AFTD Auto-Formalizing Theoretical Domains

walrasian_budget_exhaustion

theoremverifiedformalization

In a pure exchange economy E with finite agents Agent and finite, nonempty commodity type Good, if price vector p has strictly positive prices, agent i has a strictly monotone utility function, and (p, x) is a Walrasian equilibrium, then agent i exhausts their budget at equilibrium: the expenditure on allocation x_i equals the value of agent i's initial endowment (that is, ∑ g, p g * x i g = ∑ g, p g * E.endowment i g).

Statement

theorem walrasian_budget_exhaustion
    {Agent Good : Type*} [Fintype Agent] [Fintype Good] [Nonempty Good]
    (E : ExchangeEconomy Agent Good) (p : Good → ℝ) (x : Agent → Good → ℝ)
    (i : Agent)
    (hp : ∀ g, 0 < p g)
    (h_mono : StrictMono (E.utility i))
    (h_walras : is_walrasian_equilibrium E p x) :
    ∑ g, p g * x i g = ∑ g, p g * E.endowment i g
Formalizes
Mas-Colell, Whinston & Green, Proposition 3.D.2(i)
Verified
01 Oct 2026
Axioms
Classical.choiceQuot.soundpropext
Built on 2
  • ExchangeEconomyA pure exchange economy with agent type Agent and commodity type Good consists of a non-negative initial endowment of each good for each…
  • is_walrasian_equilibriumIn a pure exchange economy E with finite agents Agent and finite commodities Good, a price vector p : Good → ℝ and an allocation x : Agent…

Read back from the Lean

Let Agent and Good be finite types, with Good nonempty. Let E be an exchange economy with agents Agent and goods Good, let p : Good → ℝ be a price vector, let x : Agent → Good → ℝ be an allocation, and let i : Agent be an agent. Assume that: 1. Every good has strictly positive price: ∀ g, 0 < p g. 2. Agent i's utility function E.utility i is strictly monotone (StrictMono (E.utility i)) with respect to the pointwise partial order on Good → ℝ. 3. (p, x) is a Walrasian equilibrium for E. Then agent i exhausts their budget: ∑ g, p g * x i g = ∑ g, p g * E.endowment i g.

Written by a model that saw only the Lean, never the English above. If the two disagree, that is worth a challenge.

Lean source view module on GitHub

/-- At a Walrasian equilibrium with strictly positive prices and strictly monotone preferences, every agent's budget constraint binds with equality. -/theorem walrasian_budget_exhaustion    {Agent Good : Type*} [Fintype Agent] [Fintype Good] [Nonempty Good]    (E : ExchangeEconomy Agent Good) (p : Good → ℝ) (x : Agent → Good → ℝ)    (i : Agent)    (hp : ∀ g, 0 < p g)    (h_mono : StrictMono (E.utility i))    (h_walras : is_walrasian_equilibrium E p x) :    ∑ g, p g * x i g = ∑ g, p g * E.endowment i g := by  have h_le := (h_walras.2 i).1  have h_max := (h_walras.2 i).2  have h_feas_nonneg := h_walras.1.1 i  apply le_antisymm h_le  by_contra! hlt  set n : ℝ := (Fintype.card Good : ℝ)  have hn : 0 < n := Nat.cast_pos.mpr Fintype.card_pos  set Δ := ∑ g, p g * E.endowment i g - ∑ g, p g * x i g  have hΔ : 0 < Δ := sub_pos.mpr hlt  set c := Δ / n  have hc : 0 < c := div_pos hΔ hn  set y : Good → ℝ := fun g => x i g + c / p g  have hpy : ∀ g, p g * y g = p g * x i g + c := by    intro g    dsimp [y]    rw [mul_add, mul_div_cancel₀ _ (ne_of_gt (hp g))]  have hsum : ∑ g, p g * y g = ∑ g, p g * E.endowment i g := by    simp_rw [hpy, Finset.sum_add_distrib, Finset.sum_const, nsmul_eq_mul]    have : (Finset.univ.card : ℝ) = n := rfl    rw [this, mul_div_cancel₀ _ (ne_of_gt hn)]    exact add_sub_cancel _ _  have hy_gt : ∀ g, x i g < y g := by    intro g    dsimp [y]    have : 0 < c / p g := div_pos hc (hp g)    linarith  have hy_nonneg : ∀ g, 0 ≤ y g := by    intro g    exact le_trans (h_feas_nonneg g) (le_of_lt (hy_gt g))  have hxy_lt : x i < y := by    rw [lt_iff_le_and_ne]    refine ⟨fun g => le_of_lt (hy_gt g), ?_⟩    intro h_eq    obtain ⟨g0⟩ := ‹Nonempty Good›    have h1 : x i g0 = y g0 := congr_fun h_eq g0    have h2 : x i g0 < y g0 := hy_gt g0    linarith  have h_util_lt : E.utility i (x i) < E.utility i y := h_mono hxy_lt  have h_util_le : E.utility i y ≤ E.utility i (x i) := h_max y hy_nonneg (le_of_eq hsum)  exact not_le_of_gt h_util_lt h_util_le
See it in the graphDiscuss this resultChallenge it