We study a model of financial reporting where investors infer the precision of reported earnings. Reporting a larger earnings surprise reduces the inferred earnings precision, dampening the impact on firm value of reporting higher earnings, and providing a natural demand for smoother earnings. We show that for sufficiently “bad” news, the manager under‐reports earnings by the maximum, preferring to take a “big bath” in the current period in order to report higher future earnings. If the news is “good,” the manager smoothes earnings, with the amount of smoothing depending on the level of cashflows observed. He either over‐reports or partially under‐reports for slightly good news, and gradually increases his under‐reporting as the news gets better, until he is under‐reporting the maximum amount for sufficiently good news. This result holds both when investors are “naïve” and ignore management’s ability to manipulate earnings, or “sophisticated” and correctly infer management’s disclosure strategy.

We use a simulation model for a pharmaceutical R&D program to examine the tradeoff between objectivity and relevance of accounting information under various methods of R&D reporting. A simple capitalization rule, similar to the successful‐efforts method of capitalizing oil and gas exploration costs, provides a stronger relation between accounting information and economic values than immediate expensing of R&D outlays or capitalizing the full cost of outlays. The superior relevance of this “successful‐efforts” method persists even when earnings management is widespread.

We use Dealscan , a database of private corporate lending agreements, to provide large–sample tests of the debt covenant hypothesis. Dealscan offers several advantages over the data available in previous studies, principally larger and more representative samples and the availability of extensive actual covenant detail. These advantages allow us to construct powerful tests in which we find clear support for the debt covenant hypothesis. We also use these data to provide broad evidence on the economic role of debt covenants. We find that private lenders set debt covenants tightly and use them as “trip wires” for borrowers, that technical violations occur relatively often, and that violations are not necessarily associated with financial distress. Finally, since we measure covenant slack directly, we report evidence that the extensively–used leverage variable is a relatively noisy proxy for closeness to covenants.

This study examines the association between activity‐based costing and manufacturing performance. Results using a cross‐sectional sample of manufacturing plants indicates that extensive ABC use is associated with higher quality levels and greater improvements in cycle time and quality, and is indirectly associated with manufacturing cost reductions through quality and cycle time improvements. However, on average, extensive ABC use has no significant association with return on assets. Instead, we find weak evidence that the association between ABC and accounting profitability is contingent on the plant’s operational characteristics.

Using a field study approach, we examine two competing perspectives on direct labor variance reporting: some argue that direct labor variance reporting is costly and cumbersome, and should be eliminated; whereas others contend that without direct labor variance information, managers will not be able to monitor workers effectively, causing workers to shirk and worker productivity to decline. Specifically, we investigate the productivity and quality impacts of eliminating direct labor variance reporting with panel data containing 36 months of data from seven experimental plants that eliminated direct labor variance reporting and 11 control plants that did not. The experimental plants experienced a significant decline in labor productivity compared to the control plants. Also, the experimental plants showed an improvement in product quality, indicating that workers reallocate their efforts to other tasks as a result of the change in the information set available to evaluate them.