What is the but-for clause? The test that decides factory incentives

The “but for” test asks what a manufacturer would do without public support. It is the counterfactual behind many discretionary grants and tax credits, and it can determine both whether support is awarded and how much is justified. Real cases from Ohio, Germany, Hungary and Canada show that contemporaneous evidence of a live commercial choice matters more than a list of eligible costs. Formula-based schemes, such as India’s electronics Production Linked Incentive programme, work differently.

Anduril Industries was not asking Ohio to refund a machinery invoice.

The US defence-technology company’s Arsenal-1 advanced-manufacturing project received an estimated $452.3 million, 30-year Job Creation Tax Credit in January 2025. JobsOhio, the state’s private economic-development organisation, subsequently executed a $310 million grant. The obligations were formidable: 4,008 jobs, $530 million in annual payroll and at least $910.5 million of investment, extending across the 30-year agreement.

That scale makes people reach for the obvious explanation. Ohio must have liked the equipment. Or the jobs. Or both.

But Ohio law asks a more difficult question. Its Tax Credit Authority must determine that the credit is a “major factor” in beginning, continuing or completing the project. This is the central thesis: in discretionary manufacturing incentives, public money buys changed decisions, not eligible spending.

Anduril’s Ohio project shows how a discretionary package is tied to long-term jobs, payroll and investment obligations. Photo: Sergey Platonov / Pexels, Pexels licence (free commercial use).

That sounds like a technical distinction. It is not. It separates a credible incentive application from an expensive request to subsidise a decision already made.

The question hiding behind the grant

Imagine you have approved a new factory line. The equipment is eligible. The construction is eligible. New jobs are eligible. You have even found a programme that advertises a generous percentage of qualifying costs.

Now the public funder asks: what would have happened without us?

Would the project have gone elsewhere? Been smaller? Arrived later? Not happened at all?

That is the “but for” test, also called additionality, inducement, financial need or, in Ohio, a major-factor test. It creates a counterfactual, meaning the plausible no-support version of events. The point is not to demand a tale of corporate distress. A profitable company can have a genuine choice between locations, project sizes or launch dates.

Most companies treat incentives as a catalogue. Find eligible spending. Claim a percentage. That works in some formula programmes. It is the wrong mental model for many discretionary awards.

Public funders use additionality tests for a reason larger than avoiding an embarrassing windfall. They need to identify the public outcome being purchased, then establish the smallest support that changes the decision. Without that discipline, an authority cannot distinguish between money that creates jobs, investment or production in its area and money that merely improves the economics of activity that was coming anyway.

This is where most people stop looking. They ask whether a cost qualifies. The better question is whether the funder can see a believable fork in the road.

Timothy Bartik’s July 2018 review of 34 estimates for the Upjohn Institute found that typical US incentives tipped only 2% to 25% of incented location, expansion or retention decisions. In at least 75% of cases, a similar decision probably would have occurred anyway. That finding explains the suspicion built into discretionary schemes. The project may be real. The spending may be real. The claimed influence of the subsidy may still be weak.

Anduril Industries. Photo: User: Dugan Meyer / Wikimedia Commons, CC BY 4.0.

BMW found the ceiling was not the prize

BMW, the German vehicle manufacturer, learned this lesson in Leipzig.

Germany planned €45.26 million of regional aid for BMW’s electric-vehicle investment there. The eligible investment was €392 million. At first glance, the obvious argument was that Leipzig fell within the programme and BMW’s spending supported a sizeable award.

BMW’s own board documents told a narrower story. Over the life of the project, Leipzig was €17 million more expensive than Munich. In July 2014, the European Commission authorised €17 million of aid and rejected the remaining €28.26 million as unnecessary.

BMW’s Leipzig case shows that a documented location disadvantage, not the maximum aid ceiling, can determine the support awarded. Photo: BMW Werk Leipzig / Wikimedia Commons, CC BY-SA 2.0 de.

Here is the twist. The €45.26 million was not the answer. It was the ceiling.

The documented €17 million location disadvantage was what established the amount considered necessary to change the location decision. The Commission was not judging whether BMW could afford electric-vehicle production. It was judging what was required to make Leipzig competitive with Munich.

Put Anduril and BMW side by side and a pattern appears that neither case says outright: the more a programme is discretionary, the less useful a generic cost schedule becomes. Evidence of the alternative becomes the real currency.

UK subsidy guidance recognises that support can change a project’s location, scale, scope or timing. US CHIPS guidance uses a similarly practical standard: awards require investment that would not occur, or would differ in scale or speed, without assistance. That matters because it rejects a common assumption. A company does not always have to prove that the entire factory disappears without aid. It must show what materially changes.

For a wider explanation of the first, easier screen, see what manufacturing projects qualify for incentives. Eligibility is necessary. It is often not sufficient.

The evidence has to exist before the story is polished

Samsung SDI, the South Korean battery-cell producer, had a more familiar contest: Hungary versus China.

Hungary proposed €108 million for Samsung SDI’s approximately €1.19 billion battery-cell expansion in Göd. The company said the production could otherwise have gone to Xi’an, China. Its internal documents showed that Hungary’s offer was considered during the final location comparison.

That was enough to establish incentive effect. It was not enough to secure every euro requested.

The European Commission rejected parts of Samsung’s cost assumptions and approved €89.6 million in February 2023. Arguments about Chinese tax treatment and local sourcing did not carry the same force where Samsung had not demonstrably relied on them when making the decision.

What surprised us was not that the Commission tested Samsung’s model. That is its job. It was how clearly the decision-time record separated evidence from reconstruction.

A board paper made while the choice is live is not glamorous. A later spreadsheet can be beautifully made and still be weaker. Samsung’s case shows why: the question is not which explanation sounds best during review. It is what management actually considered when it chose between Göd and Xi’an.

Ferrero did not need saving

The word “additionality” can make readers picture a failing business asking for rescue money. Ferrero makes that picture hard to sustain.

FedDev Ontario, the Canadian federal economic-development agency for southern Ontario, reported that Ferrero chose Canadian production expansion and new product lines in Brantford over alternatives in Italy, Germany, Ireland and Mexico in part because of federal assistance. The expansion created 118 jobs, prompted seven suppliers to consider locating in Brantford and had generated $153 million in sales by the evaluation date.

This was not a question of whether the group could make chocolate and other products somewhere. It was a question of where the investment would land.

That distinction is crucial for manufacturers. A credible but-for case can be geographical. The company may have capital and a viable project, yet the country, region or town seeking the investment faces genuine competition from another location. Ferrero’s Brantford case is the cleanest answer to the claim that only financially distressed businesses can demonstrate additionality.

The order that can end the conversation

The most painful failures often happen before anyone writes the application.

EU regional-aid rules generally require an application before “start of works”. That can mean the beginning of construction or the first firm order for equipment. A company that signs a binding equipment order and then seeks support may have already disqualified the project under that rule.

This is not bureaucratic fussiness. Once the first irreversible commitment is made, the funder has a fair question: how could its assistance have changed a decision already locked in?

UK authorities likewise must identify the no-subsidy baseline and exclude costs the beneficiary would fund anyway. So do not confuse normal planning with commitment. Manufacturers can assess sites, develop budgets and compare suppliers. But they need to know the programme’s timing rule before they cross the line it treats as irreversible.

A useful discipline is to write two versions of the project while both remain plausible: the no-support version and the supported version. Then preserve the papers showing why they differ. That is more valuable than claiming, vaguely, that support is “important”.

GI Network’s view: The but-for case should read like a decision record, not a grant-writing exercise. Show the live alternative, the commercial consequence of each option and the specific decision public support changes before the business commits.

When the opposite is true

Not every manufacturing incentive requires this kind of corporate autobiography.

Foxconn, the electronics manufacturer operating under India’s Production Linked Incentive scheme for large-scale electronics manufacturing, sits on the other side of the line. The scheme pays 4% to 6% of incremental sales over a base year when companies meet predefined investment and production thresholds. According to India’s Ministry of Electronics and Information Technology, Foxconn received approval based on verified incremental investment and sales and was reported as receiving ₹357.17 crore in FY2022-23 disbursements.

The programme does not individually decide whether each rupee changed Foxconn’s location choice. It measures additional output against a defined baseline. Meet the rules and verified performance triggers payment.

South Africa offers another useful distinction. The country’s Green Fund expressly requires a project to show that it cannot proceed without support and to demonstrate a funding gap. That is a direct financial-additionality test, rather than a payment simply triggered by a pre-set output measure.

The lesson is not that one system is tougher than another. They are buying different things. The Green Fund tests whether the funding gap is real. India’s scheme pays for verified incremental output.

So the first question is never simply, “Can equipment be funded?” It is, “What is this programme designed to purchase?”

What operators should do before they commit

For founders and operators, begin with the decision you still control.

Identify whether the programme is discretionary or formula-based. Read its rules on application timing before signing an equipment order or beginning construction. Set out the credible alternatives: another site, a smaller line, a later launch, reduced scope or no project. Then keep the board, management, financial and location-comparison materials that show those alternatives were real.

Here is a live capital-allocation use for the Changed Decision Test. Imagine management is ready to approve a larger production line at Site A, while Site B remains a real alternative. If the incentive means Site A can support the larger line, record that choice before approval: without support, Site B wins or Site A proceeds with reduced scope; with support, Site A receives the larger investment. If the board has already approved the full Site A project regardless of assistance, the test reveals the problem early. The award may still be welcome, but it is much harder to describe as the cause of the decision.

Do not build the entire capital plan around an uncertain discretionary award. A grant can alter the need for debt, equity or internal cash, but it does not erase that need. GI Network would test the project’s underlying cash flows, identify commitments that could undermine the incentive case, align the evidence pack with the funder’s decision standard and rehearse the questions an investment committee will ask about timing, location and funding gaps. The practical work is less about a polished deck than a file that can survive scrutiny, as what the process for applying for funding really requires explains.

What investors should see in the file

For investors, a but-for claim is both an opportunity and a warning light.

A strong file shows management understands its timetable, its alternatives and the amount of support actually needed. BMW’s case is the reminder that a headline maximum is not the same as an underwriteable award. Samsung’s is the reminder that assumptions must be supported by real decision records. Anduril’s shows that even very large packages come with future performance obligations.

The practical discipline is straightforward: value the project first without the discretionary incentive. Then test a second case in which support arrives later, is smaller than expected or does not arrive. The difference is the incentive downside. It should affect expected returns because it may require more equity, more borrowing or less cash available from the project.

That downside should also shape covenants, meaning promises made to lenders. If the project needs an award to meet agreed financial conditions, the financing documents should not quietly assume payment at the most optimistic date. Investors should ask what funding covers the gap before a decision, what happens if performance obligations are missed, and whether the business has cash set aside for that contingency. This is not pessimism. It is the difference between treating public support as a possible improvement and treating it as cash already in the bank.

A weak file can reveal a more serious problem: a project whose economics only work if a discretionary award arrives, even though management has already committed to the project. Investors should ask what happens if the award is smaller, later or unavailable. They should also ask whether the programme pays for output, investment, jobs, a location decision or something else entirely.

The Changed Decision Test is the simplest checklist to keep:

1. 1.Choice: What real decision remains open?
2. 2.Change: Does support alter location, scale, scope, timing or whether the project proceeds?
3. 3.Proof: Was that alternative documented when management considered it?
4. 4.Commitment: Has construction or a firm equipment order already crossed the programme’s start-of-works line?
5. 5.Programme: Is this a discretionary additionality test or a formula payment against verified thresholds?

If those five answers are clear, the incentive conversation becomes much less mysterious. If they are not, eligible spending will not save it.