The live sheep export phase-out

The second truck is the problem. A livestock carrier in the eastern wheatbelt of Western Australia — invented here, but assembled from circumstances that are entirely ordinary in that part of the state — bought his first prime mover and crate in his late twenties and his second a decade later, in 2019, when the work was steady enough to justify it. The second truck came with a loan, and the loan came with the assumption that the run he had been doing for fifteen years would still exist in some form. That run is sheep, from the assembly points and feedlots of the Great Southern down to the wharf at Fremantle. It is not the only thing he carts. It is, in a normal year, the thing that pays the second loan.

From 1 May 2028 that run does not exist. Not a smaller version of it, not a version with tighter rules and more paperwork: none of it. The Export Control Amendment (Ending Live Sheep Exports by Sea) Act 2024 prohibits the export of live sheep by sea from Australia from that date. His trucks will still be worth something. His licence will still be valid. His customers, mostly, will still be farming. But the specific stream of work that services the specific loan he signed in 2019 will have been legislated out of existence by a parliament sitting four thousand kilometres away, and no amount of working harder will bring it back.

That is the honest human stake in this policy, and it is worth stating precisely, because almost everything said publicly about the live sheep export phase-out overstates it in one direction or the other. On the evidence, this is not a large economic event for Western Australian agriculture. The modelled hit to the industry is around $123 million and roughly 400 jobs (Western Australian state government modelling, 2024), on a trade that accounts for less than one per cent of the state's agricultural output. Those are small numbers by the standards of the sector. Nor is it a costless act of moral tidying-up, as some of its supporters imply. For the carrier with two trucks, for the shearing contractor whose run is built around export lines, for the small feedlot near the port, the aggregate is irrelevant. Their exposure is not one per cent. It is all of it.

What the government did

The commitment was made before the 2022 election and delivered in the term that followed. In May 2024 the Australian Government announced that the export of live sheep by sea would end on 1 May 2028, with an initial transition assistance package of $107 million announced in the 2024-25 Budget (Budget Paper No. 2, May 2024). The legislation followed within months: the Export Control Amendment (Ending Live Sheep Exports by Sea) Act 2024 passed the Parliament in 2024, inserting the prohibition into the Commonwealth's export control framework and fixing the commencement date in statute rather than leaving it to regulation. The transition package was subsequently expanded to approximately $140 million (Department of Agriculture, Fisheries and Forestry, transition assistance program materials, 2024–2026).

Four features of the design matter and are frequently misdescribed, so they are worth setting out plainly.

First, the ban is confined to sheep, exported by sea. Live cattle exports by sea are unaffected. Live goat exports are unaffected. Breeding stock and animals exported by air are outside the prohibition. The Australian live export trade continues after 1 May 2028; it is one leg of it that stops.

Second, the mechanism is a hard statutory date, not a licence wind-down or a quota taper. From 1 May 2028 the activity is unlawful. There is no residual permit, no transition licence, no reduced volume. That design choice is deliberate and it has consequences in both directions: it removes the uncertainty of a discretionary phase-down, and it removes any possibility of partial survival for a business that cannot pivot.

Third, the notice period is four years. The date sits four years after the announcement, which places it well beyond the electoral cycle in which it was legislated and gives every affected business a known, fixed horizon. By the standards of Australian industry adjustment — where regulatory change frequently arrives with twelve or eighteen months' notice, and occasionally with none — four years is a long runway.

Fourth, the decision followed a formal review rather than an announcement. An independent panel appointed by the Minister for Agriculture, Fisheries and Forestry consulted across the industry and reported in October 2023 on how, not whether, a phase-out should be implemented, recommending a transition period and an accompanying assistance package (Independent Panel on the Phase Out of Live Sheep Exports by Sea, report to the Minister, October 2023). The Bill was then referred to the Senate Rural and Regional Affairs and Transport Legislation Committee, which took submissions from producers, exporters, processors, veterinarians and animal welfare organisations before reporting in 2024. The parliamentary record on this policy is unusually complete.

The evidence: a small aggregate sitting on a very narrow base

Start with the size of the thing, because the public argument is conducted almost entirely without it.

The Western Australian state government's modelling of the phase-out put the impact on the industry at approximately $123 million and job losses at approximately 400 (WA state modelling, 2024; this is state government modelling, not Commonwealth Treasury analysis, and should be read as one estimate among several). Set against Western Australian agriculture as a whole — an industry with a gross value of production in the order of ten billion dollars a year — the live sheep trade represents less than one per cent of output. Set against the national economy, $123 million is a rounding error: it is roughly the cost of a few hundred metres of urban motorway, or about one two-hundredth of the annual growth in the Commonwealth's public debt interest bill.

The trade's decline is the second fact that has to be on the table before anything else is argued. At its peak in the late 1980s Australia shipped more than six million live sheep a year. By the early 2000s that was in the order of four million. In recent years it has run well under one million head, with almost all of the remaining volume leaving from Western Australia (approx.; ABARES and DAFF livestock export statistics, various periods). The export value in recent years has been well under $100 million annually (approx.), against Australian sheepmeat and wool exports measured in billions.

A reader who stops there will conclude the policy is trivial, and a great many commentators have. That conclusion is wrong, and the reason it is wrong is the single most useful idea in this article.

Aggregate is not incidence

An aggregate tells you the total size of a loss. Incidence tells you who bears it. They are different questions, and for a policy like this one they give almost opposite answers.

A $123 million loss spread evenly across the 2.9 million people of Western Australia would be about $42 each — an afternoon's inconvenience. But the loss is not spread evenly. It is concentrated on a set of businesses that is small enough to count: the producers in the wheatbelt and Great Southern whose flock composition and lambing calendar are built around export lines rather than domestic processing specifications; the handful of licensed exporters; the feedlots and assembly depots near the port; the shearing contractors and stock agents whose annual run is timed to export shipments; and the livestock transport operators, many of them one- or two-truck businesses, whose route structure exists because the wharf exists.

For those businesses the correct denominator is not Western Australian agriculture. It is their own turnover. And for the ones most exposed, the exposure is not one per cent. It is a large fraction of everything, up to and including all of it.

A cost that is one per cent of a state can be one hundred per cent of a business, and the two facts are not in conflict.

This is not a rhetorical point. It is the reason national-aggregate reasoning systematically misjudges policies of this shape, and the misjudgment runs in a predictable direction. Aggregate reasoning treats a loss as small when it is small relative to the whole. But the whole does not bear the loss; individuals do. A policy that costs one hundred people everything and everyone else nothing produces a modest aggregate and a catastrophic incidence. A policy that costs everyone a small amount produces the same aggregate and almost no hardship. Cost-benefit analysis conducted at the national level cannot distinguish between those two worlds, because the arithmetic that makes it tractable — adding everyone's gains and losses into a single number — is precisely the step that discards the information about who.

There are three specific reasons the concentrated party fares worse than the aggregate implies, and they compound.

The first is the absence of insurance. A diversified household absorbs a small loss out of savings. A business whose single line of work is removed has no internal diversification to absorb it with, and there is no market from which it could have bought insurance against a change in the law. Regulatory risk of this kind is uninsurable in practice.

The second is asset specificity. Capital that serves one purpose cannot be redeployed at full value. A stock crate configured for sheep, a feedlot sited for proximity to a port, a shearing plant, a flock bred for a particular market's specifications: these are worth substantially less in any alternative use, and in a thin regional market, worth less again because everyone is selling the same asset at the same time. The 2028 date, which is a gift in every other respect, guarantees that a large number of similar assets hit the market in a similar window.

The third is geography. A worker who loses a job in a metropolitan labour market can look for another one without moving house. A shearing contractor in a wheatbelt town of eight hundred people faces a labour market of that town, plus whatever is within a two-hour drive. Adjustment costs that economists model as frictional are, at that scale, the choice between staying and going.

The compensation arithmetic, and what it does and does not tell you

Now the number that makes this policy genuinely unusual. The transition assistance package is approximately $140 million (DAFF, 2024–2026, following an initial $107 million announced in the May 2024 Budget). The modelled industry impact is approximately $123 million (WA state government modelling, 2024). The compensation exceeds the modelled harm, by something in the order of fourteen per cent.

That is rare. Australian industry adjustment more commonly delivers a fraction of the modelled loss, and frequently delivers nothing at all — regulatory change that reduces the value of a business is generally treated as a risk of doing business rather than an acquisition of property requiring compensation. On the headline arithmetic, this package is more generous than the norm, and the government is entitled to say so.

But a ratio is not a design, and three questions have to be asked before the ratio means anything.

Are the two numbers measuring the same thing? Not exactly. The modelled impact is a state government estimate of the economic effect on the industry, produced with a particular set of assumptions about substitution, sheep prices and the counterfactual rate of decline. The package is a Commonwealth appropriation covering a wider set of purposes, some of which — market development, processing capacity, animal welfare research, community support programs — are not compensation for a private loss at all. Comparing them directly is comparing a modelled economic aggregate with a budget line. The two numbers are close enough to be interesting and different enough that the fourteen per cent margin should not be treated as a genuine surplus.

Does the money reach the people who bear the loss? This is the question that decides whether the package is good policy or good arithmetic. The published program architecture is built substantially around grant rounds, industry programs and support services rather than a single direct payment to affected businesses (DAFF transition assistance materials, 2024–2026). Grant-based delivery has real virtues: it can fund genuine transformation, it avoids paying people who were leaving anyway, and it produces assets that outlast the payment. It also has a well-documented failure mode. Grant programs reward applicants who can write applications. They favour larger, better-resourced, better-advised businesses over the two-truck operator who has never filled in a Commonwealth grant form. They frequently exclude the parts of the supply chain that are not the primary producer — the contractor, the carrier, the depot — because eligibility is drawn around the industry as it is conventionally defined rather than around the incidence of the loss.

Was the number calibrated analytically or politically? The honest answer is that it is not possible to tell from the outside, and the sequence is at least suggestive. The package was announced at $107 million alongside the policy, and subsequently grew to approximately $140 million. An amount that rises after announcement, in response to industry and parliamentary pressure, is behaving like a negotiated figure rather than a calculated one. That is not a scandal — negotiated compensation is how most political systems settle these questions — but it means the reassuring fact that $140 million exceeds $123 million is a coincidence of two independently-derived numbers, not evidence that the loss was measured and then covered.

The critique: what the design gets wrong

The critique of this policy is not that it is expensive, because it is not. It is that a policy with an unusually small aggregate cost and an unusually generous headline package can still fail the people it was designed to protect, in three specific ways.

1. Compensating an industry is not compensating the losers

The deepest design flaw is a mismatch between where the money is aimed and where the loss lands. The package is oriented towards the sheep industry — producers, processing capacity, market development, sector transformation. The loss is distributed across a supply chain that is wider than the sheep industry as conventionally defined.

Consider the carrier from the opening of this article. He is not a sheep producer. He does not hold an export licence. He is not a processor. On most conventional definitions of "the industry", he is not in it — he is a road transport business that happens to derive a large share of its revenue from one commodity flow. The same is true of the shearing contractor, the stock agent, the depot operator, the rural mechanic whose workshop revenue tracks the shipping calendar, and the small-town business whose customers are all of the above.

A well-designed adjustment package would follow the incidence: identify the businesses with high revenue concentration in the affected activity, verify the concentration against their own records, and compensate proportionately to demonstrated exposure. That is administratively harder than running grant rounds, and it requires the government to accept that some of the money will go to businesses that are not photogenic representatives of the sector. But it is the only design that matches the money to the harm. Where a package is instead organised around a sector, the reliable outcome is that the best-connected participants in that sector capture a disproportionate share and the peripheral, highly-exposed businesses capture little.

This is a testable proposition, and it should be tested. The relevant evidence — how much of the $140 million was ultimately paid, to whom, and against what measure of exposure — will exist in departmental program reporting and is a proper subject for the Australian National Audit Office. Until that evidence exists, the claim that affected parties have been made whole is an assertion, not a finding.

2. The compensation arrives as programs; the loss arrives as a date

There is a timing mismatch as well as a targeting one. The loss is discrete and dated: on 1 May 2028 a revenue stream stops. Grant funding is distributed across rounds over several years, against applications, with acquittals and milestones.

For a business carrying debt, the difference between a certain payment and a possible grant is the difference between being able to plan and not. A carrier deciding in 2026 whether to refinance, sell a truck, or take on a different contract needs to know what he will receive and when. If the answer is that he may be eligible for a program that opens next year, subject to criteria not yet published, he cannot plan, and the four-year notice period — the single best feature of this policy — is partly wasted. Notice is only valuable to the extent the person given it can act on certain information. Four years of warning attached to four years of uncertainty about the compensation is worth considerably less than the calendar suggests.

3. Regulation was working, and prohibition foreclosed it

This is the strongest genuine critique available, and it deserves to be argued properly rather than asserted.

The regulatory response to the 2018 findings was substantial. The Australian Standards for the Export of Livestock were rewritten, with ASEL 3.0 taking effect from 2020, tightening stocking density requirements — the space each animal must be allocated — along with ventilation, feed and water provisions. A prohibition was imposed on exporting sheep to or through the Middle East during the northern hemisphere summer, the period the McCarthy Review identified as carrying the greatest heat-stress risk. Independent observers were placed on voyages, with their reports published. The Exporter Supply Chain Assurance System, in place since 2011, already required exporters to demonstrate control and traceability of animals through the destination supply chain, with published investigations into non-compliance.

And the measurable outcome moved. Reported voyage mortality rates for sheep fell substantially following those reforms, to levels well below one per cent and in recent reporting periods to a small fraction of that (approx.; DAFF reports to Parliament on the live animal export trade). Mortality is an imperfect proxy for welfare — an animal can suffer badly and survive the voyage, and critics of the trade make that point correctly — but it is the primary published indicator, it is measured consistently, and it moved decisively in the right direction.

If the regulation was working, the case for prohibition has to rest on something other than the failure of regulation — and the government has never quite said what.

The government's answer, as far as it is discernible from the parliamentary record, is that the trade's inherent characteristics — long voyages, high stocking densities, destination markets outside Australian jurisdiction — make an acceptable standard unachievable regardless of regulation, and that the community had settled its view. That is a coherent position. But it is a different argument from "regulation failed", and the two are frequently blurred together in public advocacy for the ban. If regulation was in fact delivering measurable improvement, then the case for prohibition is a judgment that the residual risk is unacceptable in principle, not a finding that the regulatory approach was exhausted.

That distinction matters because it changes what other industries should expect. A government that prohibits after regulation demonstrably fails is applying a standard other sectors can understand and plan around. A government that prohibits while regulation is improving is applying a different standard — one where measurable compliance does not secure a licence to continue operating. Both are legitimate exercises of parliamentary power. They are not equally predictable, and predictability is what investment depends on.

What they got right

Three things on this policy were done well, and each should be said plainly before the risk section, because the critique above is only credible if the credit comes first.

The notice period is genuinely generous. Four years, fixed in statute, announced in advance, and deliberately placed beyond the term in which it was legislated. Compare that with the standard experience of Australian industry adjustment: regulatory changes that arrive with a year's notice, licence conditions altered at short notice, and — in this very trade — a suspension imposed in 2011 with effectively no notice at all, later found by the Federal Court to have been unlawfully made. Whatever else is arguable here, the government did not do that. Producers have been given a known date and years to act on it. That is not a small thing; it is the single most important determinant of how well an affected business comes through.

The package exceeds the modelled harm. On the headline numbers — approximately $140 million against approximately $123 million — the Commonwealth has appropriated more than the modelled impact. The caveats set out above are real: the two figures measure different things, and delivery design determines whether the money reaches the right people. But most regulatory change in this country delivers affected parties nothing, on the principle that no one has a property right in the continuation of a legal market. Against that baseline, funding the transition at all reflects a defensible view that when the community decides to end an activity for the community's reasons, the community should carry the cost rather than leaving it with whoever happened to be operating.

The process was a process. An independent panel reported in October 2023 on how to implement a phase-out. A Senate committee took submissions and reported on the Bill. The commencement date was placed in primary legislation, which means it can only be changed by another Act of Parliament — a higher and more visible bar than a regulation the executive can amend. This was not a snap announcement, and characterising it as one is not accurate.

The honest caveats on that credit are equally clear. Four years of notice attached to an unresolved compensation design is worth less than four years of certainty. A package that exceeds modelled harm in total may still under-compensate the most exposed participants and over-compensate the least. And a thorough process that consulted extensively on how to phase out the trade is not the same as a process that tested whether a continued regulatory path could have achieved an acceptable outcome — the panel's terms were framed around implementation, and that framing settled a question rather than examining it.

The precedent, and what it costs

The dollars in this policy will be forgotten within a decade. The precedent will not.

Australia is entering a long period in which similar decisions will be made repeatedly. Land use, water allocation, native forestry, emissions-intensive processing, land clearing rules, animal welfare standards in other sectors: each involves a legal activity that a democratic majority may decide to restrict or end, with benefits spread thinly across the population and costs concentrated on a small number of identifiable businesses in identifiable places. The live sheep phase-out is the cleanest recent example of that structure, and it is small enough to reason about clearly. Whatever practice it establishes will be cited.

The practice it establishes is currently ambiguous, and that ambiguity is the real problem. On the generous reading, the precedent is: a government that ends a legal industry gives four years' notice, funds a package that exceeds the modelled harm, and legislates the date so it cannot be moved by executive fiat. That would be a good precedent, and other industries facing similar pressure would be right to take some comfort from it. On the sceptical reading, the precedent is: a government ends a legal industry, announces a headline compensation figure that is negotiated upward under pressure, delivers it through grant programs that reach the well-organised and miss the most exposed, and never publishes an assessment of whether affected parties were actually made whole. That is a much worse precedent, and it is worse precisely because the headline number makes it look generous.

Which reading is correct will be determined not by the legislation but by the acquittal — by what the departmental program reporting and, in due course, any audit of the transition assistance actually shows about where the money went. That evidence does not yet exist in public form. Anyone claiming today that the affected parties have been fairly compensated is making an assumption, and anyone claiming they have been abandoned is making a different one. The correct position is that the question is open and answerable, and that it should be answered.