Routine vs Entrepreneurial: DEMPE and Who Captures the Profit

Almost every transfer pricing dispute in the modern era boils down to one question: is this entity routine, or is it entrepreneurial? A routine entity performs functions for a predictable, benchmarkable return (cost plus, TNMM). An entrepreneurial entity owns unique intangibles or bears un-benchmarkable risk and captures the residual profit — and the method, the PLI and the comparable pool all follow from the answer.

The routine / entrepreneurial spectrum

There is no bright line, but the working spectrum looks like this:

Position	Functions	Assets	Risks	Profit character
Pure routine	Defined, specified activities (manufacturing to spec, data processing, routine R&D services)	Employed, not owned	Contractual, operational only	Cost-plus return, fully benchmarkable
Limited risk	Market-facing functions under contractual shields (limited-risk distributor)	Working capital; IP licensed from group	Inventory/credit risk absorbed by principal	Return on narrow function (TNMM)
Hybrid	Mix of routine delivery and some commercial discretion	Some owned IP or customer relationships	Partial risk	Split treatment needed
Entrepreneurial	Innovation, product development, brand creation, go-to-market strategy	Owns/controls unique intangibles	R&D failure, market, product risk	Residual profit — not benchmarkable by one-sided methods

The two poles are unambiguous; the middle is where the work is. A captive that “provides R&D services” can sit anywhere from pure routine (fixed-fee research projects) to hybrid (development of owned platform technology) — and the difference is worth the entire residual profit pool.

DEMPE: the modern test for intangibles

Post-BEPS (OECD Ch. 5), “who gets the intangible profit” is answered by DEMPE — which entity performs Development, Enhancement, Maintenance, Protection and Exploitation of the intangible, and — critically — which entity controls the risk of each function:

1. Identify the intangibles and who legally owns them (legal title is the starting point, not the answer).
2. Map DEMPE functions to entities — who actually develops the code, enhances the platform, protects the patents, exploits the brand.
3. Assign control of risk — who bears the economic consequence when a development fails, a product underperforms, a patent is challenged?
4. Allocate profit — returns follow DEMPE contribution and risk control, not legal title. A shell entity that holds a patent but performs no DEMPE and controls no risk is not the profit owner.

The classic Indian pattern: an overseas parent holds the IP, an Indian captive develops and enhances it, and the contract says “IP remains with the parent.” Post-2017, the TPO and the ITAT read that pattern through DEMPE: if the captive performs the developmental functions and the parent’s contribution is funding, the question is whether the captive’s compensation is a routine cost-plus on development services — or whether it should capture part of the residual. That is exactly the question a defensible functional analysis must answer before any pricing.

Control of risk — the word that decides cases

“Ownership without risk” is the recurring fact pattern. The OECD’s rule: the entity that controls the risk (can decide whether to accept or reject it, and can mitigate it) is the one whose compensation reflects it. Three probes:

• Would the entity’s P&L move if the risk materialised? A captive whose fee is fixed at cost-plus does not economically bear R&D-failure risk even if the contract names it the “owner” of development work.
• Does the entity make the go/no-go decisions? Investment, launch, and kill decisions are risk-control signals.
• Does the entity have the resources to mitigate? A party that cannot mitigate the risk (no capital, no optionality) is not a genuine risk-bearer.

If the “risk-bearing” party fails all three probes, the risk allocation in the contract is not respected for TP purposes — and the profit reallocation follows.

How the divide drives the method choice

Position	Method logic
Pure routine	One-sided methods: cost plus or TNMM on the routine function. Fully benchmarkable.
Limited risk	TNMM (or RPM) on the narrow function; Amount B where conditions hold
Hybrid	Two analyses: routine part benchmarked, entrepreneurial part allocated
Entrepreneurial	Profit split — one-sided methods fail because the residual cannot be measured against comparables

This is why method selection always starts with the FAR profile, not with data availability: the data says what you can compute; the FAR says what you may compute.

The two failure modes

• Calling the entrepreneurial entity routine. The under-compensation pattern: an entity that actually develops the platform is priced at cost-plus, and the residual is left overseas. The TPO’s most aggressive (and most frequent) adjustment theory — the “benefit test” and DEMPE recharacterisation both point here.
• Calling the routine entity entrepreneurial. The over-compensation pattern: a limited-risk distributor or captive is priced above its function because the group “needs” it to earn there. The defence must show the function, the risk and the asset behind the extra return — or the TPO re-characterises and adjusts down.

In both directions the cure is the same: a contemporaneous FAR profile with DEMPE mapped, risk control probed, and the method chosen from that evidence.