#553 1966 · Carl Stevens / Major League Baseball · Labor relations / dispute resolution mechanism design
An economist fixed arbitration's incentive to lie by taking away the arbitrator's power to compromise
the problem
Conventional arbitrators split the difference, so both sides padded their opening offers
background
Labor and commercial disputes that couldn't be resolved through direct negotiation often ended up in front of an arbitrator empowered to hear both sides and issue a binding decision — and in practice, arbitrators facing two competing positions very often landed somewhere between them, a pattern both sides came to expect. Once that expectation was widely understood, it created its own incentive problem: a party that opened with a genuinely fair, moderate position was simply giving up ground, since the arbitrator's eventual compromise would still pull the outcome away from that fair number toward whatever extreme the other side had staked out.
Labor economist Carl Stevens studied this dynamic and, in a 1966 paper, argued that conventional arbitration actively discouraged the honest bargaining it was supposed to produce, because knowing a compromise was coming gave both sides every reason to posture rather than negotiate in good faith.
what everyone would do
Keep using conventional arbitration, where the arbitrator listens to both sides and crafts a compromise somewhere in between — the standard model for resolving disputes when two parties can't agree, and one that seems fairer precisely because it doesn't force an all-or-nothing outcome.
what they saw
The compromise itself was the problem: if both sides knew the arbitrator would split the difference between their positions, the rational move for each side was to open as far from a fair number as they could plausibly defend, since a more extreme opening position pulled the eventual midpoint further in their favor. Carl Stevens saw that the fix wasn't a smarter arbitrator or better guidelines for what counted as reasonable — it was removing the arbitrator's power to compromise at all. If the arbitrator had to pick one full offer or the other with nothing in between, submitting an extreme number no longer helped; it only made an opponent's more reasonable offer look better by comparison.
the move
Stevens proposed final-offer arbitration: instead of hearing both sides and crafting whatever settlement seemed fair, the arbitrator would be handed exactly two numbers — each side's best and final offer — and required to pick one of them in full, with no power to split the difference or compromise at all.
why it works
Because the arbitrator in final-offer arbitration can only choose one side's number in full, an extreme, self-serving offer becomes a liability rather than an advantage — it makes the other side's number look moderate and defensible by contrast, increasing the odds the arbitrator picks against you. That flips both sides' incentives toward converging on genuinely reasonable numbers close to what an independent observer would consider fair, which in turn makes the two positions similar enough that many disputes settle before a hearing even happens, once each side can see how close a reasonable version of the other's number would land to their own.
the payoff
Major League Baseball adopted final-offer arbitration for player salary disputes in the 1973-74 offseason, and the mechanism has held ever since: because an unreasonable filing risks losing entirely to the other side's more defensible number, most cases that reach a formal filing still settle before an actual hearing takes place, as each side recalibrates once it sees how a genuinely reasonable version of the other's number would compare to its own. The 2022 salary dispute between the New York Yankees and Aaron Judge — filed at $21 million versus $17 million, settled at $19 million before a hearing — is a modern instance of exactly the dynamic Stevens predicted in 1966.
where it breaks
It only disciplines behavior if both sides genuinely fear losing outright — a party confident the arbitrator will side with them regardless of how extreme their number is, because of bias, precedent, or a weak counterpart, has no reason to moderate, and the whole incentive structure collapses back into one-sided extremity. And it depends on there being a credible independent arbitrator whose choice both sides actually respect and will abide by; without that credibility, neither side faces real risk in submitting an unreasonable number, since the threat of losing the hearing carries no weight.
what came after
Final-offer, or "baseball," arbitration has since spread well beyond baseball into commercial contract disputes, public-sector labor negotiations, and other fields where a neutral third party needs to resolve a dispute without inadvertently rewarding whichever side stakes out the most extreme position.
references
- [1]The Negotiation Tactic That Leads to More Reasonable SettlementsHarvard Business School Working Knowledge, 2023library.hbs.edu
- [2]Salary Arbitration: Burden or Benefit?Society for American Baseball Research, 2021sabr.org