Theorem Minimal Decisive Coalition Has Size One
theorem staged

Minimal Decisive Coalition Has Size One

Theorem. Let $F$ be a social welfare function on a finite nonempty voter type with $|A| \ge 3$, satisfying unanimity ([[social_choice.voting.unanimity]]) and IIA ([[social_choice.voting.iia]]). Then $F$ admits a decisive coalition of size exactly one.

Formally: the cardinality predicate $n \mapsto \exists C\ \mathrm{decisive\ of\ size}\ n$ is well-founded on $\mathbb{N}$, and its minimum equals $1$.

In Lean: SocialChoice.Voting.decisive_minimal (proves Minimal (exists_nonempty_decisive_of_size F) 1), with the existence half SocialChoice.Voting.exists_minimal_decisive_coalition. Both are public lemmas of the Arrow development in Decisive.lean, over the predicate SocialChoice.Voting.exists_nonempty_decisive_of_size.

Proof

sketch

  1. The grand coalition is decisive of size $|N|$ ([[social_choice.voting.unanimity_univ_decisive]]), so the cardinality predicate is nonempty. Take any minimal $n$.

  2. $n \ne 0$: a decisive coalition is nonempty by definition.

  3. $n < 2$: if $n \ge 2$, the contraction lemma ([[social_choice.voting.decisive_contraction]]) produces a strictly smaller decisive coalition, contradicting minimality.

Combining $n \ne 0$ and $n < 2$ forces $n = 1$. $\square$

References

  • [MSZ Chapter 21] Maschler, Solan, and Zamir, Game Theory. Minimality step in the decisive-coalitions proof.

Used by

Also in