Tax

Cost Sharing Agreement Rejected by Tax Office: Recognition Conditions

Close-up of a workspace with a calculator, coins, and glasses on a notepad.

A cost sharing arrangement (CSA) is the mechanism multinational groups use to split joint research and development costs by expected benefit shares, and it is among the structures Vietnamese authorities reject most readily, because Vietnamese law has no CSA regime of its own: every allocation charged into Vietnam is “translated” back into transactions domestic law knows — research service fees, technology royalties, or consideration-less charges — and examined under Decree 255/2026/ND-CP (earlier periods: Decree 132/2020/ND-CP) and the deductibility conditions of Corporate Income Tax Law 67/2025/QH15. Unlike loss allocation (EXP-6008), which transfers a negative outcome with no consideration, a CSA has real consideration in rights over research outcomes; unlike group service fees (EXP-5992), the CSA question is not only “was the service real” but “is the subsidiary genuinely a participant bearing risk and holding rights in the research programme”.

Why CSAs get rejected: five practical recognition conditions

One, a written agreement signed before or when the costs arise, identifying the participants, the programme scope, each party’s rights over outcomes and its contribution duties. Two, an allocation key grounded in verifiable expected benefits (projected revenue from products using the research outcomes, markets licensed to exploit them), not a round percentage imposed one-way by the group. Three, evidence the research programme is real and running: project files, research personnel, stage outcomes actually transferred to the subsidiary. Four, the subsidiary’s substantive rights over outcomes within its scope (usable without paying additional royalties for the same outcomes). Five, a true-up mechanism when actual benefit departs from projections, plus exit and settlement terms.

Inspection practice shows the two most commonly missing conditions are substantive rights over outcomes (a subsidiary paying both CSA contributions and royalties for the very technology the programme produced — a self-cancelling contradiction) and an allocation key traceable to no expected benefit at all. Missing either point, the whole allocation is treated as a cost not serving the enterprise’s own activity.

Allocated costs without participating in the research: why it fails

A CSA’s essence is sharing among parties who jointly participate in and jointly bear the risk of a research programme: participants contribute by expected benefit and receive rights over outcomes even if the research fails. A subsidiary with no research function, no related personnel, no outcomes transferred to it and products that do not use those outcomes is not, functionally, a CSA participant — money imposed on it is a one-way group cost allocation, rejected on both grounds: no benefit consideration, and not a cost serving its own activity.

The subtler case is a subsidiary whose products do embed the research outcomes: its true relationship is then that of a buyer of goods or a licensee of technology from the outcome owner, and the cost must travel through goods pricing or a valued royalty channel, not through a CSA. Using the wrong channel not only loses the deduction but also misses the contractor-tax duty on the right channel.

After rejection, where the cost goes — and the late-entry buy-in

When a CSA is rejected, the booked allocation has no route to deductibility as a CSA cost; the company can salvage only the part provable as the cost of a specific research service actually supplied to it (deliverables, acceptance records), re-examined as a service fee under the full conditions of a related-party service transaction — with the corresponding contractor-tax duty where the provider is offshore. The unprovable remainder leaves deductible costs, with tax collection and late-payment charges under the tax handling frame; there is no “automatic conversion” from CSA to service fee.

On buy-in: a party joining a CSA after the programme has already created value (technology formed, patents filed) must pay an entry amount for the pre-existing intangibles it gains rights to exploit — the standard CSA principle worldwide. That buy-in is a separate intangible transaction requiring independent valuation; joining late while skipping the buy-in, or folding it into the periodic allocation ratio to dodge valuation, are both inspection targets. The “retroactivity” scrutiny is this: an agreement signed later cannot create effects for periods in which the subsidiary was not a participant and bore no risk.

FAQ

What are the conditions for a CSA to be recognized?

In practice one or more of five elements is missing: a written agreement signed before the costs arose with clear scope and party rights; an allocation key traceable to expected benefits; files proving the research programme is real with outcomes transferred; the subsidiary’s substantive rights over outcomes (no additional royalty for the same outcomes); and a true-up mechanism with exit terms. The two most frequently missing are substantive rights over outcomes and a grounded allocation key.

What is a buy-in payment?

Not valid. A CSA participant must jointly take part in and bear the risk of the research programme, with substantive benefit from its outcomes. A subsidiary with no research function, receiving no outcomes and whose products do not use them, is subject only to a one-way cost allocation, rejected for lack of benefit consideration and for not being a cost of its own activity under Corporate Income Tax Law 67/2025/QH15.

Which missing clause leads to rejection?

There is no automatic conversion. Any part proven to be the cost of a specific research service actually performed and delivered to the subsidiary may be re-examined as a related-party service fee, with full evidence conditions and the corresponding contractor-tax duty; the rest is removed from deductible costs with tax collection and late-payment charges. The company should restructure its evidence toward the service character proactively during explanation, not wait for the authority to suggest it.

How should the contract be rewritten?

An agreement signed later creates no retroactive effect for periods when the subsidiary was not a participant and bore no research risk — allocations computed for periods before signing will be rejected. Late entry into a programme that has already created value also triggers a separate transaction: a buy-in payment for rights over the formed intangibles, independently valued under Decree 255/2026/ND-CP (earlier periods: Decree 132/2020/ND-CP) and kept apart from the periodic cost allocation ratio.

Related articles

Reviewed by the Legal Practice Group – FLAT LAW FIRM

This legal analysis has been verified against current Vietnamese regulations and cross-border FDI transaction practices by the attorneys at FLAT LAW FIRM.