. The continent has read the first to the decimal and barely opened the second; what follows tries to open it, one measurable line at a time.

Does the proxy everyone uses actually measure ownership?

The case that ownership is coming home rests on one number: since 2019, Africa-based investors have led 59% of million-dollar-plus deals for Big Four start-ups, against 37% elsewhere.³ The number is real, and the wrong instrument for the claim.

Lead-investor nationality tells you who runs the round, not who owns the company. The most active "African" investors this cycle are, to a striking degree, foreign development-finance institutions, Britain's BII, the World Bank's IFC, France's Proparco, the Netherlands' FMO, Germany's DEG and America's DFC, and every African Series A in early 2026 carried at least one.¹⁶ A Nairobi manager deploying European limited-partner money is placing offshore capital, not holding domestic ownership, and counting him as the latter is the quiet substitution the optimistic reading depends on.

The "domestic investor" bucket compounds this, folding in African corporates whose treasuries write cheques on strategic, not patient-venture, mandates, mixing genuine local risk capital, recycled development finance and balance-sheet plays that behave nothing alike.¹⁶ None of this makes the 59% false; it makes it carry far less than asked, and the honest response is to look for measures that speak to ownership directly.

Two measurements speak to ownership without a proxy. The first is macroeconomic. When foreigners own domestic assets, the income those assets throw off leaves the country as investment income on the balance of payments, in the primary-income account; a persistent, widening primary-income deficit is the national accounts confessing that more of what the economy produces accrues to owners abroad.¹² It is blunt, but it is the closest thing to a hard ownership-flow number the statistics produce.

The second measurement is transactional. At a company's largest liquidity event, the acquisition or listing, public records show where control goes.

Both measurements point the opposite way from the proxy. At the macro level a growing share of what these economies produce is paid to owners abroad; at the deal level, control of the best companies relocates at the exit. Neither is fatal to circulation: the jobs and taxes are a separate account that reads far better. But this is the ownership account, and it has been left unread.

One more exhibit the title implies and honest reporting cannot yet deliver: a hand-built sample of fifteen to twenty disclosed Series B cap tables, tracing how founder and local equity dilutes as Delaware incorporation and foreign leads take the stock. It does not exist in public form anywhere on the continent, because private African cap tables are not disclosed.

Operating value stays home almost regardless of structure. Ownership does not.

The earlier version of this instrument produced a single figure, 57¢ of every dollar of value staying home, from no model. This rebuild replaces it with three companies that actually file their numbers, and it separates two things the 57¢ silently fused: where the operating value lands, and who owns the company.

Read "home" here as the African market or markets a company operates in, and "abroad" as everything else. For each of three publicly reporting firms we can put a defensible figure on two questions. The first is where the annual value a company distributes, to staff as wages, suppliers as procurement, governments as tax, shareholders as dividends, actually comes to rest. The second is who holds the equity, and so the claim on every future dollar. The instrument below carries both, because the gap between them is the argument of this brief in two bars.

The instrument makes a point the original got half-right and stated backwards. Its predecessor called incorporation a red herring and operation the lever, and for the operating account that broadly holds: a Delaware or Berlin parent barely moves where wages, procurement and tax land, because those follow the work, and the work stays. The ownership account tells the opposite story, and it is the one the title is about. Safaricom and MTN keep most of their ownership at home because they list at home; Jumia keeps almost none, a German company listed in New York, and no amount of African operating spend changes who holds the upside. Incorporation is a red herring for jobs and taxes and decisive for ownership, and conflating the two is how a real insight about operations got mistaken for a reassurance about ownership.

Debt at 41% of capital has two readings. The original brief printed one.

"Debt finances companies built to operate, not flip" is a good line and exactly half the meaning. The other half: debt is what founders raise when equity dries up, and a rising debt share in a market with almost no exits can signal a stalled equity engine as easily as a maturing one.

To adjudicate, look at who is lending, on what terms, and in what form, and the evidence tilts hard. Development-finance institutions, chief among them Britain's British International Investment, the World Bank's IFC and France's Proparco, have topped debt capital to African tech for two years running, each closing three or more deals a year while commercial venture lenders stayed thin.¹⁶ Most of that debt is dollar-denominated while borrowers earn in naira, cedi and shilling, a mismatch that turned brutal when the naira lost more than half its purchasing power between January 2023 and January 2024 and the shilling shed 22%.¹⁶ Local-currency facilities are the sensible structure and remain the exception.

Form matters as much as currency. Much of what 2025's headlines filed under record rounds was, on inspection, project finance and asset-backed securitisation rather than venture debt: Sun King's $156 m receivables securitisation put commercial banks in the senior tranche and development institutions in the riskier mezzanine; SolarAfrica's $98 m was project finance against a solar installation; Wave's $137 m carried concessional development-finance participation behind a commercial lead.¹⁷ There the commercial money takes minimal exposure and the development institutions absorb the downside, which is intelligent engineering and is not private venture capital returning to Africa. Concessional development debt and commercial venture debt point in opposite directions, and the headline 41% fuses them into one number that reads cleanly neither way.

Where does the circulation story break down?

A brief arguing for circulation owes the case where it fails, and the clearest is the departure lounge of nurses, anaesthetists and general practitioners. Here the offset the earlier sections leaned on vanishes.

The World Health Organization sets a minimum density of 4.45 doctors, nurses and midwives per 1,000 below which basic care suffers, and counts 83 countries beneath it, most of them African; as of 2022 only Seychelles, Namibia, Mauritius and South Africa cleared it.²² Roughly one in five African-born physicians now works in a high-income country, the OECD records emigration rates above 70% for Liberia, and in 2023 nearly half of new UK doctors had trained abroad.²³ The African Development Bank puts the annual cost of health-sector emigration at around $2 bn.¹⁰ A poor system that trains a doctor who then registers in Manchester has made a direct capital transfer to a rich one, and leaves nothing on any register to reactivate.

Every mechanism that makes founder migration recoverable is absent for the clinician. The departures are largely permanent, not the commuting Saxenian saw; a remittance buys groceries and school fees but does not staff an intensive-care unit; there is no equity to keep a departed nurse tied to the ward she left; and the loss falls on the very public systems this brief shows are hollowing out. Circulation works for sectors that are tradeable, capital-light and network-dependent, and fails for licensed, place-bound, public-service work, where the output is a person at a bedside.

Why Taiwan, India and Israel circulated, and whether the preconditions travel

The success stories are invoked so often that their preconditions have been sanded off. Taiwan's ran through Hsinchu Science Park, more than half of whose firms were founded by Silicon Valley returnees, and through a deliberate state effort to pull them back.²⁵ Israel's ran through the state-created Yozma programme of the early 1990s, a venture-capital industry built from scratch and fed by US returnees, a Russian immigration wave, and a defence establishment that worked as a national engineering network.²⁵ India's ran on a domestic market and IT-services base big enough to give returnees something to return to. Each case combined a strategically connected home industry, a deliberate programme to pull talent back, and a proven exit.

Set that against Lesotho, whose main human export is labour to South African mines and hospitals and which has no returnee programme nor a capital market to return capital to; or a South Africa that has the market and the exchange but built no Yozma and watches its clinicians leave for good. And even the best case carries an awkward line the literature skips: Israel kept listing its companies on foreign exchanges, so most of the financial gains flowed to overseas investors.²⁵ Circulation brought the industry home; ownership of it largely stayed abroad, the same two-column split this brief keeps finding.

A South African interlude, written from under the jacarandas.

In Pretoria you can date a memory by the jacarandas. They bloom in October, the streets go purple, and every year somebody at the braai says it: did you hear, so-and-so's daughter is in Amsterdam now. The doctor cousin is in Perth. The fintech guy from varsity is in London, doing well, and nobody is quite sure how to feel about it.

The numbers say the braai is not just anecdote. More than a million South Africans now live abroad,⁵ and Afrobarometer finds 27% have considered emigrating, rising to 38% among the most educated and 42% among the wealthiest.⁶ In early 2026 national unemployment was 32.7%, youth unemployment among 15-to-24-year-olds 60.9%, and roughly 8.1 million people were out of work.¹⁴ The people most courted by Amsterdam and London are, disproportionately, the ones who might have built the firms to employ the 60.9%. South Africa is not short of people who need jobs. It is short of owners of the things that make them.

And yet the corridor complicates the grievance. It keeps furnishing evidence for circulation: Mark Shuttleworth sold Thawte, left, then funded South African ventures from abroad for two decades, while Naspers, a Cape Town media house, built one of the world's most valuable internet holdings and chose Amsterdam for the Prosus listing.⁷ The current data is just as stubborn. In 2025 South African start-ups raised about $600 m, up 51%, more than 90% of it equity, the largest equity share on the continent, and took first place in equity funding since 2017.¹¹ Departure and dynamism sit in the same dataset.

The lived truth is messier than either camp admits. The operator lost to Amsterdam on Monday is often, three years later, the angel cheque, the warm introduction to a Dutch pension fund, the first international customer. The loss is immediate and legible, the return delayed and diffuse; whoever manages only what is legible will misprice both. But the jacaranda street knows what the spreadsheet resists: the doctor cousin in Perth is not coming back, and that is a different loss from the fintech founder's.

Remittances are the largest unmanaged investment programme in Africa. Twice, someone tried to manage a slice.

Strip the $35 bn Ras El Hekma megaproject, funded by Abu Dhabi's ADQ, from 2024's headline $97 bn in African foreign direct investment, and the broad-based figure falls to about $62 bn, below the roughly $95 bn the diaspora wired home that year.¹³¹⁸ The diaspora out-invested the world. It just files it under groceries.

The flow deserves the comparison because it behaves better than the capital the continent chases. Over the past decade, remittances to low- and middle-income countries rose 57% while foreign direct investment to them fell 41%; the money is stable, counter-cyclical and loyal, arriving when portfolio capital flees.¹³ Remittances behave better than the capital we chase, yet almost every recipient country leaves a flow of that size unstructured. The instrument built to structure a slice is the diaspora bond, whose African record is a short lesson in what the idea needs.

Ethiopia is the cautionary case, twice over. It issued a diaspora bond in 2008 for the state electricity utility and again in 2011 for the Grand Ethiopian Renaissance Dam, and both fell short, undermined by distrust of the government and a default risk no patriotic premium could offset.²⁴ The 2011 offering was not registered with the US Securities and Exchange Commission, and in 2016 the utility settled SEC charges of selling unregistered securities to US residents of Ethiopian descent, unwinding roughly $5.8 m raised from more than 3,100 investors: a bond that ended as a rescission notice.²⁴

Nigeria is the apparent success that underperforms differently. Its 2017 issue of $300 m in 5.625% diaspora bonds due 2022 was oversubscribed by about 130%, and it worked because Nigeria did the legal work Ethiopia skipped: SEC and UK registration, a London listing.²⁴ But $300 m raised once, against a remittance flow of roughly $20 bn a year,⁴ is a rounding error dressed as a milestone, never repeated or scaled. Issuance is the easy part; the binding constraints are trust and structure.

What a bond would need follows from those failures. It must be registered where the diaspora lives, both to sell and to avoid the enforcement that ended Ethiopia's; it must ring-fence proceeds and report against them, because migrants forgive a low coupon but not mismanagement; it must price the patriotic discount honestly, since an under-market bond simply fails to sell; and it must offer a local-currency option to neutralise the exchange risk that deters a dollar-and-pound diaspora from lending in naira or birr. The places that made diaspora finance work, Israel across seven decades and India at the scale of billions, had the trust and dense ties first, and no prospectus can manufacture that.

Stop counting departures. Start auditing ownership.

The model worth keeping is the wedge. Every economy builds value and lets some of the ownership settle elsewhere; the gap runs wider than the funding totals suggest. The talent debate, the brain-drain panic and the guilt at the gate all argue about who builds. The number that decides the next decade is who owns, and it points the other way from the reassurance those totals usually give.

Once the two accounts are separate, the policy noise quiets. Exit taxes, loyalty campaigns and visa walls all try to pin the builder in place and fail, because a builder forced to stay is not an owner you kept but a talent you wasted. The lever for the owners' account is the register, not the border. It works for tradeable, capital-light sectors and not for the clinician, whose loss no cap table can recover. And it moves ownership abroad by default, through Delaware parents and foreign listings and foreign leads, so keeping a share at home is not automatic; a board, an exchange and a policymaker have to build it on purpose.

So the questions change. A board stops asking how to keep people from leaving and asks how to stay on their cap table after they go. An investor underwrites the corridor, not the company. A policymaker stops taxing the runway and builds the exchange that lets ownership rest at home. None of this asks anyone to stay; it asks someone to own, and admits that for the doctor cousin in Perth none of it will be enough. Her loss sits on a balance sheet no register can reopen. A country does not grow rich by keeping its builders at home. It grows rich by owning a share of what they build, wherever they build it, and by being honest about the exports, clinical labour above all, from which it will collect no share. Look again at the wedge: it widens by default and narrows only on purpose. Talent does not owe geography its loyalty; geography has to earn it back, with interest.