27 August 2013

Five finger discounts

'Cleaning House: The Impact of Information Technology Monitoring on Employee Theft and Productivity' [PDF] by Lamar Pierce, Daniel Snow and Andrew McAfee examines how US "firm investments in technology-based employee monitoring impact both misconduct and productivity".
We use unique and detailed theft and sales data from 392 restaurant locations from five different firms that adopt a theft monitoring information technology (IT) product. Since the specific timing of individual locations’ technology adoption is plausibly exogenous, we can use difference-in-differences models to estimate the treatment effect of IT monitoring on theft and productivity within each location for all employees. We find significant treatment effects in reduced theft and improved productivity that appear to be driven by changing the behavior of individual workers rather than selection effects. Although workers with past patterns of theft appear more likely to leave treated locations than others, individual behavioral changes by existing workers drive restaurant-level improvements. These findings suggest multi-tasking by employees under a pay-for-performance system, as they increase effort toward sales following monitoring implementation in order to compensate for lost theft income. This suggests that employee misconduct is primarily a result of managerial policies rather than individual differences in ethics or morality.
The authors conclude -
In this paper, we show evidence that the adoption of information technology can substantially impact both the productive and corrupt behaviors of employees. Our results suggest that when management implements increased monitoring under a pay for performance scheme, employees will redirect effort toward productivity because their incentives have been realigned. Furthermore, our results suggest that the majority of improvement in organizational performance and productivity stems from the improved behavior of existing employees, not from the firing of those engaged in theft. The treatment of individual workers, not worker selection, appears to drive most productivity improvements and theft reductions. This does not mean that worker selection is unimportant in our story. In fact, those workers who stole under the weaker monitoring regime appear to self-select out of the more highly-monitored restaurants, perhaps to other more easily pilfered establishments.
Each of these results is highly consistent with a multi-tasking story where workers (agents) seek to trade off costly effort for income from either productivity under a pay-for-performance scheme or theft. Increased monitoring by management (principal) reduced the net gains from theft, necessarily increasing the equilibrium effort allocation toward productive behavior. In such a model, where workers are free to select out of the firm, increased monitoring also makes outside options more attractive, thereby increasing the likelihood of attrition for all workers who previously derived any income from theft. We note, however, that other cost-based worker activities remain unobservable in our data. We cannot, for example, observe whether reducing one type of theft (stealing revenue) through monitoring increases other forms of theft or misconduct such as inventory shrinkage. Given Olken’s (2007) results on substitution across types of corruption, such costs may very well exist and thereby reduce the profit gains from monitoring.
Another possible explanation for our results is that Restaurant Guard, by reducing the effort or attention required of managers toward theft monitoring, frees them to focus on managing service and food quality. Such a reallocation of managerial effort across dimensions could also result in the productivity and service quality improvements observed in our data. Although we are unable to separate these two mechanisms, both are based in the fundament multitasking tradeoff between misconduct and productivity. A technology tool that improves monitoring of employee misconduct has the potential to improve productivity both by providing financial incentives for employees to redirect effort and through freeing managerial attention toward improving production efficiency.
The results in this paper are important for at least three reasons. First, they represent the measurement of an important economic activity, employee theft, that has largely been observed only indirectly or anecdotally in firms. Although there is a considerable literature on corruption (e.g. Olken and Barron 2009), direct evidence on illegal behavior by firm employees is rare in the economics literature. Nagin et al. (2002) is an exception, demonstrating employee reductions in fraud following audit increases. We are able to show not only the direct effect of monitoring on theft, as they do, but also the secondary employee adjustments to other productive tasks to account for lost income. Second, our results illustrate the value of information technology when it complements human resource practices that motivate productive effort. Similar to arguments made by Bloom and colleagues (2012), the pay for performance system in American restaurants is likely important in how the IT monitoring system in our setting redirects effort from theft toward productivity. Finally, our results suggest a counterintuitive and hopeful pattern in human behavior: employee theft is a remediable problem at the individual employee level. While individual differences in moral preferences may indeed exist, realigning incentives through organizational design can have a powerful effect in reducing corrupt behaviors. This runs counter to the common view in the human resource management literature that productivity and integrity is largely about selection rather than managerial practice or technology (e.g. Ones et al. 1993). We show that firms can use information about employee theft not simply to fire the culprits, but rather to alter their behavior in ways that improve productivity.