On the Optimal Disclosure of Queue Length Information

Consider an information disclosure policy to be implemented by a service provider which informs customers about the current queue length when the queue is short (i.e., it is less than some threshold) and conceals the information when the queue is long. Simhon et al. (2016) proved that such a threshold policy, seemingly intuitive, never maximizes the server’s revenue in M/M/1 queues – when customers are aware of the policy and the threshold. We show that it is indeed optimal when customers do not have any information of the policy. Furthermore, we identify the socially optimal queue length disclosure policy. Interestingly, it is to hide the queue length information when the queue is short and announcing it when the queue is long. Such a policy even yields higher consumer welfare than full queue length disclosure (i.e., an observable queue).