Lies, Labels, and Mechanisms

We test whether lying aversion can steer equilibrium selection in mechanism design. In a principal-worker environment, the direct mechanism admits two dominant-strategy equilibria: the designer’s target and a worker-optimal outcome. We show this limitation persists for all robust mechanisms, then ask whether framing misreports as explicit lies helps. We develop a 2X2 experiment that varies direct vs. extended mechanisms with implicit vs. explicit messages. We find that framing misreporting of type as an explicit lie shifts play away from the worker-optimal outcome toward truthful reporting, raising designer payoffs with minimal efficiency loss. These findings indicate that lying aversion is an effective lever for aligning behavior with social objectives.

💡 Research Summary

The paper investigates whether the behavioral tendency known as “lie aversion” can be used as a design lever to resolve the classic problem of multiple equilibria in mechanism design. The authors focus on a principal‑worker adverse‑selection environment in which the principal (designer) wishes to assign tasks and salaries based on workers’ private expertise (expert or beginner). In the standard direct‑revelation mechanism, two dominant‑strategy equilibria exist: (i) a truthful equilibrium that implements the designer’s social‑choice function (SCF) and (ii) a worker‑optimal equilibrium in which both workers misreport their type as “expert,” thereby achieving a Pareto‑dominant outcome for themselves.

The first theoretical contribution shows that this limitation is generic: any mechanism that robustly implements the designer’s SCF (in the sense of Bergemann and Morris, 2005) also robustly implements the worker‑optimal SCF. Proposition 1, a generalization of Repullo (1985), formalizes that any ex‑post implementation of the designer’s SCF necessarily admits an ex‑post equilibrium that reproduces the worker‑optimal SCF. Hence, robust implementation alone cannot eliminate the undesirable equilibrium.

To break this impasse, the authors turn to a well‑documented behavioral regularity: people are averse to lying. Meta‑analyses of die‑rolling experiments (e.g., Fischbacher & Föllmi‑Heusi, 2013) find that subjects forgo roughly three‑quarters of the monetary gain from lying. The authors hypothesize that if misreporting is framed as an explicit lie, lie‑averse workers will prefer truth‑telling even when a cooperative deviation would increase their payoffs.

The experimental design is a 2 × 2 between‑subjects factorial that varies (1) the mechanism type—direct (2 × 2) versus extended (3 × 3) with a “leave blank” option that removes weak dominance of the deceptive action, and (2) the framing of messages—implicit (neutral labels) versus explicit (labels that explicitly call the deceptive report a “lie”). The four treatments are denoted 2 × 2‑I, 2 × 2‑E, 3 × 3‑I, and 3 × 3‑E. In each round, two workers are randomly assigned private types (expert or beginner) and simultaneously choose a message; a third participant (the “staffer”) plays the role of the mechanism designer, receiving no strategic choices but earning payoffs that depend on the workers’ types and reports.

Results are striking. In the explicit‑framing conditions (2 × 2‑E and 3 × 3‑E), the frequency of the worker‑optimal equilibrium drops dramatically, while the truthful equilibrium becomes the dominant observed outcome. This shift translates into higher designer payoffs with only a modest loss in overall efficiency (the extended mechanism’s “blank” option slightly reduces the effect but still outperforms the implicit conditions). In contrast, the implicit‑framing conditions (2 × 2‑I and 3 × 3‑I) replicate the theoretical prediction: the worker‑optimal equilibrium is observed frequently, confirming that robust implementation alone does not eliminate the undesirable outcome.

The authors interpret these findings as evidence that lie aversion can be harnessed as a practical design tool. By simply labeling a deceptive report as a “lie,” the mechanism taps into a psychological cost that outweighs the monetary incentive to coordinate on the worker‑optimal outcome. The extended mechanism further shows that weakening the strategic dominance of the deceptive action (through a “leave blank” option) can complement the framing effect, though the framing alone yields the larger impact.

The paper contributes on three fronts. First, it extends the impossibility result of robust implementation to a concrete principal‑worker setting, clarifying that any mechanism robustly implementing the designer’s SCF also implements the worker‑optimal SCF. Second, it demonstrates experimentally that behavioral preferences—specifically lie aversion—can break this impossibility, shifting equilibrium selection toward the designer’s desired outcome. Third, it offers a low‑cost, easily implementable policy prescription: framing misreports as explicit lies can improve mechanism performance in a variety of real‑world contexts such as employee evaluations, public service applications, or any setting where self‑reported private information determines allocation.

Overall, the study bridges the gap between abstract implementation theory and behavioral economics, showing that incorporating well‑known psychological biases into mechanism design can resolve otherwise intractable equilibrium selection problems.