When a separate vehicle is worth it
Contracts coordinate. They are poor at holding things. When an arrangement will accumulate assets — a plant, a licence, a body of research, a payroll, a brand of its own — those things have to sit somewhere, and dividing them line by line across two parents becomes unworkable quickly.
The test is straightforward. If, three years in, a stranger would have to ask whose is this? about the important assets, and the honest answer is both, then a vehicle is doing real work. If everything of value stays clearly on one side or the other, the vehicle is overhead.
Contributions and the argument they cause
Each parent contributes something: money, equipment, staff, intellectual property, access, a licence, a customer contract. The share each takes of the results is meant to reflect those contributions, and the difficulty is that the contributions are not the same kind of thing.
Cash is easy to compare. Fifteen experienced engineers seconded for two years is not, and neither is an existing relationship with a customer, or a process nobody else is allowed to use. Parties usually settle this by agreeing values during the talks, when goodwill is high. The wise addition is a stated method for revaluing contributions later, because the balance always shifts: one parent's contribution was a one-off, the other's is a continuing commitment that grows heavier every year.
Governing the vehicle
The vehicle's board is where the parents' interests meet, and it is the reason joint ventures acquire a reputation for slowness. A few structural choices carry most of the weight.
Deadlock: with two parents holding equal shares, an unresolved disagreement stops everything. Arrangements deal with this by giving one parent a casting decision in defined areas, by escalating to named senior people on each side, or by defining an ending that both would rather avoid, so both prefer to settle.
Reserved matters: a short list of decisions that need both parents' agreement — new borrowing, a change of business, a large disposal — while everything else is left to the vehicle's own management. Short lists work. Long lists mean the vehicle has no management, only a queue.
Staffing: people seconded from a parent keep their careers, their loyalties and their pension in the parent. People employed by the vehicle acquire an interest in the vehicle surviving. This is not a small detail; it changes how the venture behaves under stress.
Consortia: acting together without a vehicle
A consortium is a lighter answer to a specific problem: a piece of work is too large, too broad or too specialised for one party, and the customer wants to deal with one arrangement rather than five.
Usually no company is formed. Instead the members agree among themselves how the work divides, and one of them takes the lead position facing the customer. That lead member carries a real exposure, because from the customer's point of view it is answerable for the whole thing, including the parts it does not perform.
Consortium agreements therefore concentrate on three things: which member does which scope, how payment received by the lead is passed on, and what happens when one member fails to deliver. The third is the one that gets skimmed during optimistic negotiations and matters more than the other two combined.
What each is good and bad at
A joint venture is good at holding shared assets, sustaining long work, and giving joint activity its own identity and management attention. It is bad at speed, bad at reversibility, and prone to becoming an orphan when both parents change strategy.
A consortium is good at assembling capability quickly for defined work, and it leaves each member independent afterwards. It is bad at anything requiring shared investment, and it exposes every member to the weakest member's performance.
Planning the ending at the start
The most useful hour spent in a joint-venture negotiation is the one spent on how it stops. Who may buy out whom, on what notice, at a price set how; what happens to seconded staff; who keeps jointly developed work and on what basis the other may keep using it. Parties resist this discussion because it feels like planning for failure. It is the opposite: arrangements with a clear, survivable ending are the ones parties are willing to enter properly in the first place.
Where this leads
Whatever the structure, the arrangement is run day to day by people who did not negotiate it. Governance is where that handover succeeds or fails.