Queueing Systems with Rationally Inattentive Customers

Problem definition: Classical models of queueing systems with rational and strategic customers assume queues to be either fully visible or invisible, while service parameters are known certainty. In practice, however, people only have “partial information” on the environment, in sense that they not able discern prevalent uncertainties. This is because assessing possible delays rewards costly, as it requires time, attention, cognitive capacity, which all limited. On other hand, also adaptive endogenously respond information frictions. Methodology: We develop an equilibrium model for a single-server system having limited attention. Following theory inattention, we optimize their learning strategies by deciding type amount acquire act accordingly internalizing associated costs. Results: establish existence uniqueness customer when allocate attention learn uncertain queue lengths delineate impact characteristics. provide complete spectrum costs throughput show numerically might nonmonotone. reflected social welfare if firm’s profit margin high enough, although always suffers from Managerial implications: identify settings where firms planners should most cautious customers’ translate our results advisable provision design. For example, recommend avoid partial hindrance queue-length low-demand highly valued customers. popular value reasonably highly, particularly advisable. Academic/practical relevance: propose microfounded framework behavior links beliefs, rewards, It offers holistic perspective prevalence (and frictions) operational performance can extended analyze richer complex structures, rendering valuable tool Supplemental Material: The online appendix available at https://doi.org/10.1287/msom.2021.1032 .