We show that firms׳ ability to avoid taxes is affected by the quality of their internal information environment, with lower effective tax rates (ETRs) for firms that have high internal information quality. The effect of internal information quality on tax avoidance is stronger for firms in which information is likely to play a more important role. For example, firms with greater coordination needs because of a dispersed geographical presence benefit more from high internal information quality. Similarly, firms operating in a more uncertain environment benefit more from the quality of their internal information in helping them to reduce ETRs. In addition, we provide evidence that high internal information quality allows firms to achieve lower ETRs without increasing the risk of their tax strategies (as measured by ETR volatility). Overall, our study contributes to the literature on tax avoidance by providing evidence that the internal information environment of the firm is important for understanding its tax avoidance outcomes.

Although employees' time estimates are used extensively for costing purposes, they are prone to measurement error. In an experimental setting, we research how measurement error in time estimates varies with: (1) the level of aggregation in the definition of costing system activities (aggregated or disaggregated); (2) task coherence (the extent to which the activities that require time estimates present themselves coherently or incoherently); and (3) when notice is given that time estimates will be required (in advance or after the fact), that is, whether participants know that time estimates will be required before they perform the activities. We also test on response mode (estimates in percentages or absolute time units). The results suggest an important trade-off between the level of aggregation and measurement error: increasing aggregation in the definition of activities leads to lower measurement error. Also, advance notification reduces measurement error, especially in settings with aggregated activities or incoherent tasks. Finally, we find a strong overestimation bias when participants provide time estimates in minutes, which may be problematic for Time-Driven Activity-Based Costing that advocates the use of estimates in minutes. These results are relevant to accountants and decision makers who want to assess and control the measurement error in their costing system and to professionals in related areas that make use of time estimates (e.g., billing, tendering).

Cost accounting systems provide accurate costs only under stringent conditions. However, we know little about the nature, level, and bias of costing errors. This paper reports the results of a simulation study of two-stage cost allocation systems that provide the following main insights: (1) partial improvement in the costing system usually increases the overall accuracy of reported product costs except in specific cases identified in this paper where errors have an offsetting effect, most notably when there is aggregation error in the activity cost pools and measurement error in the resource drivers; (2) the impact of Stage II costing errors on overall accuracy is stronger than that of Stage I errors, so system refinements should focus on Stage II; and (3) the presence of aggregation and measurement errors usually results in relatively more products being under- than over-costed, with large amounts of over-costing for a few “big-ticket” (in dollar terms) products, and small amounts of under-costing for a larger number of less expensive products.

ABSTRACT Using Census microdata on 28,000 manufacturing plants, we examine how firms manage employee retention concerns. In response to reductions in the local unemployment rate, plants take additional steps beyond increasing compensation. First, plants adjust bonus architecture to ensure bonuses can be paid. Second, plants offer more agency to employees by deploying high-involvement work practices that generate longer-term commitment. Third, plants pull these retention levers less when they have high availability and use of data as this reduces the adverse effects of employee turnover on organizational knowledge. These results are robust to using the fracking revolution as a shock increasing firms’ retention concerns. Additionally, we observe that although compensation increases tend to spill over to other plants within the same firm—aligning with theories of inequity aversion—adjustments to bonus architecture and the provision of employee agency do not, suggesting these may be more cost-effective strategies for multiplant firms. JEL Classifications: J63; M51; M54.

We examine the relation between plant-level predictive analytics use and centralization of authority for more than 25,000 manufacturing plants using proprietary US Census data. We focus on headquarters' authority over plants through delegation of decision-making and design of performance-based incentives. We find that increased predictive analytics use is associated with reduced delegation of decision-rights to local managers, increased centralization of control over data gathering and reduced plant managerial payrolls. In terms of incentives, predictive analytics use is associated with more accurate targets and tighter linkages between rewards to workers (performance-based bonuses, promotions and firings) and measured performance. Overall, our findings suggest that predictive analytics use is associated with increased centralization of authority in headquarters.

ABSTRACT We investigate the dynamics of internal forecasting in multi‐location firms and the relations between forecast characteristics and investment. Using U.S. Census microdata on plant‐level growth expectations, we find that plants within multi‐location firms make forecasts that are both more certain and less accurate than those of standalone plants. We provide evidence suggesting that headquarters infers uncertainty from inter‐plant forecast disagreement, and that headquarters is able to facilitate the use of relevant information held by one plant but applicable to another. Differences between peer and focal plants' forecasts predict forecast errors and relate to investment decisions at the focal plant, suggesting that information from multiple sources is integrated into capital allocation decisions. Headquarters' heavier use of peer plant forecast information when focal plants are more uncertain likely weakens focal plant managers' incentives to consider extreme scenarios when forecasting.