We study the economic consequences of opportunities for dishonesty in an environment where efficiency relevant behaviour is not contractible, but rather incentivized by informal agreements in an ongoing relationship. We document the repeated interaction between a principal and an agent who, within our main treatment, was privately informed about the costs of effort provision being either high or low. At the beginning of the interaction, an agent could either truthfully report the cost type to the principal or choose to lie about it. We find that a substantial fraction of low cost agents decided to signal high costs. Dishonest low cost and honest high cost agents pool on the complete information outcome with high costs, as measured in our control treatment. The outcome of such pooling is less efficient than for honest low cost agents. Moreover, principals who face dishonest agents earn substantially less profits than those facing honest agents. Our evidence therefore suggests that informal agreements in a repeated interaction generate less efficient outcomes if dishonesty is possible but, at the same time, are robust to substantial degrees of deception. We furthermore show that our experimental findings can be organized using the logic of repeated games.
