The True Cost of "Cheap" Imported Fuel
It looked like economics while the tankers kept arriving — fuel bought cheap, no questions asked. Count the whole ledger and the verdict reverses.
Australia closed its refineries, sold its fleet, and ran its reserves down because importing fuel was cheaper per litre on a normal day. That price was correct, but it left out most of the costs: the reserves the country now has to hold, the freight and insurance, the roughly $59 billion a year (2023) that leaves the country, and the cost of a supply shock like 2026. This memo adds up those costs. Counted in full, importing is the expensive option. Producing fuel at home costs more upfront, keeps the money in the country, removes the risk, and builds an industry — and the same capability lets the country defend itself.
1. The decision, and the one number behind it
Between 2014 and 2021, Australia closed most of its oil refineries — Kurnell, Clyde, Bulwer Island, Altona, and finally Kwinana, the largest in the country. Over the same years it let its merchant fleet shrink to almost nothing and allowed its emergency fuel reserve to sit, year after year, below the level it had promised the world it would hold.
Each decision was defended with the same argument, and in its own terms the argument was sound. Australia's refineries were old and sub-scale; on the delivered cost of a litre under normal conditions, imports won. Keeping an Australian-flagged ship at sea cost more than chartering a foreign one. Holding a litre in a tank cost more than not holding it. On a calm day, every one of those numbers favoured the cheaper path, and the country took it.
The companies said as much at the time. Closing Kwinana — the largest refinery in the country — BP stated that the growth of large-scale, export-oriented refineries across Asia and the Middle East had structurally changed the market, and that "regional oversupply and sustained low refining margins" left the plant no longer economically viable; its Australia head added that the decision was "not in any way a result of local policy settings." ExxonMobil closed Altona, the smallest, as "no longer economically viable" after a review. Ampol's Lytton refinery lost $145 million in 2020 on depressed regional margins. The federal government summarised the Kwinana closure as based on "commercial and international factors, including the age of the refinery and overseas competition." Every reason given was a per-litre, single-firm margin reason. None of them weighed the cost to the nation of importing the lot — because that was nobody's job to weigh.
This memo does not dispute those commercial facts. The plants really were beaten on margin. The point is narrower: the firms optimised their own books, correctly, and no one ran the national ledger at all.
2. One number, doing the work of many
The price of a litre at the dock tells you what the fuel costs to buy. It tells you nothing about what it costs to keep that fuel arriving: the reserves the country has to hold, the shipping, the insurance, and the bill that lands when the supply is cut.
For a household filling up, the per-litre price is the whole decision — and rightly so. For a country deciding whether to make its own fuel or import all of it, that price is one number among many. Australia decided on the one number and left the rest out.
Set those costs side by side — importing versus producing here — and the picture changes. That is the comparison this memo runs, section by section, and pulls together in the ledger that follows.
3. The ledger
| Factor | Importing it (the status quo) | Producing it here |
|---|---|---|
| — COST — | ||
| Headline price per litre, calm day | Lowest — the number the closures were judged on | Higher today (conceded) |
| Annual spend, and where it goes | ~$59bn in 2023 (and rising) leaves the country, permanently | The same spend stays onshore and recirculates |
| Economic multiplier | ~Nil here — margin, wages, tax compound offshore | Wages → spending → tax compound at home |
| Reserve to hold | Must buy a stockpile (the $3.2bn tank, inside the $14.8bn package) | Far less to hold — the resource is the reserve |
| Security margin (90-day obligation) | Buy the whole margin as a standalone asset — $3.2bn reaches only 50 days; ~$20bn+ more to reach 90, then yearly holding cost | Carried inside domestic production; never separately bought |
| Inflation pass-through | Imported inflation, currency-multiplied | Domestic, AUD-priced, insulated |
| — RISK — | ||
| Price volatility | Full world price, no buffer | Domestic buffer, stable |
| Currency (FX) | USD-priced; AUD falls in the same shocks — hit twice | AUD-priced; no FX exposure |
| Freight + war-risk insurance | Thousands of km; spikes thousands-fold in a crisis | None — made and used here |
| Shipping / fleet | Foreign ships; zero Australian-flagged fuel tankers | Minimal shipping; sovereign control |
| Chokepoint / supply chain | Hormuz + Asian refineries + tanker availability | Immune to chokepoints |
| Tail risk (the shock) | Priced at zero; ~3–5c/litre hidden premium (§5) | Falls as the import task shrinks |
| — NATIONAL BENEFIT (the column imports leave blank) — | ||
| Jobs | Offshore | Construction + operations + regional |
| Tax & royalties | Foreign refiners pay foreign tax | Royalties + company + payroll tax, here |
| Industrial capability | None retained; refining skills lost | Permanent sovereign capability — an asset |
| Resource base / options | Treated as irrelevant — "we have no oil" | False: a spectrum of sovereign options (§6) |
| Defence / strategic | Dependent, vulnerable, slow to fix | Plants, rail and refineries are themselves defence assets |
| — THE HONEST COSTS of producing here — | ||
| Upfront capital | None — you just buy | Real and large — refineries, exploration, electrification, backbone |
| Build time | Immediate | Years |
| Carbon | Refining emissions offshore = no Australian carbon cost | Safeguard ~$37/t onshore — but counted and reducible |
| Environmental trade-off | Hidden offshore | Clean path preferred; unconventional fallbacks contested |
Importing wins one row: the headline price on a normal day. It loses or leaves blank every other row. The one number Australia decided on is the only number that favours importing.
4. The cost that leaves, and the cost that builds
Two rows carry most of the weight, because they are about where the money goes.
Australia spent about $59 billion importing petroleum in 2023 — roughly $51 billion in refined product and $8 billion in crude. The figure moves with the oil price and the dollar, and the volume underneath it keeps rising, but it is tens of billions every year, and it leaves the country entirely. It pays the refining margin, the shipping, the insurance, and the wages of other nations. It lands in no Australian payroll, no Australian plant, and no Australian tax return. Spent, and gone.
A dollar spent on domestic fuel buys the fuel and pays an Australian worker, an Australian supplier, and Australian tax. An imported dollar buys only the fuel; the wages, plant and tax go to the exporting country. Both deliver the litre. Only one builds anything here. That is what the per-litre price cannot show.
There is a second cost that only the importing country pays: the reserve. Because Australia makes almost none of its own fuel, it has to buy its entire security margin and store it, where a producing country carries that margin inside its own production for nothing. The obligation Australia signed up to as an IEA member is 90 days of cover. The country holds around 37, and the $3.2 billion reserve just funded reaches only about 50. Getting to 90 — the level Australia has been below since 2012 — means buying and storing on the order of six billion more litres of fuel: indicatively another twenty billion dollars and more in fuel and tankage, and then a permanent annual bill to insure, guard and cycle it so it does not degrade. That is the price of holding a margin you could otherwise keep in the ground. A producing nation never writes the cheque.
5. The number nobody computed: the price of the shock
The cost of a disruption is its probability times its impact. The decision to import all the fuel never priced it — the risk was simply left out of the calculation.
In 2026 the impact term arrived. When the Strait of Hormuz closed, Australia — holding the thinnest fuel reserves in the developed world — watched diesel pass three dollars a litre, forecourts run dry, and officials model rationing if stocks fell toward ten days. The government's response was a $14.8 billion Strengthening Australia's Fuel Resilience Package — and it is worth reading what sits inside that single headline, because the components tell the story and they are parts of the total, not additions to it. Of the $14.8 billion: $7.5 billion is an underwriting facility that secured around 740 million litres (740 ML) of diesel and 150 million litres (150 ML) of jet fuel the market would not have delivered; $3.2 billion builds a permanent government reserve — the "big tank," lifting cover to 50 days from barely 21; and $2.9 billion halves the fuel excise and zeroes the heavy-vehicle road-user charge for three months. The reserve drawdown, the relaxed fuel standards and the rationing modelling sit alongside. Every dollar of it bought, borrowed, stored or discounted existing fuel — the $3.2 billion tank most of all, which Section 6 takes up. None of it made a litre.
Put even a modest annual probability on an event of that size and the number stops being abstract. Australia burns on the order of 50 billion litres of transport fuel a year. A shock costing the economy in the order of $15 billion, occurring just once a decade, works out to a hidden risk premium of roughly three to five cents on every litre imported across that ten-year span — money the per-litre comparison never charged, because the risk it created was left out of the price. (The range reflects the spread between the headline crisis cost and a stricter, unbundled figure; either way the premium is real and recurring.) The cheap litre was only ever cheap because nobody put that line in the ledger.
The shock is not only a cost. It is also a transfer. When the oil price spikes, Australia's oil producers — who export more than ninety per cent of what they pump — earn more, while Australian motorists pay more for the imported fuel that comes back. The same chokepoint that drained the country's pumps was a windfall for the companies extracting its oil for export. Across 2026 the six largest oil majors were projected to earn around ninety-four billion US dollars for the year, about thirty-seven million dollars a day above the year before, with executives describing the volatility itself as a trading opportunity. The profit and the pain landed on opposite sides of the same event.
6. The reserve we already have, and the oil we are told we don't
Two objections have to be met, because the whole import case rests on them.
"We have to stockpile, and that's expensive." It is — but a built stockpile is the worst form of reserve. The $14.8 billion tank costs billions to fill, costs money every year to store, insure, and cycle, degrades because fuel does not keep, and drains: once you start drawing it down, the clock runs and nothing sits behind it. It is a depreciating asset you pay to hold and hope never to use. A reserve in the ground is the opposite on every count. Coal, gas, shale, and a standing canola crop cost nothing to "store," cannot be sunk or embargoed or turned back at sea, sit dispersed across the continent rather than concentrated in tank farms that are themselves a target, and never empty — you draw on them by producing, and the canola replenishes every harvest while the coal and gas are measured in centuries. A tank is a bucket. A resource is a tap. The country is spending billions to hold a worse version of a reserve it already owns.
"Australia doesn't have the oil." This confuses no cheap conventional oil at today's price with no resource. Australia holds some of the world's largest coal reserves, vast gas, and significant undeveloped shale basins. The technology to turn those into liquid fuel is proven and running commercially elsewhere: coal-to-liquids has fuelled much of South Africa's economy for decades, gas-to-liquids operates at scale in Qatar, and tight-oil fracking made the United States the world's largest oil producer. Australia has not developed these because, on a calm-day per-litre basis, importing was cheaper — the same incomplete number, again. The constraint is not geology. It is choice.
And the country has barely looked. Dorado, off the Pilbara coast in the lightly drilled Bedout Sub-basin, was the largest oil discovery in Western Australia this century — found in 2018 by a well that was chasing gas, in a basin that had seen around seventeen exploration wells against more than fifteen hundred in the basin next door. It is light, low-emission crude, headed for first oil around 2026 at up to 100,000 barrels a day, and on the operator's figures it pays back its build cost inside roughly a year. Dorado is not a lone hit: the same sub-basin holds a string of further oil and gas discoveries and leads — Pavo, Apus, Roc and others — and remains, on the explorers' own assessment, one of the most prospective and least-drilled petroleum provinces in the country. The Great Australian Bight, opened and then abandoned by the majors, is essentially untested. A nation that finds its biggest oil field in decades almost by accident, in a basin it had hardly drilled, has not run out of oil — it has stopped looking for it. The first step is to start: a serious, sovereign exploration effort across the frontier basins, begun without delay.
These options are not equal, and the honest path says so. The cleanest and best is to electrify the bulk of demand and make the residual from renewables and the nation's own biomass — sovereign and low-emission. Coal-to-liquids and shale are higher-emission and contested, and they are the fallback, not the plan. But their existence settles the point that matters: a nation that can make fuel from coal, gas, shale, sun, wind, and its own canola crop does not lack options. It lacks the decision to use them.
7. The honest other side
A full-cost case is worthless if it ignores what the import case gets right, so let it be said plainly. The refineries really were uncompetitive. Reopening them would not fix the dependency, and not only because they run on imported crude: Australia's legacy plants were engineered decades ago for light, sweet domestic crude, while the global slate has shifted heavier and more sour, so the surviving refineries are configured for feedstock the country no longer produces in quantity and must import to suit. "Reopen the old gates" is therefore a fantasy twice over — the plants are uncompetitive and mismatched to today's crude. That is exactly why the viable domestic path is not old-style refining but a structural pivot to converting the resources Australia does have: electrify the demand, and make the residual from biomass, gas and renewable power. Domestic fuel of that kind costs more per litre today, and not every litre can be replaced quickly; the domestic path carries real upfront capital and real build time, measured in years.
The pivot also does not mean rebuilding a refinery on the old scale. The choice was never a quarter-million-barrel mega-plant or nothing. Modular refineries — skid-mounted process units built in a factory and bolted together on site — reach completion in well under a year rather than five, at a fraction of the upfront capital, and scale in small increments. A simple topping or condensate-splitter configuration is matched to exactly the light condensate Australia produces and currently exports, rather than the heavy imported crude the old plants were built for. They are honest about their limits: small refineries have weaker unit economics and a narrower product slate, so they are not cheaper per litre than an Asian mega-refinery. What they offer is speed, low upfront cost, flexibility, and a match to the crude the country actually has — built where the oil is, sized to it, and added to as demand requires.
Producing at home is not free. But the money builds something in the country and removes the risk. Importing does neither, at any price.
8. The comparison, run properly — and what it builds
Run the ledger with every column filled, and the result inverts. The import path is cheapest on the one calm-day number and dearer on all of them once you count the outflow, the risk, and the benefit foregone. The production path costs more upfront and returns that spend to the country, builds an industry, and retires the risk. Over a decade, the investment pays itself back twice — in money that stays home instead of leaving, and in crises that do not happen.
And the spending does not buy fuel alone. The same capital builds refineries and processing, opens exploration, stands up a biodiesel industry across regional Australia, electrifies the freight task, and lays the corridor backbone — onshore, now, employing Australians. This is not an expense to be minimised. It is a New Deal to be kick-started: the moment a nation turns a permanent loss into permanent capability.
It is also defence. A country that can fuel and move itself can defend itself; one that imports nine litres in ten and owns none of the ships cannot. The fuel plant, the refinery, the electrified line, and the reserve in the ground are strategic assets. The comparison with the submarine program is direct. As set out in the companion memo The AUKUS Cost Blowout, that program is costed in the hundreds of billions over its life, with boats arriving through the 2040s. At about $59 billion in 2023 and rising, a single decade of imported fuel runs to $500–600 billion or more — more than the entire lifetime cost of the submarines — and it is spent every year, offshore, building nothing here. The same money put into fuel and energy capability secures the industrial base now, not in the 2040s.
9. The decision rule
The lesson is not really about fuel. It is about how a nation should evaluate any choice between making something and importing it: not on the price of the thing on a calm day, but on the total system cost and the total national benefit — the reserves, the volatility, the leakage, the tail risk, and what gets built or lost at home.
Australia was warned. The risk of import dependence was flagged in national security assessments in 2009 and 2011; a refinery-loss supply shock was modelled in 2011; a parliamentary committee asked for a fuel-security review in 2018; an interim review was published in 2019 and the final report was never released. Across that decade and a half, under both major parties, the refineries closed, the fleet was sold, and the reserve stayed below the line. The warnings were not missing. The full-cost analysis was.
How a country talks itself into this is not a mystery. Three forces pushed the same way. Decisions were made on narrow financial measures — the delivered cost of a litre, the return on a plant, the next result — by firms and treasuries each optimising its own books, with no one accountable for the national whole; every closure was locally rational and nationally costly. A carbon policy added to it: it priced the two domestic refineries for the emissions they produced onshore, while the imported litre's emissions, made in Singapore or Korea, carried no Australian cost at all — so it raised the cost of producing here and moved the emissions offshore rather than cutting them. And the working assumption underneath all of it was that demand would fall and imports would always be there. Britain made the same choices on the same logic, closed most of its refineries, and ended up just as exposed. None of the three forces set out to hollow out the country's fuel supply. Together, and accountable to no one for the sum, that is what they did.
A bigger reserve, a few new ships, and eight submarines each buy time. None of them fixes the dependency, and none of them builds anything in the country. Producing what we use does both. The number that closed the refineries was never the full number. The full number says build it here.
References
- Refinery closures and timeline — Australian Petroleum Statistics (DCCEEW/Department of Industry); Liquid Fuel Security Review interim report, April 2019.
- Stated reasons for closure (companies’ own words) — BP media statement on Kwinana, 30 October 2020 ("regional oversupply and sustained low refining margins… no longer economically viable"; "not in any way a result of local policy settings"); ExxonMobil statement on Altona, February 2021; Ampol 2020 results (Lytton loss of $145m); Minister for Energy media release on Kwinana, 2020 ("commercial and international factors, including the age of the refinery and overseas competition").
- Import dependence (~90% of refined product; ~71% of onshore-refined oil is imported crude) — Australian Petroleum Statistics; IEA.
- Annual petroleum import bill — ~A$59bn in 2023 (refined product ~A$51bn + crude ~A$8bn), price- and FX-dependent and rising with volume — DFAT trade statistics / Australian Petroleum Statistics (DCCEEW).
- Crude/condensate exports (~94–96% exported; A$13.2bn earnings, 2022–23) — Geoscience Australia, Australia's Energy Commodity Resources.
- 2026 crisis response — Budget 2026–27 fuel-security papers; ministerial releases (the $7.5bn Fuel and Fertiliser Security Facility; ~740ML diesel / ~150ML jet secured; excise halving and road-user-charge pause; ~20% reserve drawdown; $14.8bn resilience package).
- Reserve position (~37 days; high of 310 days in 2002; IEA 90-day obligation unmet since 2012; member average ~140) — IEA oil-stock data; DCCEEW; contemporary reporting, 2026.
- Carbon treatment — Safeguard Mechanism (DCCEEW): default price ~A$36.82/t (interim to March 2026); A$75/t ACCU price cap; applies to facilities over 100,000 t CO₂-e; government leakage review recommended a border carbon adjustment for import-exposed goods.
- Resource base and conversion technology — coal-to-liquids (Sasol, South Africa); gas-to-liquids (Shell Pearl, Qatar); tight/shale oil (US tight-oil production); Geoscience Australia resource assessments (Cooper, Canning, Beetaloo basins).
- Canola/biodiesel reserve (5.84Mt in storage ≈ 2.2bn L; ~34–37 days diesel) — MMP fuel-crisis letter series (Letter 7, 31 March 2026), drawing on ABARES crop data.
- Warnings timeline — National Energy Security Assessments 2009 and 2011; ACIL Tasman Liquid Fuels Vulnerability Assessment 2011; PJCIS recommendation 2018; Liquid Fuel Security Review interim report 2019 (final never released).
- Oil-producer windfall during the 2026 shock (six majors projected ~US$94bn for the year, ~US$37m/day above 2025; volatility as a trading opportunity) — Oxfam International analysis, 2026; Fortune; company Q1 2026 results.
- Modular/mini refineries (skid-mounted, <12 months to completion vs ~5 years, lower upfront capital, 1,000–40,000 b/d, topping/splitter suited to light condensate; weaker unit economics than mega-refineries) — industry sources (Honeywell UOP modular trains; Howe Baker; modular-refinery technical literature).
- UK refinery closures and import exposure — UK refining-capacity decline, parallel trajectory.