Private Information and Non-Participant Welfare: A General Impossibility Theorem

No mechanism can protect outsiders while keeping participants truthful, willing, and budget-balanced

The Non-Participant Welfare Impossibility Theorem proves that in any direct-revelation environment containing a bilateral-trade subproblem with overlapping private values, five constraints cannot hold simultaneously: Bayesian incentive compatibility, interim individual rationality, weak budget balance, ex-post welfare-neutrality for non-participants, and constrained efficiency on the outsider-neutral set.

• Non-participants have no message space, so the designer cannot elicit their types or use transfer tools to compensate them
• The proof reduces to Myerson-Satterthwaite (1983): information rents consume surplus, leaving no slack for outsider protection
• Any mechanism that satisfies the first four constraints must leave some outsider-neutral gains from trade unrealized

Every Satterthwaite impossibility evaluates welfare only over participants

From voting (1975) to bilateral trade (1983) to double auctions (1989, 1994) to physician markets (1992) to hospital mergers (2003) to dynamic matching (2007) to merger policy (2020), each result sacrifices some desirable property under private information—but always over the mechanism's participants, never over affected outsiders.

• Participants: voters, buyer and seller, traders, patients, merging firms and their consumers
• Outsiders: creditors, employees, suppliers, communities, uninsured patients, displaced workers
• The paper asks: what happens when we add outsider welfare as a binding constraint?

Participants have private types; non-participants have no message space

Participant class P has private one-dimensional types θ drawn from continuous distributions, quasi-linear utility, and a direct-revelation mechanism. Non-participant class N has types θ ^N that affect their welfare but cannot be elicited—they are not players in the game.

• Designer's objective: Π(M) = E[Σ v_i] + λ·E[W_N], where λ ≥ 0 weights outsider welfare
• Constraints: BIC, IIR, weak budget balance, and ex-post welfare-neutrality for N (never worse off than status quo)
• Ex-post welfare-neutrality is the analog of individual rationality for outsiders who cannot walk away

The 1983 theorem shows four constraints are incompatible in bilateral trade

Myerson and Satterthwaite (1983) proved that with a buyer and seller holding private values on overlapping supports, no mechanism can simultaneously satisfy BIC, IIR, budget balance, and ex-post efficiency. Information rents required for truthful reporting consume the surplus needed for budget balance.

• Budget balance is structurally equivalent to a degenerate non-participant welfare constraint
• The paper adds substantive outsider welfare on top of budget balance
• The impossibility becomes: five constraints cannot hold on the outsider-neutral slice

Bounded NPWI reduction theorem: five constraints are jointly infeasible

1: In a direct-revelation environment satisfying Assumption Set A (including a bilateral-trade subproblem with overlapping support and positive gains from trade), no mechanism can simultaneously satisfy BIC, IIR, weak budget balance, ex-post welfare-neutrality for N, and constrained ex-post efficiency on the outsider-neutral feasible set.

• Assumption A3: there exists a positive-measure event set E where fixing other reports yields a bilateral-trade subproblem
• Assumption A4: outsider-neutrality does not collapse to 'never trade'—some trades are outsider-neutral, some are harmful
• Assumption A5: constrained efficiency means maximizing participant surplus within the outsider-neutral set

Reduction to Myerson-Satterthwaite on the outsider-neutral slice

Step 1: Freeze bystanders and isolate the buyer-seller pair. Step 2: Outsider-neutrality restricts the feasible allocation set. Step 3: Constrained efficiency requires ex-post efficient trade on the outsider-neutral gains region. Step 4: Myerson-Satterthwaite says no BIC+IIR+WBB mechanism can achieve ex-post efficiency on that domain. Contradiction.

• The proof does not claim a new envelope theorem—it inherits standard BIC arguments
• The contradiction shows that outsider-safe mechanisms must leave some gains unrealized
• Boundary cases: correlated types, full-surplus extraction, and 'never trade' are outside the theorem's domain

Satterthwaite's corpus collapses into one outsider-exclusion wedge

The theorem recovers Myerson-Satterthwaite (1983) as the exact benchmark (N=∅, budget balance as degenerate outsider constraint). Five other applications identify the same wedge: voting (N=losing coalitions), double auctions (N=upstream/downstream firms), physician markets (N=sicker patients), hospital mergers (N=uninsured), and merger policy (N=workers, suppliers, communities).

• Gibbard-Satterthwaite (1975): strategy-proof voting impossible because losing coalitions are harmed
• Double auctions (1989, 1994): participant efficiency converges at O(1/m²), but outsider welfare does not automatically converge
• Physician report cards (2003): information design optimizes over doctors and patients but harms sicker patients through selection

Four instruments when the designer cares about outsiders (λ > 0)

When the designer places positive weight on non-participant welfare, the optimal mechanism uses four instruments: participation restrictions (exclude some participants), allocation distortions (bend trade away from participant-efficient), payment redistributions (tax participants to compensate outsiders), and commitment to information disclosure (reveal outcomes to enable outsider adaptation).

• No instrument can fully restore first-best—the impossibility is structural
• The second-best is characterized by the trade-off between information rents and outsider protection
• Institutional analogues exist in hospital merger conditions, privacy regulations, and worker adjustment assistance

Three forces that widen the outsider-participant welfare gap

First, higher dispersion of participant types increases information rents, leaving less surplus for outsider protection. Second, fragmented non-participant classes (many small groups) make collective action costly, reducing the effectiveness of redistributive instruments. Third, temporal asymmetry: information rents accrue immediately, while outsider costs (e.g., post-merger price effects, privacy losses) manifest with delay.

• Hospital mergers and privacy markets illustrate the information-rent multiplier
• Digital privacy is the clean case of N-class fragmentation: affected users are numerous and dispersed
• Temporal discounting means even a well-intentioned designer systematically under-weights outsider harm

The NPWI theorem and the Missing System Theorem together block two escape routes

The Missing System Theorem (Postnieks 2026) shows that bilateral optimization structurally excludes system welfare even under complete information. The NPWI theorem shows that even a designer who wants to include non-participant welfare cannot do so under private information while preserving truthfulness, participation, budget balance, and constrained efficiency. Together they constitute a double impossibility.

• Complete information: system welfare is excluded by the structure of bilateral optimization
• Private information: outsider welfare cannot be added without violating incentive constraints
• The two theorems cover the full information spectrum

Mechanism design must treat outsider welfare as a separate feasibility constraint, not an afterthought

The theorem implies that participant-centered mechanism design—even when optimal for participants—does not automatically protect outsiders. Policy applications (merger review, auction design, school choice, privacy regulation) must explicitly model outsider welfare as a binding constraint, not assume it will be internalized through participant optimization.

• Hospital merger review: insured-price analysis is rigorous, but uninsured access and safety-net capacity require separate instruments
• Auction design: bidder efficiency and revenue do not capture third-party or policy objectives
• The field's participant focus is not a modeling choice but a structural limitation: outsiders have no message space