Contract R&D vs Entrepreneurial R&D: DEMPE and Cost-Based Returns

“R&D services” is the label that hides the most expensive characterization question in transfer pricing. The same invoice — an Indian entity charging its overseas affiliate for “research and development” — prices very differently depending on the answer to one question: is the Indian entity performing routine research services on another party’s R&D program, or is it a genuine R&D participant sharing in the development’s risk and return? The first earns a cost-based return. The second’s return is a question of DEMPE allocation, and the gap between the two is where the adjustment comes from.

The two fact patterns

	Contract (routine) R&D	Entrepreneurial R&D participation
The program	The commissioning party’s program, its specifications, its priorities	A joint development — the participant sets or shapes the direction
Decision rights	The commissioning party approves the scope, the milestones, the deliverables	Shared — the participant’s engineers make the development decisions
Risk of failure	The commissioning party’s — the participant is paid for the effort regardless of outcome	Shared — the participant’s return depends on the development’s success
IP ownership	The commissioning party owns the output (the participant performs, the commissioning party owns)	Allocated by agreement — the participant owns or co-owns a share
Return	Cost plus a benchmarked mark-up (a service fee)	A share of the development’s value — the residual-split territory
Method	Cost plus / TNMM on OP/OC, on the service function	Profit split (residual), with the DEMPE record as the allocation basis

The characterization is decided by conduct and rights, not the invoice title: the contract’s decision-rights clauses, the actual decision trail (who proposed, who approved, who changed scope), the IP registry (who is the recorded owner of the patents and code), the funding pattern (who paid when the milestone slipped), and the loss-bearing (who absorbed the failed program’s cost).

DEMPE and control — the allocation machinery

Where the participation is entrepreneurial, the value of the development is allocated on the DEMPE functions — Development, Enhancement, Maintenance, Protection, Exploitation — with control as the tie-breaker: who controlled the risk of each function (who had the right and the ability to make the decision, and who bore the consequence). See the routine vs entrepreneurial guide for the framework.

The allocation questions, per function:

• Development — who performed it is rarely the dispute; the dispute is whether the development was routine execution (the participant’s engineers building to the commissioning party’s architecture) or key development (the participant’s engineers making the architectural decisions the product lives or dies on). The decision trail is the evidence.
• Enhancement — the ongoing improvement stream: same question, continuous form. A participant that genuinely enhances (not just maintains) the platform is doing more than a contract service.
• Protection and Exploitation — usually the commissioning party’s (the IP owner, the commercial side). Where the participant protects or exploits (files the patents in its own name, licenses the output elsewhere), the characterization has already moved.

The cost-based return, for the routine side

Where the characterization is contract R&D — the routine case, and the case Indian authorities examine most often — the return is a service return:

• Method: cost plus on the R&D cost base (the contract manufacturing discipline applies to the cost base: line by line, stated, consistent), or TNMM on OP/OC against comparable R&D service providers.
• The Indian safe harbour: for contract R&D (software) and contract R&D (generic pharma), the safe harbour regime prescribes OP/OC ≥ 24% (within the transaction limit) — see the safe harbour guide. The 24% harbour is the Indian fact pattern where the routine R&D return is a prescribed number, and it is materially above the generic ITeS/software harbour (17–18%) because the function is knowledge-dense.
• The mark-up without the harbour: the comparable pool of independent R&D service providers (the contract research organizations, the captive- equivalent independents), on the same cost definition, with the IQR as the range.

The TPO’s standard move on a “routine” characterization it accepts: the mark-up at or near the pool’s position — and on a “routine” characterization it rejects (the Indian entity is re-characterized as a key developer), the method becomes the residual split question, which the TPO usually does not voluntarily open — the re-characterization typically works against the taxpayer’s position in the TPO’s framing (the Indian entity’s “routine” cost-plus is rejected as under-remunerating the key development, or, more often in practice, the TPO keeps TNMM on the Indian entity and the entire residual stays abroad — the taxpayer’s appeal is that the characterization, and therefore the method, is wrong).

The IP ownership record

The IP registry is the exhibit that decides the characterization in appeals:

• Patents and registrations — the named applicant/owner, the filing history (filed by which entity, on whose instruction, on whose funding).
• Code and technology ownership — the development agreements’ IP clauses, the repository ownership, the license-back terms (a license-back from the “owner” to the performer on broad terms is evidence the performer is not the owner).
• The economic ownership vs legal ownership gap — where legal title sits with one party and the economic benefits (the product revenue, the exploitation) sit with the other, the file must reconcile the two — the OECD’s position is that the party that performs and controls the DEMPE functions with the risk bears the economic attribution, even where the contract says otherwise. The contract that says “we own everything” while the other party funded, directed and exploited is a contract the examination reads through.

The documentation that decides it

1. The R&D agreement — scope, decision rights, IP ownership, the milestone and payment structure, the failure consequences.
2. The decision trail — the project records: who proposed, who approved, who changed course, who funded the rework.
3. The IP registry extract — the owners of the output, per the characterization.
4. The functional record of the Indian entity — the team, its skills, its decision authority (the org chart is not enough; the decision rights are).
5. The characterization memo — the conclusion (routine / entrepreneurial, with the evidence), written before the method is chosen, because the method follows the characterization and not the other way around.

See also

• Routine vs entrepreneurial
• Profit Split method
• Software & SaaS TP
• Intangibles TP