Deception: The role of uncertain consequences

We study how decisions to lie extend to risky environments. We provide experimental evidence from a sender-receiver game where there is uncertainty over the amount by which a sender's lie reduces its receiver's payoff, which is known only to potential liar. Even though all reduction amounts are equiprobable, ex-post beliefs elicited from senders suggest that, unlike truth-tellers, most liars underestimate the extent of the actual reduction in the receiver's payoff and appear to exploit this self-serving bias, resulting in substantially more lying relative to a baseline treatment without the uncertainty. Subsequent treatments confirm the bias by either providing additional evidence or by removing possible confounds. An intervention treatment nudging senders toward correcting the bias reduces lying.