The Farm Before the Factory

Part one of three on How Asia Works.

I've been reading How Asia Works on and off, mostly on planes, and the first chapter did something I didn't enjoy at the time. It took a thing I believed without ever having thought about it and showed me I'd got it backwards.

The thing I believed: big farms beat small farms. Obviously. One operator, a thousand hectares, a combine harvester, against a thousand families with jembes. You don't need economics for that, you just need eyes.

Except it turns out "beat" is doing a lot of work in that sentence.

Big mechanised farms win on output per worker. Small farms, worked hard by families who own them, win on output per hectare, sometimes by a lot. The reason isn't mystical. A household that keeps what it grows will weed by hand, intercrop, plant in stages, carry manure to the plot, and pick over the same ground four times in a season. Nobody supervises that. It's a garden, not a field, and gardens are more productive than fields per unit of land. They're just brutal on labour.

And for a poor country, per hectare is the number that counts. Land is fixed. Labour you have too much of. Capital you don't have at all. Building your agriculture around expensive machines and few workers means you've optimised for somebody else's constraint.

That's the factor endowment argument and I find it hard to unsee now. In America labour is dear and capital is cheap, so spending half a million dollars on a machine that replaces thirty people is just arithmetic. Flip both terms and the same arithmetic gives you the opposite answer. Most of the development advice I absorbed in my twenties was American arithmetic with the terms unflipped.

What the land reform was actually for

Japan, Korea, Taiwan, later China: break the landlord estates, hand the land to the families already working it. Yields rose. But Studwell's point isn't really the yields. It's where the surplus went afterwards.

A tenant produces and most of it leaves the village. An owner produces and keeps it. Do that across several million households and things start compounding in ways nobody planned in detail. You get a rural middle class where there used to be a rural underclass and a handful of very rich men. Those households put children in school, and twenty years later the factories have someone to hire. They also start buying things, so when manufacturing does show up there's already a domestic market with money in its pocket rather than an export-or-die problem from day one.

The fourth effect is the one I keep coming back to. The cities filled slowly. A family making a decent living off its land doesn't send its sons to sleep in a city on the chance of work. The migration happened, but late, and toward jobs that existed. Asia didn't urbanise and then industrialise. It made the countryside bearable first and let the cities pull rather than the countryside push.

The half everybody skips

Here's what caught me off guard: redistributing the land was the cheap part.

What came after was the expensive part. Extension officers who actually turned up, who knew the soil in that specific valley. Credit that could reach a man with one hectare and no title to pledge. Seed and fertiliser arriving before planting rather than in week three. Boards that bought the crop at a price you could plan around. And protection: tariffs, quotas, price floors, so a shipment of cheap imported rice or a wave of food aid couldn't wipe out the reason to plant next season.

Without all that, redistribution is just poverty changing hands. The land moves and nothing else does.

I think that sentence is the most important one in the chapter and I've never seen it in a summary of this book.

Which brings me somewhere I didn't want to go

The natural next move is to write "and so Africa needs land reform." I can't write it.

Look at the continent. The great majority of African farms are already smallholdings, already family-held. And Kenya specifically has run this exercise more than once: the consolidation and titling push of the 1950s, then the settlement schemes after independence that moved the White Highlands into African hands, then sixty years of subdivision on top of that.

We did the redistribution. We're still here.

So the question the book leaves me holding isn't the one I opened it expecting. It isn't how do we get land to farmers. It's: why does a hectare in Kenya yield so much less than a hectare in Taiwan did in 1965, when from a distance the farm structure looks similar?

My honest guess is we did the cheap half and skipped the expensive one. Extension thinned out to almost nothing. Smallholder credit never arrived at any real scale. Inputs are late and unreliable and everyone knows it. The marketing boards we built got captured and turned into political instruments instead of price-stabilising ones, which is arguably worse than not having built them. And a plot that fed a family in 1965 has been cut three ways twice and now feeds nobody.

There's also the bit that's closest to my own work. A farmer who does produce a surplus frequently can't move it. He's eighty kilometres from a real market on a road that eats his margin, he learns the price when he arrives, and he's selling to a broker who knows more than he does about what it's worth. The surplus exists. It gets destroyed in transit, sometimes literally and sometimes just in the arithmetic.

None of that is a landlord problem. It won't respond to a landlord solution.

On what's still legal

There's a fatalist version of this essay and I don't want to write it, partly because I don't believe it and partly because I've noticed it's very comfortable to believe.

Yes, the toolkit narrowed. The technology transfer Japanese firms squeezed out of American ones in the sixties would get you sanctioned now. A lot of what built Korean industry sits badly against WTO commitments. The rich world climbed a ladder and then went to some trouble to negotiate it out of reach, which wasn't malice, just self-interest, which is the only motive in this whole story that never needs explaining.

Narrower isn't closed though. China joined the WTO in 2001 and did the bulk of its industrialisation afterwards, in full view of everyone, using exchange rate management, state-directed lending, SEZs, infrastructure at a scale nobody could match, and industrial policy that simply moves faster than a trade dispute can be litigated. African countries also hold preferences and LDC carve-outs Korea never had. The constraint is real but I notice it gets cited most often by people who weren't going to try anything anyway.

Same with services. I started out selling software services, so I know exactly what that road offered. AI is closing a decent chunk of it faster than we can walk down it. Fine. That's a reason to move, not a reason to sit.

The part I can't resolve

Studwell is blunt that entrepreneurs didn't do this. States did. States that forced firms to export, that cut off the ones which failed, that could put an extension officer in a village and keep him there. And behind that, pressure: develop or be eaten.

I can't build an extension service. No founder I know can. So most of this is not my job and I should say so plainly instead of implying otherwise at the end of an essay.

Where I'd push back is the claim that we've got no pressure of our own. Ours isn't an army at the border. It's the number of young people arriving into economies with nowhere to put them, every year, for the next thirty years. That's an existential clock too. It's quieter than a war and gets ignored for exactly that reason, but the arithmetic isn't paying attention to whether anyone's looking at it.

And some of the work doesn't need a ministry. The road. The market. The price you can see before you travel. The payment that clears. The container that crosses a border without three weeks of paper. That's not agricultural policy, it's plumbing, but it decides whether producing a surplus was worth anyone's time.

Where that leaves me

No playbook. I want to say that clearly because the pull at the end of a book like this is to manufacture one, and a borrowed conclusion is worse than an open question honestly stated.

What I've got is a better question than the one I started with. Not "how do we copy Asia". We can't, and on agriculture we may have already done the copyable part badly. But why our version of the same structure didn't compound, and what would need to be true for it to start.

Next: manufacturing and export discipline, where the book gets genuinely ruthless, and where I suspect the lesson travels better than this one does.