Group R&D charges allocated down to subsidiaries by a revenue key are the most contested of all shared costs, because they blend two different legal natures into one cash flow: if the charge is a research service fee, it must pass the benefit test like any intra-group service; if it is a contribution under a cost sharing arrangement (CSA), the duty to pay exists only where the subsidiary expects to benefit from the R&D results and actually participates as a beneficiary party. A Vietnamese subsidiary manufacturing on settled technology, using none of the group’s new R&D output and selling no product containing it, is entitled to refuse payment — but the right must be exercised through documentation, not by silently not paying while still booking the cost. The piece on parent management fees (FDI-LC-4688) covers management-service benefits, and the one on CSA recognition conditions (EXP-6010) covers the arrangement frame; this article covers the subsidiary’s act of refusal.
When refusal is the correct legal position
Refusal stands where three conditions coincide: the subsidiary does not participate in, or has withdrawn from, the cost sharing arrangement for the R&D program being charged; the R&D results (patents, processes, new formulas) are not transferred, licensed or applied in the subsidiary’s production or products; and the subsidiary receives no identifiable indirect benefit from that program. Where any condition flips — for example the Vietnamese plant runs a production process created by that very R&D program — the charge for the benefited portion becomes payable, and the argument moves to the reasonableness of the allocation.
Timing is decisive: refusal must be expressed when the charge is raised — a written objection to the parent, minutes recording non-participation, and no cost booked. A company that booked group R&D charges as deductible costs for years and only says “we never used it” under inspection puts itself in an inexplicable contradiction.
Proving “no benefit” with systematic negative evidence
Proving a non-existence is harder than proving an existence, and must be done with a systematic negative evidence set: the inventory of technologies and patents actually applied at the Vietnamese plant with the origin of each (self-developed, bought from third parties, or separately licensed for a royalty — if royalties are already paid for the same R&D output, the group R&D charge is duplicative); the current product list cross-checked against products containing the group R&D program’s results; and a technical department confirmation that production processes do not use the program’s output.
The strongest evidence is the group’s own paper: CSA documents whose participant list excludes the Vietnamese subsidiary, or group R&D reports describing the results’ application scope as not covering Vietnam. Request these from the group formally in writing rather than asserting on your own.
If it was paid and is adjusted: the two faces of the risk
Where the company paid and booked the charge: the authority rejects the cost under the benefit test and assesses Vietnamese tax — the company loses at both ends if it cannot recover the fee from the parent. The stranger case: the company refuses to pay but the group still invoices and carries a receivable; with no cost booked there is no cost to reject, yet the company must ensure non-payment is not read as a financing transaction (a multi-year hanging payable recharacterised as an interest-free related-party loan). The correct handling is a written termination or non-participation confirmation with the group, and cleaning the intercompany balance in accordance with it.
For future years, if the subsidiary truly begins using group R&D results, the clean path is a licence agreement or proper CSA participation with a buy-in for accumulated value — the CSA condition set is analysed separately at EXP-6010.
This article is a FLAT Law Firm draft pending review by the QA Council before publication; it provides general legal information only and is not advice for any specific matter.
FAQ
When can R&D fees be refused?
Refusal is justified where the subsidiary does not participate in the cost sharing arrangement for that R&D program, receives no transfer or application of the R&D results in its production and products, and obtains no identifiable benefit from the program. The refusal must be documented in writing to the parent and kept consistent in the books (no cost booked), under the benefit test frame of Decree 255/2026/ND-CP (earlier periods: Decree 132/2020/ND-CP).
How to prove no benefit was received?
With a negative evidence set: the inventory of technologies and patents applied in Vietnam with each technology’s origin; the current product list cross-checked against products containing the group R&D program’s output; the technical department’s confirmation that production processes do not use the program’s output; and the group’s own CSA documents showing the subsidiary is not a listed participant or within the application scope.
What does a cost sharing agreement need?
A sound cost sharing agreement needs: identified participating parties and each party’s expected benefit from the R&D results; a cost pool and allocation keys corresponding to expected benefit; a buy-in mechanism for joining mid-stream and adjustments when scope changes; and reporting and verification duties over actual results. The full CSA conditions are analysed in the dedicated piece on cost sharing agreements (EXP-6010).
What happens if adjusted?
If paid and booked, the authority may reject the entire portion whose benefit is unproven, assessing tax and penalties under the transfer pricing frame. If refused without a written non-participation confirmation, the group’s hanging receivable risks being read as a related-party financing transaction. Both cases must be closed out with written documentation with the group, not left in limbo.
This article has been quality-reviewed by the FLAW AI Legal Council
