What SA’s draft fuel stockpile policy reveals about the state that wrote it

On 9 July, the Department of Mineral and Petroleum Resources gazetted its draft Strategic Petroleum Stocks Policy for public comment. It is the first serious attempt to rebuild South Africa’s fuel reserves since the apartheid government carved crude oil caverns into the rock at Saldanha Bay in the 1970s. After thoroughly analysing the document, what I found was a policy with the right diagnosis, a workable architecture, and a text riddled with internal contradictions.

Why now?

The trigger is not mysterious. The war involving Iran has choked traffic through the Strait of Hormuz for months, oil prices have spiked, and South Africans queued at petrol stations in March while the government cut the fuel levy to soften the blow. The Cabinet approved the draft for publication on 1 July. Minister Gwede Mantashe had already trailed the numbers at the fuels industry imbizo in June.

The deeper story is structural, and that is the one I care about. South Africa has lost roughly half its refining capacity in recent years. South African Petroleum Refineries and Engen’s refinery in Durban have been closed or converted into import terminals.

With the country now importing most of its finished fuel, the diesel powering a Free State maize tractor has already spent three to six weeks at sea before reaching port. The draft policy states this plainly: imports take 21 to 42 days to reach South African ports of entry, then another 10 to 14 days for offloading and inland transport. Once self-sufficient enough to refine its own fuel, the country’s energy security is now at the mercy of shipping timetables.

This is the textbook logic of weaponised interdependence. The networks that make trade cheap in peacetime become instruments of pressure in crisis, and the states that sit far from the chokepoints but depend on flows they do not control feel it first.

Pretoria built its last stockpile because the world cut it off during apartheid. It is proposing this one because the world’s oil arteries can be squeezed by someone else’s war. The rediscovery of siege logic without a siege is, to my mind, the most interesting thing about this document.

What the policy actually proposes

The core architecture is a dual-obligation model. The state, through the South African National Petroleum Company, holds strategic stocks at Saldanha Bay and Milnerton. Licensed wholesalers and importers carry a mandatory buffer of their own, at their own cost, ending the voluntary regime that left the state as sole guarantor of supply.

Release is tiered: voluntary industry stock-sharing when 20% of refined supply is lost for more than two weeks, restricted drawdown to essential services at 40%, and mass release with fuel rationing when a ministerially declared national emergency takes out more than half of supply. There is also an economic trigger, a strategic sale of stock if prices hit $145 per barrel.

Internal contradictions

So far this strategy is defensible. However, the numbers start fighting each other.

The executive summary says the government will hold 90 days of net imports, primarily as crude at Saldanha, with industry holding an additional 14 days of refined product. The obligation table four pages later says 60 days for the state and 21 days for industry, each split into 70% crude and 30% refined product. The infrastructure table repeats the 60-plus-21 formula as the national target.

Policy Statement 1, the actual text the Cabinet is asked to adopt, requires private companies to hold “a minimum of 21 days of refined product cover” on a 70-to-30 split between crude and refined product. Read that sentence twice. Refined product cover held mostly as crude oil is not refined product cover. It is a contradiction inside a single policy statement.

The press coverage diverged along exactly these fault lines. Reuters reported 60 days state and 21 days private; BusinessTech, reading the executive summary, reported 90 days state and 14 days private. Both were faithful to the document. The document was not faithful to itself.

Section 7.1 attempts to model the economic cost of a two-week disruption, and here the wheels come off entirely. The document states that South Africa consumes “approximately 30 million litres of petroleum products annually”. Three sections earlier, the same document gives the correct figure of 27 billion litres. That is a three-orders-of-magnitude error, and the calculation built on it produces a total cost of R48-million for a two-week national fuel disruption, a figure the text then describes as roughly 0.7% of GDP. It is not. R48-million is roughly 0.0007% of GDP.

The bullet points inside the same calculation cite R4-billion in consumer effects and R4-billion in indirect losses, which cannot amount to R48-million under any arithmetic I was taught. Elsewhere, the document uses the sensible R1-billion per day estimate, except in the financing section, where the number has simply fallen out of the sentence: the cost of a fuel-less day is “estimated at R Billion per day”.

I do not raise these as linguistic pedantry. This document went through the Department of Mineral and Petroleum Resources, through the Cabinet, and into the Government Gazette without anyone catching errors that my first-year students would lose marks for. And that matters analytically, because a stockpile is not a policy document. It is a multibillion-rand physical asset that must be procured honestly, rotated every three months, guarded against the people managing it, and released competently under crisis pressure by a state-owned company that has existed for barely a year.

The state’s demonstrated capacity to produce 20 coherent pages is a leading indicator of its capacity to manage 36 million barrels of fuel. On this evidence, the indicator is flashing.

The farm angle

Here is what I did not find anywhere in the document: agriculture, except as a line item in the broken cost model, where a two-week fuel disruption supposedly costs the sector R2-million. Anyone who has tried to plant hundreds or even thousands of hectares in a 14-day weather window knows that the number is off by a large factor. Diesel is not only an input in commercial agriculture. It is the input. Cultivating, planting, spraying, harvesting and the entire cold chain run on it, and these agricultural functions work to a calendar that does not negotiate.

The policy does acknowledge that 60% of national fuel consumption is inland, in Gauteng, Limpopo, North West and the Free State. That is the food and logistics belt of South Africa. But the Level 2 trigger, the restricted release to “essential services and key economic hubs”, never defines essential services.

Is October planting in the Free State essential? Is the citrus cold chain in a Hormuz-style squeeze? The document does not say, which means the answer will be decided in a crisis by a minister under pressure from whoever lobbies loudest. We’ve seen similar pressure situations take down an agriculture minister in John Steenhuisen. My judgement call: if the agricultural sector does not put itself into that definition during the comment window, it will not be in the definition when it matters.

The crude-heavy split compounds the problem. Both the state and private buffers are mandated at 70% crude oil in a country that has lost half its refining capacity. Crude in a tank at Saldanha does not put diesel in a tractor at Bothaville. The Moerane Investigating Team identified crude-only stockholding as a core failure after the 2005 fuel shortages. The document itself recounts this history, and then the same document proposes a mix that is still 70% crude. Policy Statement 5 gestures at shifting toward finished products to bypass non-operational refineries, which quietly concedes the point the tables ignore.

What I make of it

The diagnosis is right. Import dependence plus maritime chokepoints plus a just-in-time private sector equals a country running on faith and a shipping timetable. The dual-obligation architecture is broadly sensible and matches international practice. Requiring the private sector to carry part of the burden is correct and overdue, though motorists should understand that cost will find its way to the pump.

But a stockpile policy is a bet on state capacity, and this document is an unusually candid disclosure of the current level of that capacity. The numbers contradict each other, the model does not compute, and the custodian institution carries the institutional DNA of the 2015 fire sale. That sale moved 10 million barrels at $28 a barrel when the market price was between $36 and $45. The justification given was that the oil was degrading in the tanks. The same degradation logic now underpins the rotation requirements in this draft. Whether this policy becomes a genuine buffer or another asset waiting to be stripped depends entirely on execution details the draft leaves blank: the funding instrument, the audit regime, the definition of essential services, the enforcement of private obligations.

The comment window runs for 30 days from 9 July. I intend to use it. The gap between a state that can write “R Billion per day” and a state that can defend the food system against a supply shock is measured in exactly the kind of detail a public consultation exists to fix. Reports suggest the fuels industry is preparing its own submissions. Agriculture should be in that queue, and near the front of it.

The oil is real. The chokepoints are real. The question this document cannot yet answer is whether the state is. DM