[Paper Review] Nash Equilbria for Quadratic Voting

Steven P. Lalley, E. Glen Weyl|arXiv (Cornell University)|Aug 31, 2014

Auction Theory and Applications8 references3 citations

TL;DR

This paper rigorously analyzes Quadratic Voting (QV) in a large-population, independent private values setting, proving that as the number of voters grows, the mechanism achieves near-perfect utilitarian efficiency. It establishes that welfare losses relative to the optimal outcome vanish asymptotically, even under incomplete information and bounded value distributions, by showing that marginal pivotality becomes approximately equal across voters in equilibrium.

ABSTRACT

Voters making a binary decision purchase votes from a centralized clearing house, paying the square of the number of votes purchased. The net payoff to an agent with utility $u$ who purchases $v$ votes is $Ψ(S_{n+1})u-v^{2}$, where $Ψ$ is a monotone function taking values between -1 and +1 and $S_{n+1}$ is the sum of all votes purchased by the $n+1$ voters participating in the election. The utilities of the voters are assumed to arise by random sampling from a probability distribution $F_{U}$ with compact support; each voter knows her own utility, but not those of the other voters, although she does know the sampling distribution $F_{U}$. Nash equilibria for this game are described. These results imply that the expected inefficiency of any Nash equilibrium decays like $1/n$.

Motivation & Objective

• To provide the first rigorous game-theoretic analysis of Quadratic Voting (QV) in a Bayesian-Nash equilibrium framework with incomplete information.
• To characterize the structure of Bayes-Nash equilibria in QV under bounded value distributions and independent private values.
• To establish that the mechanism achieves asymptotic utilitarian efficiency as the population size N tends to infinity.
• To resolve the discrepancy between heuristic arguments (e.g., equal marginal pivotality) and actual equilibrium behavior, showing that while marginal pivotality is not exactly equal, it becomes approximately equal in large markets.
• To formalize the conditions under which QV converges to efficient outcomes, despite the failure of the key heuristic assumption of equal marginal pivotality.

Proposed method

• The analysis is conducted in a canonical independent private values model where each voter has a privately known value u for one of two alternatives.
• The mechanism uses a quadratic cost function: voters pay the square of the number of votes they purchase, with the outcome determined by the sum of all votes.
• The authors model the system using a smoothed utility function to ensure continuity in the vote total, enabling rigorous analysis of equilibrium behavior.
• They employ large deviations and concentration of measure techniques to show that, with high probability, the ratio of marginal pivotalities across voters converges to 1 as N → ∞.
• The proof relies on analyzing the function H(ξ, w), which captures the expected payoff difference, and establishes uniqueness of the equilibrium point via continuity and monotonicity arguments.
• Lemmas are used to prove the existence and uniqueness of a critical value ξ > δ such that the maximum of H(ξ, w) over w ∈ [−δ, δ] is zero, which characterizes the equilibrium behavior.

Experimental results

Research questions

• RQ1 Does Quadratic Voting achieve asymptotic utilitarian efficiency in large markets with incomplete information and bounded private values?
• RQ2 What is the structure of Bayes-Nash equilibria in QV, and how does it differ from the heuristic assumption of equal marginal pivotality?
• RQ3 How do welfare losses in QV scale with population size, and do they vanish as N → ∞?
• RQ4 Under what conditions is the quadratic cost function uniquely efficient in aligning individual incentives with social welfare?
• RQ5 Can the heuristic rationale for QV — that equal marginal pivotality leads to efficient voting — be rigorously validated in equilibrium?

Key findings

• For all bounded value distributions, the utilitarian welfare loss of Quadratic Voting, as a proportion of maximum possible welfare, tends to zero as the population size N approaches infinity.
• Although the heuristic assumption that all voters have equal marginal pivotality fails in equilibrium, the ratio of marginal pivotalities between any two randomly selected voters converges to 1 with high probability as N increases.
• The mechanism achieves asymptotic efficiency even when voters have private, bounded values and face incomplete information, validating its robustness in large markets.
• The existence and uniqueness of a critical value ξ > δ such that max_w∈[−δ,δ] H(ξ,w) = 0 ensures the existence of a unique equilibrium point in the model.
• The function H(ξ,w) has a unique maximizer w* ∈ (−δ,δ), and this point corresponds to the equilibrium vote demand, confirming the stability of the equilibrium structure.
• The proof establishes that the function h(α) := max_w∈[−δ,δ] H(α,w) is continuous and strictly decreasing for α ≥ δ, which is essential for proving the existence of the unique ξ satisfying the equilibrium condition.