WHERE AUSTRIAN ECONOMICS MEETS REAL WORLD ACTION

Nineteen months ago, Cyril Ramaphosa signed the Expropriation Act into law.

Since then the state has not expropriated a single farm under it. Not one warehouse, not one plot. It cannot — the President has never issued the proclamation that brings the Act into operation. It sits on the statute book: signed, gazetted, litigated, dormant.

And in those nineteen months it has been changing how South African land is valued, borrowed against and improved.

Earlier this month, three judges in Cape Town spent five days hearing argument about whether that Act is constitutional. It is the right question for a courtroom. It is the wrong question for anyone trying to understand why the land market has been repricing since long before this Act existed.

Because the Act did not create the problem. It finished a job the Constitution started in 1996.

On 3 August, a full bench of the Western Cape High Court began hearing a challenge to the Expropriation Act 13 of 2024. The applicants are the Democratic Alliance, AfriForum, the Institute of Race Relations and the Vaderland Stigting — a strange line-up, given the DA sits inside the coalition government it is suing.

The Act replaces apartheid-era legislation from 1975 and sets out how the state must go about taking property: who may expropriate, what process applies, how compensation is determined, what rights owners have. The contested provision is section 12(3), which identifies circumstances in which nil compensation may be just and equitable.

Three arguments were run:

— The list that doesn’t end. AfriForum argued that “including but not limited to” in the nil-compensation provision creates an unlimited category of situations where the state could decide nothing is owed. A list that announces it does not end is not a list. It is a discretion.

— The broken dispute mechanism. The DA argued the Act is irrational because it creates an unworkable process for resolving compensation disputes. The state largely agreed: lawyers for the President and the National Council of Provinces conceded that a section dealing with compensation disputes contains an error creating a “circular process,” and proposed a court-ordered remedy. Which means a plausible outcome here is not that the Act falls, but that it survives with judicial repairs bolted on.

— The vote. The DA argued that fewer than five provincial delegations voted in favour with lawful authority, making the passage through the National Council of Provinces unlawful — a defect they say invalidates the whole Act by itself.

Judgment was reserved. Wherever this lands, it is going to the Constitutional Court.

And the government’s case deserves stating properly. The 1913 Natives Land Act restricted land ownership for the black majority and allocated most farmland to white settlers — a statute with a paper trail, not an inference from inequality data. The programme that followed democracy aimed to transfer 30% of agricultural land within five years and fell far short. The state’s position is that willing-buyer-willing-seller failed. On its own targets, it did.

Take every word of that as true. What follows is unaffected — which is exactly why it is worth saying.

Section 25 does not require the state to pay market value. It requires compensation that is “just and equitable,” reflecting a balance between the public interest and the interests of those affected, having regard to all relevant circumstances — including the property’s current use, the history of its acquisition, its market value, the extent of state investment in it, and the purpose of the expropriation.

Read that list again, slowly, and two things become visible.

First, it has a direction. Market value is one factor among five. Every other factor on that list is a reason to pay less — history of acquisition, state subsidy already received, the public purpose being served. Not one of them is a reason to pay more. This is not a balancing test. It is a discount with a list of justifications attached.

Second, and more important: the list does not end either. The Constitution says “all relevant circumstances, including.” So when AfriForum stands up in court and attacks “including but not limited to” in the Act, they are attacking a drafting move the Constitution made first, twenty-eight years earlier.

Now the economic question, which nobody in that courtroom is asking.

Market value is discoverable. Not perfect, not precise, but findable — you look at what comparable property actually changed hands for between two parties who could each have walked away. It is a fact about the world, arrived at by people revealing what things are worth to them through their willingness to part with money.

Every other factor on that list is a judgement. How many rand is “history of acquisition” worth? There is no procedure that answers that, because it is not a magnitude anyone can observe. It is an opinion about the moral weight of the past, converted into a number by someone with the authority to do so.

So the formula takes a figure partly grounded in observed exchange and replaces it with a figure grounded in an official’s assessment. And once you have done that, there is no principled place to stop — because nothing in the formula tells you where the bottom is.

That is the trap. Not nil compensation. The absence of a floor.

Section 12(3) does not depart from the constitutional logic. It completes it. It is the first statute honest enough to write down where a formula with no floor eventually arrives.

The Act extends the problem in three specific ways:

— It names the terminus. Everyone could work out that “just and equitable” might mean very little. Section 12(3) says so out loud — and what people believe about the range is what gets priced.

— It removes the last discipline. An exhaustive list of nil-compensation circumstances would at least be readable; you could check whether you were in it. “Including but not limited to” means you cannot.

— It breaks the way out. The compensation mechanism its own drafters conceded is circular means the question may not be answerable at all. An uncertain formula is bad. An uncertain formula with a broken procedure for resolving disputes about it is something else.

Worth noting: Parliament tried in 2021 to amend section 25 to make nil compensation explicit, and could not reach the two-thirds majority. What could not be done by amendment is now being attempted by ordinary statute. That, in substance, is what those three judges are being asked to decide.

Prices are not labels. They are information — and they exist only because people who securely own things trade them.

Take away confidence in the ownership and you do not get distorted prices. You get prices that no longer tell anyone anything.

Watch it work. Every land valuation in South Africa carries an invisible term for what “just and equitable” will turn out to mean when somebody finally applies it. Nobody can size that term. The formula has no floor, the nil-compensation category is open-ended by design, the dispute mechanism is conceded broken, the judgment is unwritten. So the buyer discounts, the seller resists, and neither can say how much of the gap between them is the land and how much is the politics.

The farmer weighing a twenty-year irrigation project cannot calculate it — he does not know what he will own in year eight. The bank pricing a land-backed loan does not know what the collateral is. The valuer signing the number cannot defend it.

And we have watched this run before. In 2017 the governing party resolved to pursue expropriation without compensation. No bill. No statute. Nothing taken from anybody. Farmland prices fell anyway, and agricultural economists at the time attributed the fall to uncertainty over property rights and the opacity of the government’s reform plans.

Nothing needed to be expropriated. It only needed to become unknowable.

You can price a rule. You cannot price a discretion.

But something has to be done about the past.

Agreed — and notice that the case for doing something does not tell you which something. Two very different things get called by the same name.

Restitution is specific. A named property. A documented taking. A claimant who can prove the chain from the original owner to himself. Where that is established, returning the property does not violate property rights — it enforces them. It says titles are real, traceable, and good even against the state that took them. A country that does this ends up with stronger property rights, not weaker.

Redistribution is statistical. It starts from a group-level disparity and works backwards to a transfer, assuming the disparity reflects wrong in some proportion nobody can specify.

Here is the test that separates them. Land was taken from a man a century ago. Today he has thirty descendants. Which of them owns it? Divided how? Decided by whom, on what evidence?

There is no answer, and the absence is not a gap for a formula to fill — it is the claim failing. Whoever asserts a right to another person’s property carries the burden of proof, and if he cannot carry it, the current title stands.

Underneath that sits the problem that really settles it. The statistical case rests on a counterfactual: this family would today hold this land and its accumulated value, had the taking not happened. Nobody can know that, because wealth does not sit still.

Consider Rockefeller — the largest fortune in America, built in one lifetime. Where is it now? Spread across roughly two hundred descendants, not one of whom appears near a rich list. Nothing was stolen. A century simply happened to it: division, sale, diversification, ventures that failed. The same forces run at every scale.

And one more thing, which is really the root of all of it. Groups do not hold title deeds. Sixty million separate people do, or do not, over particular pieces of ground. A statistic about land ownership by group is real as a statistic and is not a fact about any individual — inside any category are people who were dispossessed, people who never were, people who bought since, and people who arrived after. Act on the average and you get individual cases wrong in both directions. Some receive land that was never theirs. Some lose land they came by honestly.

Which brings us to the thing this is really about.

The half-measure is the problem. Not because a small taking is as bad as a large one — but because once compensation can be set below what a buyer would pay, the question of how much you own has no answer that does not depend on somebody’s judgement.

There is a sharper way to put it. Anyone who has not sold their land at market value, by definition, values it above market value — otherwise they would have sold. So even full market value sits below the owner’s own valuation. Market value is not generous compensation. It is the least dishonest number available. Cutting below it is not trimming excess. It is taking the difference.

A right whose extent is decided by the party who wants your land is not a weaker form of ownership. It is a different thing wearing ownership’s name.

And it applies to whoever receives the land afterwards on precisely the same terms. That is the part nobody says out loud. A title acquired through discretionary transfer is held exactly as contingently as the one it replaced — good only until the next official takes a different view.

That is not ownership. It is occupancy with permission.

If you own or lend against South African property

If you are reading this from outside South Africa

Do not file this under African politics. Your pension holds emerging-market debt and equity, and this is what “country risk” means once you unpack it — not the risk of confiscation, but the risk of not knowing. Capital leaves jurisdictions with unpredictable rules faster than jurisdictions with bad ones. Predictable bad rules attract more investment than unpredictable ones. Deeply unintuitive, and among the more reliable findings in economics.

Which is why the question to ask of any jurisdiction you hold assets in is never whether the state has a power to take property. It always does. The question is whether the compensation is tied to something discoverable, or to somebody’s opinion — and whether the formula that determines it has a floor.

None of this is financial or legal advice — it is commentary. Take your own advice on your own situation.

Start with the compensation formula, because that is where this begins.

Market value is not generosity to landowners. It is the only figure in this entire debate that is not somebody’s opinion. It is what two parties who could each have said no actually agreed. Every alternative on offer replaces a discoverable number with an assessed one — and once you have done that, the question of what you own has no fixed answer.

So: compensation at market value, determined before transfer rather than litigated afterwards through a mechanism its own drafters admit is circular. Claims about historical injustice heard as claims — a named claimant, a named property, a documented taking, the burden of proof on whoever brings it, and the state as a party to the case rather than the judge of it. No residual category of “and whatever else we decide.”

Do it that way and something useful happens. Property rights come out stronger. Every proven claim becomes evidence that title is real, that records matter, and that the state can be held to them.

And there is an awkward fact sitting underneath the whole debate. South Africa’s 2017 land audit found 28 million hectares — roughly 23% of the country — already in state ownership. Before any question of taking anything from anybody, the state holds close to a quarter of the land. What it does with that is a test of intent requiring no new powers, no constitutional argument and no court.

Stay free,

Jean-Pierre

The Austrian Actor

P.S. The full archive will always be available at theaustrianactor.com