Automation, Job Design, and Productivity : Field Evidence

In jobs where the cost of effort exhibits increasing differences in separate tasks, automation increases productivity by directly eliminating the automated tasks and indirectly by reducing the marginal cost of non-automated tasks. Here, I report a field experiment rotating supermarket cashiers between conventional (where they scanned and collected payment) and scan-only checkouts. Consistent with increasing differences in separate tasks, at conventional checkouts, cashiers who scanned faster collected payments more slowly. At scan-only checkouts, cashiers scanned 10 percent faster, consistent with lower marginal cost of effort in the non-automated task. The faster scanning was not due to learning, less task-switching, or differential shirking