Preview

PDF, English - main document Download (716kB) | Terms of use

Abstract

In a simplifying analytical framework with endogenous levels of actual and self-reported emissions, we consolidate the existing literature into three main hypotheses about the relative merits, for a resource-constrained regulator, of random (RAM) and competitive (CAM) audit mechanisms in the presence or absence of peer information about actual emissions. Testing the three hypotheses in a quasi-laboratory experiment (N = 131), we find supportive evidence that CAM always induce more truthful reporting than RAM. Moreover, we provide the empirical validation of the theoretical prediction that CAM can succeed in aligning actual emissions more closely with the social optimum in the presence of peer information when RAM cannot. Behavioral mechanisms prevent reaching the first-best outcome.