Here’s an interesting story from the BBC showing how incentives can lead to odd choices in managing queues (Royal Cornwall Hospital patients ‘jumping waiting list’, Aug 27).
Up to 670 people had been waiting for more than the government waiting time target, of 18 weeks, the hospital said.
It explained it was treating the newer referrals first to avoid penalties for further target breaches. …
Orthopaedic Surgeon, Andrew Lee, told BBC News: “We’re doing our very best not to make any more breaches.
“We’re therefore treating the people under 18 weeks as well as dealing with the breached patients as fast as we can.”
While this sounds, frankly, kind of awful, lots of incentives systems for managing queues would lead to this behavior. Indeed, anything that measures outliers relative to a target may lead to this. For example, call centers are often evaluated on some sort of percentile wait (e.g., 80% of calls are answered in 20 seconds are less). Such a scheme has the obvious appeal of assuring a large fraction of customers receive good service. However, it also means that a customer that has waited 21 seconds already counts as a bad outcome and might as well as take one for the team if another customer has been waiting for 17 seconds when an agent becomes available.
Clearly, if you are the patient left waiting for surgery or the caller cooling his heels on hold, this will seem terribly unfair. On the other hand, the people making the decisions are just responding to the incentives presented to them and (at least at a call center) the front line service providers may not have an active role in choosing who goes next.
Addendum: If you want a specific citation on how delay percentile contracts can lead to unfortunate outcomes, check out this article by our colleagues Tava Olsen and Joe Milner:
Service-Level Agreements in Call Centers: Perils and Prescriptions
MANAGEMENT SCIENCE 2008 54:238-252
As Tava points out in an email, evaluating them on the average wait past the target would largely take care of the problem.