Two-Level Games

In brief

In 1988 the political scientist Robert Putnam published an article in the journal International Organization titled "Diplomacy and Domestic Politics: The Logic of Two-Level Games." Its claim was simple and, once stated, hard to unsee. A national leader at the negotiating table is playing two games at the same time. At one table, Level I, they bargain with their foreign counterparts. At the other, Level II, they face legislators, interest groups, party factions, and voters who must accept whatever deal comes home. Neither game can be won alone. A pact that delights foreign partners is worthless if the domestic side rejects it, and a deal built to please the home audience may be one no foreign partner will sign. The negotiator sits at both tables, and every move at one is read at the other.

The full treatment

The problem it answers

Traditional international relations theory treated states as unitary actors: black boxes pursuing a single national interest, their internal politics irrelevant to how they behaved abroad. That abstraction is clean but often wrong. It cannot explain why leaders sign agreements their own legislatures then torpedo, why identical external pressures produce opposite responses in different countries, or why a president spends as much energy managing Congress as managing the counterpart across the table. Putnam wanted a framework that took domestic politics seriously without abandoning the strategic logic of bargaining between states. The two-level game is that framework: it keeps the state as an actor at Level I while opening the box to reveal the ratification fight at Level II.

How it works: win-sets

The engine of the model is the win-set. For any given negotiator, the win-set is the set of all possible Level I agreements that would win the necessary approval among their Level II constituents. Ratification need not be a formal legislative vote; it can be any process by which domestic actors can accept or veto a deal, from a parliamentary majority to the acquiescence of a governing coalition.

The central results follow from one geometric fact. An agreement is possible only where the win-sets of all parties overlap. If the sets do not intersect, no deal exists that everyone can ratify, and the talks fail no matter how skilled the diplomats. Larger win-sets are more likely to overlap, so they make agreement more likely. But larger win-sets are also weaker: a negotiator willing to accept a wide range of outcomes will be pushed toward the edge of their set that favors the other side. This is the model's signature tension. A small win-set makes agreement harder to reach but improves your terms when one is reached, because your partner knows there is little you can concede.

Why a weak hand can be a strong card

The most quoted insight comes from this tension, and it inverts common intuition. A negotiator whose domestic win-set is narrow, whose hands are visibly tied at home, can bargain harder abroad. "I would love to accept your terms, but my parliament will never ratify them" is a credible way to hold firm. The constraint becomes a resource. Putnam drew this directly from Thomas Schelling, whose 1960 book The Strategy of Conflict showed that the power to bind yourself, to burn a bridge or throw away the steering wheel, can be a source of bargaining strength. A leader who can point to a hostile legislature, an angry public, or a restive coalition has thrown away part of the wheel in full view of the other driver.

The mirror image is that a leader may try to expand a counterpart's win-set from the outside. Side payments, linkage to other issues, and appeals aimed over the head of a foreign government directly at its public are all moves that change what the other side can ratify at home.

The distinctions that matter

Putnam separated the negotiator's own preferences from the constraints they operate under, since a leader may personally want a deal their constituents will block, or may hide behind constituents to reject a deal they privately dislike. He distinguished voluntary defection (a leader breaking a deal because it serves them) from involuntary defection (a leader unable to deliver ratification), a distinction that matters because the second kind cannot be deterred by the usual threats. And he noted that the chief negotiator sits uniquely at both tables, which gives them room to manipulate information, telling each audience a version of the constraints that serves the overall bargain.

Lineage

The model is a piece of applied game theory, and it inherits that field's core assumption: outcomes depend on the interlocking strategies of rational players. Its most direct debt is to Schelling, who supplied the paradox of the bound negotiator. It arrived during a broader turn in international relations, in the same years, toward opening the unitary-state black box, a turn shared by liberal and institutionalist work on how domestic actors and international regimes interact. The intellectual cousin here is the study of bureaucratic and organizational politics inside government, whose landmark is Graham Allison's analysis of the Cuban missile crisis in Essence of Decision. Where Allison opened the state to reveal competing agencies and standard routines, Putnam opened it to reveal the ratification constraint that shadows every international bargain.

The strongest case for it

The model earns its place by explaining things the unitary-actor view cannot. It makes sense of ratification failures, the treaty signed in good faith and then rejected at home, as a structural feature rather than an accident or a lie. It clarifies why the same external offer meets acceptance in one country and refusal in another: the win-sets differ. It gives a precise vocabulary, win-set, ratification, the size and location of the set, to phenomena diplomats had always felt but rarely named. And it generates non-obvious, testable predictions, above all that domestic weakness can translate into international strength, which practitioners recognize the moment they hear it. Because it sits on game-theoretic foundations, it connects cleanly to the wider study of strategic bargaining while remaining usable by historians working through a single case. The 1993 volume Double-Edged Diplomacy, edited by Peter Evans, Harold Jacobson, and Putnam, showed the framework surviving contact with a range of real negotiations.

The strongest case against it

The sharpest objection is that the model is a rich metaphor more than a predictive theory. Its central object, the win-set, is nearly impossible to measure before a negotiation concludes. A win-set is often only revealed by whether a deal ratified, which makes the explanation dangerously circular: we know the sets overlapped because a deal happened, and we know they did not because it failed. Critics within the rationalist camp have pressed for tighter formal models that pin down these quantities rather than invoking them after the fact.

A second objection targets the two-level count itself. Real bargaining rarely stops at two tables. Coalition governments negotiate internally before facing anyone abroad; multilateral talks stack many domestic games onto one international game; supranational bodies and transnational lobbies add layers the metaphor does not name. Scholars have proposed three-level and multi-level extensions precisely because two levels flatten a deeper hierarchy.

Third, the assumption of a single chief negotiator who straddles both tables fits a strong presidency better than a fragmented executive, a fractious coalition, or a system where several officials speak abroad at once. When authority is dispersed, the tidy image of one player at two tables breaks down. Constructivist critics add a further charge: the model treats domestic preferences as fixed inputs, when in practice leaders and publics argue, persuade, and reframe what counts as an acceptable deal, so the win-set is not a given boundary but something politics itself produces.

Where it stands now

The two-level game is a fixture of how international relations is taught and how negotiations are dissected, from trade rounds to climate agreements to arms control. Its core terms, especially win-set and the strategic uses of a tied hand, have passed into the working vocabulary of the field and of many practitioners. It has not become a fully formalized predictive theory, and the measurement problem its critics identified remains real. But as a lens that forces the analyst to ask "who has to ratify this, and what will they accept," it has proved durable. Later work extended it to more levels and married it to formal bargaining models, yet the 1988 article is still the reference point. Its lasting achievement was to make the boundary between domestic and foreign politics, long treated as a wall, into the very hinge on which negotiation turns.

Test yourself

Think of a recent international agreement your country signed, or failed to sign. Before asking whether the foreign terms were good, ask a different question: who at home had to approve it, and what was the range of deals they could have lived with? Now ask whether the negotiator's public complaints about their own constraints were a genuine limit or a bargaining tactic. If you cannot tell, notice that the ambiguity is exactly where the two-level game locates its power.