A solicitor calls to discuss a case. Her client was dismissed after raising a grievance about a bullying line manager. The employer says it was redundancy. The client says it was victimisation. The solicitor has a conference bundle, a schedule of loss, and a question: what are the prospects?
The standard approach is to read the papers, form a view on the evidence, and give a percentage. Sixty-forty. Fifty-fifty. Seventy-thirty if the documents are strong and the respondent's witnesses are weak. This is not wrong. But it answers only one question — who is likely to win? — and that is not the only question, and often not the most important one.
The question that usually matters more, and that practitioners almost never ask explicitly, is: what kind of strategic interaction is this?
Seven Problems, Not One
Game theory — the mathematical study of strategic interaction between rational agents — identifies a small number of canonical structures that recur across every domain where parties with competing interests must make decisions that depend on each other's choices. Employment litigation produces all of them.
The Prisoner's Dilemma arises when both sides would benefit from settling but each fears being exploited by the other's aggressive litigation. The dominant strategy for each individual — prepare for war — produces the worst collective outcome. Legal costs consume the settlement surplus. Neither side wanted a hearing; both end up in one.
The Game of Chicken arises when one or both sides are threatening escalation to force concession. Costs warnings, derisory Part 36 offers, refusal to engage with ACAS — all are variants of driving toward a head-on collision and daring the other side to swerve. The danger is that neither swerves. The game has two equilibria, both asymmetric: one side yields while the other does not. Which side yields depends on who can more credibly commit to not backing down. Schelling (1960) showed that the power to eliminate your own options — to throw away the steering wheel — is a source of bargaining strength. The employee who resigns and files a constructive dismissal claim has crashed. The employer who begins without prejudice discussions after receiving the ACAS notification has swerved. Both moves are rational within the chicken framework, and neither looks rational from outside it.
The Stag Hunt appears in cases where both sides would genuinely benefit from a cooperative approach — joint expert instructions, agreed facts, structured mediation — but cooperation only works if both commit. If one side defects (ambushes with new evidence, resiles from agreed positions), the cooperator is worse off than if they had never tried. The challenge is not incentives but trust.
The Battle of the Sexes is the quantum negotiation. Both sides prefer settlement to tribunal. They agree on that. They disagree on the figure. The problem is coordination, not conflict: how to converge on a number when each prefers a different one. Anchoring effects are decisive. The first credible offer frames the negotiation.
Deadlock is rare but important: both sides genuinely want a tribunal ruling. A union-backed claimant developing the law. An employer seeking binding EAT authority. Settlement would leave the legal question unresolved. Mutual litigation is not a failure of negotiation — it is the equilibrium. Practitioners who try to settle a deadlock case waste costs and miss the point.
The Volunteer's Dilemma governs group grievances and whistleblowing. Everyone benefits if someone acts, but the person who acts bears the personal cost. Retaliation, career damage, the well-documented pattern of institutional hostility toward those who raise uncomfortable truths. The equilibrium is that one person volunteers while everyone else free-rides. This is why whistleblowers are so often alone, and why collective action — when it can be coordinated — transforms the dynamics entirely.
The War of Attrition describes the employer who fights every claim to deter future ones. Selten's Chain Store Paradox (1978): individually irrational to fight a meritorious claim, but across a portfolio of potential claims the deterrent value of never settling can far exceed any single defence cost. The claimant is playing a one-shot game against an opponent playing a repeated one. The asymmetry is structural, not just financial.
Why This Matters
A practitioner who identifies the game structure knows something that a practitioner focused solely on merits does not. They know what the equilibrium looks like — not what should happen, but what will happen if both sides play rationally. They know where the equilibrium diverges from the efficient outcome, and therefore where intervention (mediation, Part 36 offers, costs applications) can shift the dynamics. They know which levers matter.
In a Prisoner's Dilemma, the lever is trust: creating mechanisms that make settlement safe. In Chicken, the lever is commitment: credible signals that you will not swerve. In the Stag Hunt, the lever is incremental cooperation: small gestures before large commitments. In Battle of the Sexes, the lever is anchoring: the first credible number wins. In a War of Attrition, the lever is cost: making the fight expensive enough to be irrational even within the deterrent framework. In each case, the strategic prescription follows from the structure. Get the structure wrong and the prescription is not just unhelpful — it is counterproductive. You cannot solve a coordination problem (Battle of the Sexes) with commitment devices (Chicken). You cannot solve a trust problem (Stag Hunt) with anchoring (Battle of the Sexes).
This is not abstract. A practitioner advising a client to "hold firm" in a Prisoner's Dilemma — where the equilibrium is mutual litigation and the efficient outcome is mutual settlement — is telling them to play the dominant strategy that produces the worst collective result. A practitioner advising "reasonable compromise" in a Game of Chicken is telling them to swerve, which may be sensible or may be catastrophic depending on whether the other side reads the compromise as cooperation or weakness. The advice is identical in both cases: hold firm, or be reasonable. The strategic context makes it right in one and wrong in the other.
The Mixed-Mode Problem
The most treacherous feature of real disputes is that the parties are often in different games. Rapoport's taxonomy of 2×2 games (1967) identified that the same objective conflict can be classified as different game types depending on each party's subjective model. Harsanyi's incomplete information framework (1967–68) formalised this: when players have different beliefs about the game structure, rational play by one side looks irrational to the other.
This happens constantly in practice. An employer conducting a protracted grievance process believes it is in a War of Attrition — time is on its side, the employee will tire. The employee, having reached breaking point, is in a Game of Chicken — every day is a decision to crash (resign, file) or swerve (accept, stay). The employer's bureaucratic delay, rational within its attrition frame, is experienced by the employee as escalation within a chicken frame. The employer is blindsided by the constructive dismissal claim because it thought there was still time. There was not. The employee had already crossed the threshold.
Conversely, an employer in chicken mode — calculating whether to settle or hold firm — may face an employee in attrition mode who keeps filing appeals and raising new complaints, believing persistence will eventually force concession. The employer's decisive move (reject outright, or make a final offer) arrives while the employee is still expecting continued process. The mismatch in strategic frames, not the disagreement about substance, is what causes the breakdown.
The practical implication is that diagnosing your own game is not enough. You need to diagnose what game the other side thinks it is playing. The single most valuable piece of strategic intelligence in any negotiation is not what the other side wants — it is what the other side thinks is happening.
What Game Theory Does Not Do
It does not predict outcomes. It does not replace judgment. It does not tell you whether the tribunal will believe your client's account of the meeting on 14 March. The payoff matrices are ordinal, not cardinal — they tell you which outcomes each side prefers, not what those outcomes are worth. The models assume rationality, and real litigants are not always rational. They assume complete information, and real litigants rarely have it.
What game theory does is something more modest and more useful than prediction. It provides a diagnostic framework. When a negotiation stalls, the question is not just why? but what kind of stall is this? A trust failure (Prisoner's Dilemma) requires different treatment from a coordination failure (Battle of the Sexes), which requires different treatment from a commitment contest (Chicken). A practitioner who can diagnose the structure can prescribe the intervention. A practitioner who cannot is guessing — and the guess, however experienced, is unstructured.
The seven archetypes are not a complete theory of employment litigation. They are a vocabulary. They give practitioners a way to name what they already see, to distinguish problems that look similar but are structurally different, and to match their strategic advice to the actual dynamics of the dispute rather than to a generic template of "be firm" or "be reasonable" that may or may not apply.
The Uncomfortable Inference
If game theory has been applied to contract law (Baird, Gertner and Picker, 1994), corporate governance, international relations, evolutionary biology, auction design, and the pricing of financial derivatives, the question is not whether it applies to employment litigation. It is why employment practitioners have not used it. The answer, probably, is the same answer that explains the resistance to base rates, Bayesian updating, and every other formal framework that has been offered to a profession that prizes judgment over method: it feels reductive. It feels like it diminishes the role of experience, intuition, and the irreducible complexity of human conflict.
The feeling is understandable. It is also wrong. Game theory does not replace experience. It organises it. A practitioner who has handled hundreds of settlement negotiations has an intuitive sense of when to push and when to concede, when to be firm and when to be flexible. That intuition is real and valuable. What game theory adds is a structure that makes the intuition testable — that converts I think we should hold firm here into this is a commitment contest, and holding firm is the rational play because the other side's threat is not credible, and here is why. The conclusion is the same. The reasoning is visible. And visible reasoning can be examined, challenged, and improved in ways that intuition cannot.
Employment law is not special. The disputes it generates are strategic interactions between parties with competing interests, incomplete information, and choices that depend on each other's actions. That is the definition of a game. The only question is whether practitioners want to play it with or without a map.
Explore all seven archetypes interactively — with payoff matrices, outcome scenarios, and strategic guidance for each.
Open the Strategy Lab →References
- Baird, D. G., Gertner, R. H. & Picker, R. C. (1994), Game Theory and the Law, Harvard University Press
- Harsanyi, J. C. (1967–68), 'Games with Incomplete Information Played by "Bayesian" Players', Management Science, 14(3/5/7)
- Rapoport, A. (1967), 'Exploiter, Leader, Hero, and Martyr: The Four Archetypes of the 2×2 Game', Behavioral Science, 12(2), 81–84
- Schelling, T. C. (1960), The Strategy of Conflict, Harvard University Press
- Selten, R. (1978), 'The Chain Store Paradox', Theory and Decision, 9(2), 127–159