The number that stopped me was from 1986.
Measured in constant 2015 dollars, Cameroon’s GDP per capita peaked that year at about $1,843 (1,051,608.24 FCFA). In 2024 it was roughly $1,481 (845,052.53 FCFA). That is less growth, not slower growth which I would have greatly preferred. The avaerage Cameroonian today is measurably poorer than someone in the same place forty years ago.
South Korea’s figure on the same measure was about $1,028 in 1960. Last year it was around $36,239.
I started this looking for why similar inouts produce succh different outcomes. I now think the framing was wrong, and dealing with that hoonestly is where the argument has to start.
The inputs I thought were similar never the same
The story that Korea and Ghana were equally poor in 1960 gets repeated because it is rhetorically useful, it implies that destiny is a choice. The underlying data shows that is real but the GDP was hiding almost everything, especially those that mattered.
Korea in 1960 had near-universal primary enrolment, a land reform that had already broken the landlord class, linguistic and ethnic uniformity, a functioning tax bureaucracy, an American security guarantee, and a preferential access to the largest consumer market on earth. North Korea was actually the richer half at the time, by 30 to 50 percent, on the back of Japanese industrialisation in the 1930s.
So the honest version is that Korea was not poorer than us in any sense except that one economist could measure.
That should be the end of the argument. It is not, because of what the missing input actualy was. Korea’s advantage was not capital or resources. We have more of the second(resources) than they ever did. It was a state that could make a commitment and be relied on to honour it. Which is not a historical accident you inherit. It is something you need to build, ad it is the right thing we haven’t.
Predictability is the input
You cannot build a self-driving car on a road system where lane markings are advisory, where the same journey takes twenty minutes or ninety depending on nothing you can observe, where a road can be closed for a week without reaching anyone who needed to know.
The instinct is to say we should fix the roads first, and that’s true but too shallow. The deeper issue is that automation is compressed assumption. Every automated system is a bet that tomorrow resembles today closely enough that a rule written yesterday still holds. Variance does not slow that down. It makes it impossible.
A system must be predictable before it can be automated. Everything we call innovation is a bet that tomorrow will look like today.
Look at what that means for power. Cameroon’s grid runs at a SAIDI of about 51.5 hours; the total time an average customer spends without electricity (could be more) each year; accross 20 seperate interruptions. The government’s own projection is to bring that to 45.8 hours and 18.3 interruptions by 2027. That is an improvment, and it is also an admisiion that a business palnning production in 2027 should expect to lose power eighteen times.
Eighty percent of the companies in Gecam name electricity as their primary operational problem. Not cost, ot tax, not access to credit. The lights
Now ask what that does to everything built on top. Cold chain, and there agro-processing. Continuous manufacturing, and therefore anything with a kiln or a line. Servers, and therefore any sofware business that wants to host domestically rather than pay Vercel or google in dollars. Each of these is not blocked by a shortage of skill or ambition. Each is blocked by variance.
Why the workaround becoms the ceiling
Here us where I would push back on how this is usually described, including how I used to describe it.
It is tempiting to say people here have stopped aiming high. That we have settled for mediocrity and lost the appetite for excellence. I said versions of this for a long time. I think it is a wrong way of lookig at it, and wring in a way that makes it useless, because it suggests the fix is exhortation.
What is actually happening is that an unprdictable environment selects for workaround thinking. If the grid fails, the correct individual response is a generator, not a grid reform; reform takes a decade and you need to run payroll in March. If transport is unreliable, the correct response is a moto, not transit authority. If the banking system si slow, mobile money layered on top of it beats waiting for the system underneath the change.
Every one of those decisions is right. That is exactly what makes it a trap.
Two things follow. A workaround that woirks well enough removes the political pressure that would have fixed the thing underneath; the generator is why grids stays brokens, not just a response to it. And workarounds don’t compound. A grid gets better every year someone invests in it. Your generator does not get better because you bought one. You just own a generator.
So we end up with a country full of genuinely capable people solving the same problem individually, permenently, at enormous aggregate cost, and calling it resilience.
The missing scoreboard
Korea’s planners were not more virtuous as per assume, they were more constrained.
The mechanism was export discipline. Firms got subsidised credit and protection, but conditioned on export performance, and foreign buyers don’t accept excuses. A shipment is late or it isn’t. The product meets spec or it’s returned. That created something Cameroon does not have: a scoreboard that no domestic actor could argue with, that made failure legible in public, quickly.
Our revenue comes substantially from oil, timber, cocoa. Extraction pays whether or not the domestic system works. There’s no equivalent of a rejected shipment.
This part I find hardest to explain sits here. Cameroon has had continuity of leadership since 1982, longer than Park Chung-hee’s entire rule. Long tenure should produce long horizons. It’s supposed to be the one advantage of not having elections every four years, and it’s the explicit argument for it. Yet the horizon has been very short anyway.
Which tells me tenure isn’t the variable. What;s missing insn’t time in office. It’s any mechanism that makes underperformance visible and costly to the people who could fix it.
I should be fair about couterexamples. Nachtigal came online in early 2025 adding around 360MW and something close to thirty percent to the generation mix, with about a billion dollars of private capital mobilised behind World Bank guarantees. That is a real long-horizon infrastructure project, competently delivered. It complicates my story and I would rather say so than pretend it away. The test is whether it turns out to be the first of a sequence or an isolated event, and I don’t know yet.
What I’m not sure about
I might be overweighing predictability. It’s a satisfying single variable explanation, and single-variable explanations of development have a poor record.
There is also an unconfortable implication I haven’t resolved. The states that delivered predictability fastest; Korea under Park, Singapore, arguably Rwanda did it with a lot of coercion and not much pluralism. I don’t want to conclude that authoritarianism is the price, partly because Korea’s biggest productivity gains came after democratisation in 1987, not before. But I can’t yet articulate the version that gets predictability without concentrating power, and I don’t trust an argument whose conclusion I picked first.
What I have stopped believing is that this is an information problem. Korea spends 5.21 percent of GDP on R&D, the highest of any major economy, and every paper explaining how they got there is free and online. The playbook is not hidden. A student in Bamenda with a phone has better access to development economics than a Korean planner had in 1965.
Sources. GDP per capita figures: World Bank, constant 2015 USD, via Trading Economics (Cameroon, South Korea). On the 1960 parity myth: “No, South Korea Was Not Poorer Than Kenya in 1960”, Global Developments. Grid reliability and Gecam figures: Business in Cameroon. Nachtigal: World Bank. Korean R&D intensity: OECD via TheGlobalEconomy