The Calabresian Third-Party Welfare Theorem: Rule Choice, Accident Cost Allocation, and the System Welfare Gap

Bilateral rule choice ignores a third-party welfare class S, creating a systematic gap

The Calabresi–Melamed typology evaluates property rules, liability rules, and inalienability only on A–B welfare. The theorem proves that the A–B-optimal rule diverges from the full-welfare-optimal rule whenever the welfare weight λ on the non-party class S exceeds an explicit switching threshold.

• S includes pooled risk-bearers, externality recipients, future generations, and the tax base
• Each protection regime imposes a structurally distinct welfare loss L_S(r) on S
• The switching condition: λ > (W₀(r* ) – W₀(r')) / (L_S(r') – L_S(r* ))

Calabresi identified the S problem from 1961 onward but never formalized it

Calabresi's three-cost framework (primary, secondary, tertiary) implicitly involves S: secondary costs map to pooled risk-bearers, tertiary costs to the tax base. Yet the 1972 typology and all subsequent refinements stayed bilateral.

• 1961: secondary costs (loss spreading) necessarily involve a population larger than A–B
• 1972: inalienability briefly mentions third-party effects but never decomposes S-welfare by rule type
• 1991: 'The Pointlessness of Pareto' names the gap but does not close it

Fifty years of extensions stayed bilateral; no theorem closed the S-welfare gap

Kaplow and Shavell (1996) gave the most rigorous bilateral welfare comparison; Ayres and Talley (1995) added bargaining; Bell and Parchomovsky (2002) added pliability. None gave S a distinct welfare function.

• Kaplow–Shavell: liability rules dominate for harmful externalities, but welfare metric is A–B only
• Rosenberg (1984) identified diffuse harm empirically but did not derive the welfare theorem
• No paper formally characterizes divergence between r* and r* across all three regimes

Welfare function extends Calabresi's three costs with a λ-weighted S-loss term

W(r) = U_A(r) + U_B(r) – C_prim(r) – C_sec(r) – C_tert(r) – λ·L_S(r). When λ=0, we recover the Calabresi–Melamed benchmark. S comprises four sub-populations mapped to Calabresian costs.

• S_pool: pooled risk-bearers (secondary costs)
• S_ext: externality recipients (primary costs)
• S_fut: future generations (intergenerational transmission)
• S_tax: tax base (tertiary costs)

Rule-Class Divergence: r* ≠ r* when λ exceeds the switching threshold

For any non-degenerate activity, there exists a λ* such that for λ > λ*, the full-welfare-optimal rule differs from the bilateral-optimal rule. The welfare gap Δ = W(r* ) – W(r* ) is positive exactly when λ·(L_S(r* ) – L_S(r* )) > W₀(r* ) – W₀(r* ).

• λ* = (W₀(r* ) – W₀(r')) / (L_S(r') – L_S(r* )) for the alternative rule r'
• Comparative statics: λ* increases with bilateral-welfare advantage of r* , decreases with S-loss reduction from switching
• Calabresi–Melamed (1972) is the λ=0 special case

Each rule imposes a structurally distinct welfare loss on S

Property rules generate holdout and transaction-failure externalities on S; liability rules produce systematic undercompensation of diffuse harm; inalienability creates enforcement cost and deadweight loss distributed across the tax base.

• L_S(P): holdout externalities when S cannot organize collective buyout
• L_S(L): undercompensation because court damages cannot price diffuse, non-monetizable losses
• L_S(I): enforcement costs and forgone gains from trade borne by taxpayers

Three real-world regimes show the theorem is tractable with existing data

CERCLA/Superfund (liability rule with S-welfare features), products-liability class actions (institutional response to L_S^liab), and FDA inalienability regimes (enforcement-cost quantification) demonstrate that revealed-λ methodology can be operationalized.

• CERCLA: EPA National Priorities List cost data show S-welfare features (joint and several liability, natural resource damages)
• Products liability: Federal Judicial Center settlement data reveal systematic undercompensation of diffuse harm
• FDA: National Drug Control Budget and Bureau of Justice Statistics data quantify enforcement deadweight

Five structural parameters determine when divergence occurs

Activity scale α, causal diffusion δ, pool size n, discount rate ρ, and verifiability ν all sign the comparative statics of λ* and the welfare gap.

• Higher α and δ increase L_S(L) relative to L_S(P), making property rules more attractive for S
• Larger n amplifies holdout costs under property rules, favoring liability rules
• Lower ρ (less discounting of future) raises L_S(I) because enforcement costs compound
• Lower ν (harder verification) increases L_S(L) because undercompensation worsens

Bilateral optimization defaults to the Hollow Win cell (0,1,1) for S

The eight-outcome taxonomy (C, A, B) shows that without λ-weighting, the bilateral optimum systematically produces outcomes where S loses (C=0) while A and B gain (A=1, B=1). Correcting this requires institutional mechanisms that internalize S-welfare.

• Hollow Win: S bears uncompensated loss while A and B split the surplus
• Policy levers: class actions, regulatory taxes, public-health offsets, intergenerational trust funds
• The theorem provides a formal basis for Calabresi's call for 'modified utilitarianism'

The theorem closes the gap between Calabresi's rule typology and his Pareto critique

The Calabresian Third-Party Welfare Theorem proves that rule choice systematically redistributes welfare to S, that the bilateral optimum diverges from the full optimum under identified conditions, and that the gap is empirically tractable. It recovers Calabresi–Melamed as a special case and supplies the missing formal bridge to Calabresi's later work.

• is the tort-theoretic dual of the Missing System Theorem (Postnieks 2026a)
• Posnerian wealth maximization shares the same structural blind spot (Postnieks 2026)
• Future work: dynamic rules, endogenous λ, and empirical tests of predicted regime choice