Behavioral constraints on the design of subgame-perfect implementation mechanisms

We study subgame-perfect implementation (SPI) mechanisms that have been proposed as a solution to incomplete contracting problems. We show that these mechanisms - which are based on off-equilibrium arbitration clauses that impose large fines for lying and the inappropriate use of arbitration - have severe behavioral constraints because the fines induce retaliation against legitimate uses of arbitration. Incorporating reciprocity preferences into the theory explains the observed behavioral patterns and helps us develop a new mechanism that is more robust and achieves high rates of truth-telling and efficiency. Our results highlight the importance of tailoring implementation mechanisms to the underlying behavioral environment.