Understanding the 'Numbers Game'

Two well-known stylized facts on earnings management are that the earnings surprise distribution has a discontinuity at zero, and that positive earnings surprises are associated with positive abnormal returns. We link these two facts in a model of the earnings management decision in which the manager trades off the capital market benefits of meeting earnings benchmarks against the costs of manipulation. We develop a new structural methodology to estimate the model and uncover the unobserved cost function. The estimated model parameters yield the percentage of manipulating firms, magnitude of manipulation, noise in manipulation, and sufficient statistics to evaluate proxies for identifying firms suspected of manipulation. Finally, we use the Sarbanes--Oxley Act as a policy experiment and find that by increasing costs, the Act reduced equilibrium earnings management by 36%. This reduction occurred despite an increase in benefits, consistent with the market rationally becoming less skeptical of firms that just meet benchmarks