Hidden opportunities within reach · Essai 2

An attractive deal can have no stable bargain

Finding a valuable combination is only part of finding an opportunity. There may be no division of its proceeds that keeps every credible breakaway coalition willing to stay.

The first essai explored nearby arrangements and possible moves between them. This one adds a missing test: whether an arrangement can support a stable voluntary bargain under explicitly supplied outside opportunities. Its most useful result is a refusal. Sometimes better negotiation cannot repair the arithmetic.

Three partners, ten units of value

Mara, a technical partner and a museum network can together create ten transferable surplus units. Each pair can separately create eight; alone they create zero. The proposed division is six, two, two. Every number and person in this construction is fictional. No private configuration or actual counterpart information is included.

The technical partner and museum network receive four together but could create eight without Mara. Raising their shares invites a different pair to leave. Indeed, no division of ten prevents every pair from improving: adding the three pair requirements demands twice the total payout be at least twenty-four.

See every stable division

The chart uses Mara's share horizontally and the technical partner's share vertically; the museum receives what remains. The green region contains every division that meets all individual and pair alternatives. A single point means the stable division is unique. Empty space means no stable bargain exists under these numbers.

“Stable” here means no coalition can give all of its members a strict improvement using its stated value. Equality is allowed: a coalition may be indifferent. This is neither a prediction of agreement nor a fairness judgement.

The missing value has a certificate

Let B be the total available only to the grand coalition. Each person's share must be at least their standalone value and at most B minus the opposite pair's value. Feasibility is the intersection of those three intervals with the plane whose shares sum to B.

Consequently the minimum B is the largest of five lower bounds: the sum of standalone values; each pair's value plus the remaining person's standalone value; and half the sum of all three pair values. These conditions are sufficient as well as necessary: the individual intervals are nonempty, their lower endpoints sum to at most B, and their upper endpoints sum to at least B.

The extra two units in the opening example restore the unique division four, four, four. They could represent real additional joint value or an outside contribution tied exclusively to the three-party arrangement. If departing pairs can also claim that contribution, their outside values must change and this calculation must be repeated.

Which outside offer is worth investigating?

Mara can make 8 with either partner. The three together can make 10 and nobody earns anything alone. The uncertain quantity is what the museum and technical partner could make without Mara. Test that estimate while holding the other two alternatives fixed.

Each of Mara's two alternatives leaves at most 2 for the excluded partner. The two partners can receive at most 4 together. At or below 4, the division 6, 2, 2 satisfies all three pairs. Above 4, no division of the 10 does. A conversation about the third offer can settle whether negotiation at the current total is worthwhile.

What the assignment-game source does—and does not—give us

Shapley and Shubik's assignment game connects efficient two-sided matching with a nonempty core of stable payoff divisions. That existence result depends on its structure. Three complementary parties need not inherit it. The opening counterexample is precisely where applying the attractive theorem outside its domain would mislead the opportunity search.

Our three-person transferable-utility game and its exact geometric solver are separate constructions. An independent half-unit allocation enumeration checks 3,375 cases, including the minimum-budget threshold and closest stable division to the opening offer. The proof above covers continuous shares; enumeration is corroboration, not a substitute for it.

Strongest objection. Real benefits are often non-transferable, outside options uncertain, and permissions cannot be bought with surplus. The permission switch is therefore a separate condition. A positive core is only one compatibility test; an empty core may prompt changing the proposed coalition or finding genuinely new joint value. Nothing here discovers a market, establishes anyone's willingness or turns a private ambition into a commercial claim.

Earlier essai: Nearby arrangements.

Download the model and independent reproduction check.