Conflictoring Protocol: Holmström, Milgrom, and the contracts of system welfare

Optimal bilateral contracts can create Hollow Wins

The paper turns three Holmström-Milgrom results into the Contractual System-Welfare Neglect Theorem: when bilateral surplus is easier to contract on than system welfare, the incentive contract rewards the measurable surplus task and underprovides the welfare-preserving task.

• A Hollow Win is the (0,1,1) case: system welfare loses while the principal and agent reach their bilateral optimum.
• The mechanism is contractibility, not bad intent: x_π enters the contract; W is noisier, slower, harder to verify, or legally excluded.
• The paper applies the same model to LIBOR manipulation, Volkswagen Dieselgate, and the opioid crisis.

Signals excluded from s(·) are treated as nonexistent

Holmström's informativeness principle says every signal that improves inference about effort should enter compensation. The paper uses the reverse implication: when a welfare signal cannot be verified cheaply enough, the optimal bilateral contract leaves it out.

• The formal test is whether the likelihood ratio ∂ln f(x,ya)/∂a still depends on y after conditioning on x.
• System-welfare signals often have low channel capacity: long delay, high noise, high verification cost, or legal non-contractibility.
• The principal is optimizing over S_π = s(x_π) , not over contracts that can pay directly on W.

High-powered rewards for x_π pull effort away from e_W

The paper maps Holmström and Milgrom's multitask model onto system welfare. The agent allocates effort across e_π, which produces contractible surplus, and e_W, which preserves a system variable outside the contract.

• With quadratic effort cost C(e)=½e′Ce and linear pay s(x)=α+β′x, tasks with zero incentive weight can receive too little effort even when their social marginal product is positive.
• Under strict substitution, raising effort on the surplus task raises the opportunity cost of welfare-preserving effort.
• Adding more measures does not solve the problem when the added measures are noisier or more manipulable than x_π.

Efficient opacity hides system damage until the regime changes

The paper uses Holmström's account of debt as the dynamic version of the multitasking result. Information-insensitive contracts save monitoring costs in normal times, then block early detection when the state turns distressed.

• In normal times, opacity lowers verification and monitoring costs for the contracting parties.
• After a regime switch, the suppressed information becomes decision-relevant, but the contract was built to avoid producing it.
• The paper names this dynamic the Missing System Trap: degradation remains invisible until collapse or enforcement.

The theorem requires a contractible surplus task and a noncontractible welfare task

The Contractual System-Welfare Neglect Theorem applies when a principal-agent relationship has at least two tasks, system welfare is less contractible than surplus, and the principal cannot directly observe or contract on the welfare task.

• The bilateral contract maximizes E[x−s(x)a*(s)] subject to incentive compatibility and participation.
• The planner's comparison includes W(a), either through direct effort choice or an enforceable welfare signal.
• The result is e_W < e_W^FB and, under strict substitution, e_π > e_π .

Neglect is not automatic when W becomes contractible

The paper narrows the theorem with boundary cases. If system welfare can enter compensation with verifiable marginal weight, or if agents receive direct mission utility from e_W, the Hollow Win result becomes a comparative-static risk rather than the full theorem.

• 3.1 requires that no welfare signal permits an enforceable marginal payment equal to ∂W/∂e_W at the relevant allocation.
• The theorem treats e_W as welfare-relevant but privately unrewarded inside the bilateral contract.
• The offset boundary rules out cases where higher e_π directly creates enough welfare gain to offset the loss from lower e_W.

Six outside principals convert missing welfare signals into pressure

The Conflictoring Protocol supplies signals that the bilateral contract excluded. Whistleblowers, plaintiffs, regulators, legislators, investors, and supranational bodies each make a different part of system welfare recordable or costly to ignore.

• Whistleblowers provide internal information; plaintiffs provide litigation risk; regulators provide enforcement threat.
• Legislators provide rule-change credibility; investors provide market discipline; supranational bodies provide cross-border coordination.
• When k ≥ k* agents activate at the same time, the Hollow Win becomes unsustainable because W is no longer invisible.

LIBOR rewarded trading profit while benchmark accuracy stayed outside pay

In the LIBOR case, the paper treats accurate rate submission as e_W and trading profit as e_π. Benchmark integrity was system welfare; trading gains were immediate, measurable, and privately paid.

• LIBOR referenced $350 trillion in contracts, so small benchmark distortions could transmit across a large contract base.
• Enforcement actions from 2012 to 2016 produced approximately $9 billion in sanctions against banks including Barclays, UBS, RBS, Deutsche Bank, JPMorgan, and Citigroup.
• Under Dodd-Frank §21F, whistleblower awards can be 10–30% of sanctions over $1 million; the paper argues retaliation risk kept k* unmet for too long.

The defeat device solved the wrong contract problem

Volkswagen's diesel engineers faced a measurable performance task and a less visible emissions-compliance task. Horsepower, torque, and fuel efficiency were salient to the firm and consumers; NO_x compliance was visible mainly under testing.

• The defeat device preserved performance while reducing emissions only under test conditions.
• The paper frames this as multitasking distortion: e_π was rewarded while e_W was weakly contractible.
• Barrett et al. (2015) estimated approximately 59 premature deaths in the United States from the excess emissions.

Opioid incentives paid for volume while public-health harm lagged

The opioid case fits the same structure: sales, distribution, and prescription volume were measurable surplus signals, while addiction, overdose, and community health damage were delayed system-welfare outcomes.

• The welfare task was not absent; it was outside the enforceable compensation signal at the point of sale and distribution.
• Litigation and enforcement later supplied external signals that the original bilateral arrangements did not price.
• The case shows why disclosure is too weak when the contract still rewards e_π more directly than e_W.

Field 17 makes e_W a named decision variable

Decision Accounting is the paper's operational repair. Field 17 requires each organizational decision record to include a system-welfare assessment, and the Five-Minute Test requires that an outside reader can reconstruct the decision.

• The repair changes what is contractible; it does not assume better motives from the agent or principal.
• Named accountability links welfare claims to decision makers rather than leaving W as aggregate background information.
• For benchmark integrity, emissions compliance, and public health, the record creates evidence that later principals can enforce.

The repair is incentive redesign, not more disclosure

The paper's policy conclusion is that transparency does not cure a Hollow Win when the contract continues to reward the measurable surplus task and omit the welfare task. Repair requires contractible welfare signals, named accountability, and multi-principal enforcement.

• The falsifiable comparative static is higher system-welfare destruction where measurement asymmetry between x_π and W is greatest.
• The bridge calibration reports β_W = 8.10 and Π = −7.10 per 1.00 of benchmark bilateral surplus.
• The central policy target is the feasible contract set: move from S_π = s(x_π) toward contracts and records that can attach consequences to W.