Everlasting Fraud

This paper studies the dynamic interaction between corporate fraud and regulatory detection. We develop a theoretical framework in which managerial manipulation endogenously reduces the informativeness of earnings reports, while a regulator chooses detection effort to restore it. As manipulation goes undetected, information uncertainty accumulates and investors rely more heavily on the new earnings report; detection resolves fraud-induced uncertainty but resets the cycle. In multi-firm settings, the regulator reallocates attention based on relative fraud-induced uncertainty, yielding mean reverting manipulation and, depending on parameters, either synchronized or sequential fraud patterns. We then structurally estimate the model to identify key parameters governing fraud accumulation, detection intensity, and uncertainty recovery. Finally, we present reduced-form evidence consistent with the model: earnings response coefficients drop sharply upon fraud revelation and gradually recover thereafter.