The Economics of Machine Verification: Verification-Cost Shocks and the Extensive Margin of Monitoring

Economics understands how costly verification disciplines behavior. This paper asks which claims become worth verifying at all when machine intelligence changes the technology of verification. I model verification as a technology vector: fixed and marginal costs, four pipeline reliabilities from detection to enforcement, and a false-positive rate. Three results follow. First, a verification frontier: the minimum economically contestable claim L*(τ,N) in claim value and claim volume, with component-wise comparative statics and a false-challenge-dominance condition under which monitoring entry is monotone. Second, recovery is the wrong success metric: under bounded misstatement gains and attainable full deterrence, measured recoveries are hump-shaped in verification effectiveness, and in general a technology that works can reduce measured recoveries. Third, private and social verification frontiers diverge: transfers and externalized dispute costs push private verification toward excess; uninternalized deterrence spillovers push it toward undersupply. The broader implication is institutional. Much of commercial architecture is adaptation to costly checking, and where the pipeline clears, a verification-cost collapse re-selects it. The open question is how much economic activity lies below the human verification frontier but above the machine one.