certifications7 min read

Subcontracted R&D: Why Your Client Cannot Claim Your Invoice Without Approval

The scene repeats every spring, at filing time. A design office delivered, the previous year, a development worth tens of thousands of euros for a manufacturer: a new process, genuine technical uncertainty, successive trials, iterations. Nobody disputes that this was research. And yet the client discovers, while preparing the return, that none of it can go into the crédit d’impôt recherche (CIR), France’s research tax credit. The reason is one word: the provider is not approved.

The mechanism surprises people because it is purely formal. It deserves to be understood precisely — by providers, who lose business without always knowing why, and by clients, who find out too late to fix it.

What article 244 quater B actually says

The research tax credit is set out in article 244 quater B of the French Tax Code (code général des impôts). It lets an industrial, commercial or agricultural company taxed on its actual profits deduct from its tax a share of its research and development spending: 30 % of eligible expenditure up to one hundred million euros, 5 % above that, with a higher rate in the French overseas departments. The 2026 Finance Act left this architecture unchanged.

That expenditure may be incurred in-house or entrusted to third parties. And this is where the text adds a condition: the inclusion of subcontracted research expenditure is conditional on the provider having received approval from the minister responsible for research. The rules governing that approval are set out in article 49 septies H of Annex III to the same code.

The wording is dry, and its consequence just as dry. Without approval, the amounts invoiced are excluded from the calculation, whatever the nature of the work performed. There is no tolerance, no retroactive correction, no discretionary assessment. The condition is objective: it is met or it is not.

A point of vocabulary that often causes misunderstanding: approval earns the provider nothing directly. It opens no additional tax credit for the holder. It is designed for the benefit of the holder’s clients — which explains why so many companies put off applying, until the day a prospect makes it a selection criterion.

The two ceilings clients need to know

Even with an approved provider, outsourcing is not unlimited. Expenditure entrusted to third parties counts towards the base only within an overall annual limit of ten million euros. That ceiling drops to two million euros where a relationship of dependence exists between the company and the organisation it entrusts the work to — the classic configuration of groups that house their R&D in a dedicated subsidiary. A specific, higher ceiling applies to work entrusted to public and equivalent research organisations.

A further, less well-known requirement is worth flagging: the operations must be carried out directly by the approved organisation they are entrusted to. An approved provider that merely re-invoiced work done by an unapproved third party would expose its client to a reassessment. Cascading subcontracting chains were framed precisely to prevent that workaround.

What approval proves — and what it does not

Beware the opposite excess: the provider who sells approval as a tax guarantee. Approval is a necessary condition, not a sufficient one.

The question What approval settles What still has to be shown
Does the provider have the means to do research? Yes: human and material resources reviewed by a ministry expert Nothing more, for the period covered
Does the invoiced work qualify as R&D? No Novelty, scientific or technical uncertainty, experimental approach, project by project
Can the client include the invoice in its base? It is the entry condition The real nature of the work, checkable after the fact
Was the work carried out by the approved body itself? No Traceability of execution, to be documented

In an audit, the tax authorities — assisted where necessary by an expert from the research ministry — examine the real nature of the operations invoiced. If they show neither novelty nor scientific or technical uncertainty within the meaning of the OECD Frascati Manual criteria, the expenditure is removed from the client’s base, approval or not. An honest provider explains that limit to its clients; it is also what distinguishes it from a vendor of tax advantages.

A closely studied scheme, hence a closely controlled one

The French research tax credit is one of the most studied tax expenditures in the country, which explains the rigour of the controls. Benoît Mulkay and Jacques Mairesse published an econometric assessment of the French scheme and of the 2008 reform in Oxford Economic Papers in 2013, estimated on a large panel of R&D-performing firms using a dynamic model of research demand (see the study). Antoine Bozio, Delphine Irac and Loriane Py showed, in a Banque de France working paper published in 2014, a positive effect of that reform on R&D spending both among firms already engaged in research and among new entrants, with a more modest effect on innovation itself than might have been expected (see the study).

The practical lesson for a provider is not econometric but behavioural. A scheme of this scale, regularly evaluated and regularly debated in Parliament, will not see its formal conditions loosened. Better to build your offer on approval in good standing than on the hope of leniency.

The commercial fallout, concretely

For a provider, the absence of approval produces three knock-on effects.

You drop out of tenders before being assessed. More and more clients ask for the approval number in the bid file, alongside the tax-and-social clearance certificate and the professional indemnity policy. An empty field, and the bid does not move up.

You negotiate with a price handicap. For the same work, a client able to recover 30 % of the invoice through the tax credit compares your 100,000 euros with a competitor’s net 70,000. The gap is structural, and no reasonable discount closes it.

You find out after signature. That is the worst case: the contract is performed, the invoice issued, and the client realises the following spring that nothing can be claimed. The relationship sours over an issue you could have settled eighteen months earlier.

For a client, the reflex to install is simple: check the approval before contracting, and check not only that it exists but its period of validity. The ministry publishes and updates the list of approved organisations, experts, style bureaux and stylists as open data. Approval expiring on 31 December does not cover work carried out the following January.

Should you apply?

The question is not legal — nobody is obliged to — but commercial. Three situations stand out.

  • Your clients are French companies that claim the research tax credit. Approval is not optional in practice: it gates your access to the market. See our guide on putting the approval file together.
  • You design prototypes or pilot installations of new products for SMEs. That is the scope of the innovation tax credit, with its own procedure: see our comparison of CIR and CII approval.
  • Your clients are foreign, ineligible, or do not claim any tax credit. Approval will bring nothing immediately. Keep it in reserve: the day a large French account joins your portfolio, the first-application campaign runs only from 1 January to 31 March.

That last point is decisive and too rarely anticipated. Filing windows are narrow and assessment takes months: our article on the approval timetable sets out the dates and the consequences of filing late. Finally, if your organisation also sells training, do not mix the two worlds: our comparison of CIR approval and Qualiopi certification clears up a costly confusion.

Take action

If you are a provider, ask yourself one question: how many of your prospects have asked for an approval number in the past twelve months? If the answer is not zero, your file must be ready for the 1 January window. If you are a client, add approval — and its expiry date — to your supplier qualification checklist. Scheme status, detailed procedures, frequently asked questions and a free ebook: it is all on our page dedicated to CIR-CII approval.

FAQ

Frequently asked questions

+Can my client still claim my invoice if the work is plainly research?

No. The approval requirement in article 244 quater B of the French Tax Code is a formal, objective condition: it is either met or it is not. The scientific quality of the work does not make up for its absence, and the tax authorities make no case-by-case assessment on this point. The expenditure is removed from the client's base.

+Does approval earn me, the provider, a tax credit?

No. Approval opens no tax credit right for the holder. It serves the holder's clients, who can then include subcontracted research invoices in the base of their own research tax credit. An approved provider does, separately, keep its own tax credit for work carried out on its own account.

+Is there a ceiling on subcontracted research expenditure?

Yes, on the client's side. Expenditure entrusted to third parties counts towards the research tax credit base only within an overall annual limit of ten million euros, reduced to two million euros where the company and its provider are related parties. A specific, higher ceiling applies to work entrusted to public and equivalent research organisations.

Read next