ABSTRACT We examine the causal effects of corporate governance on earnings management using shareholder‐sponsored proposals that pass or fail by a small margin of votes in annual shareholder meetings. This setting provides a causal estimate that overcomes concerns of endogeneity. Specifically, compared with firms whose shareholder proposals fall just short of a majority threshold, firms whose shareholder proposals narrowly pass have similar characteristics but a discretely higher likelihood of implementing improvements in governance. As such, we expect that firms whose shareholder proposals pass the threshold by a small margin exhibit a significantly lower level of earnings management. Employing a regression discontinuity design, we find results that support our expectation based on the propensity to just meet or beat analysts' forecasts by one cent as a proxy for earnings management. In addition, we show that the results are driven by governance changes that increase directors' monitoring. Our results are robust to using discretionary accruals as an alternative measure of earnings management. Collectively, the results suggest that improvements in corporate governance curtail earnings management, and support the underlying premise of regulators that improvements in corporate governance would improve financial reporting.

ABSTRACT We consider an investor's choice of conservative reporting, bonus payments, and investment decisions in the presence of the hidden‐information agency problem of a manager's productivity and the hidden‐action agency problem of a manager's diversion of resources. It is important to consider the hidden‐action and hidden‐information agency problems in isolation and their interaction to gain insights into the drivers of demand for conservatism. We show that the conservative (nonconservative) regime is optimal for the high‐productivity (low‐productivity) manager when both agency problems exist, even though the nonconservative regime is optimal for both the high‐ and low‐productivity managers when only the hidden‐action or the hidden‐information problem exists. Essentially, the low‐productivity manager can misrepresent as the high‐productivity manager to obtain high investment levels and divert resources only in the presence of both agency problems. This added layer of agency problem creates the demand for conservatism and highlights the importance of the interaction between the hidden‐action and hidden‐information agency problems. Furthermore, we show that as the conservatism level increases (i) the optimal investment level conditional on a good report for the high‐productivity manager increases and approaches the first‐best level (i.e., ex‐post investment efficiency increases); (ii) the expected optimal investment level for the high‐productivity manager decreases and diverges from the expected first‐best level (i.e., ex‐ante investment efficiency decreases); and (iii) the expected bonus payment to the high‐ and low‐productivity managers decreases. These findings provide insights into how the demand for conservatism arises in the presence of both hidden‐information and hidden‐action agency problems and provide empirical guidance relating conservatism to investment efficiency.