Science1 publisher2 min readPublished
Overconfidence about service times lifts the best price for queues customers cannot see
University of Florida researchers find that customers who underrate service-time variability raise the revenue-maximizing price of queues they cannot see.
The Scientist · Science desk
What happened
- University of Florida researchers report in Manufacturing & Service Operations Management that customers often underestimate how much service times vary, a bias called overprecision.
- When customers cannot see how long the line is, the study finds the optimal price is higher than traditional queueing models suggest.
- Sharing queue length improves revenue in very busy or very slow periods, while at moderate traffic the benefit to the business is less clear.
Compiled by The ScientistSomething wrong?How this is made
Why it matters
- decision An operator who prices an unseen queue from a classical model is, on this analysis, charging below the revenue-maximizing price, so the pricing model itself needs revisiting.
- exposure In hidden queues the firm's revenue interest and the customer's interest diverge in every case studied: customers come out behind and the revenue price exceeds the welfare price.
- decision Posting line length becomes a timing choice, since at high congestion the disclosure usually assumed to help customers can lower their benefit.
- constraint With no reported magnitudes, the result tells an operator which way to move a price but cannot yet be turned into a figure on a price list.
Every result in the account is scored against one benchmark: the price that "traditional models" or "classical theory" would recommend [3][5]. Those models are the control. The paper's title, "Managing Service Systems with Overconfident Customers," says what is being varied: the customer's confidence about how predictable service is [11].
"Our study shows how the common cognitive bias of overconfidence, in which decision-makers tend to make overly optimistic forecasts about uncertain events, can shape individual decisions and entire service systems," said Na Zhang, a UF Warrington graduate who is now an assistant professor at Wichita State University [8][12].
The results do not all point the same way. Only the hidden-queue result has a single sign. There, the optimal price comes out above the classical one. Where the line is visible, it can land on either side [14]. That asymmetry decides how much a manager can take from the paper. A firm running a queue customers cannot see gets a direction. A firm with a visible line gets a warning to check its congestion first [5].
Disclosure has a similar split. For moderately busy periods, Anand Paul, the E.R. Bell Professor at UF Warrington, offered a judgement call [12]. "In these cases, companies may need to be more strategic about when they share wait-time and queue-length information," Paul said [9].
Zhang pointed to one piece of outside evidence. "For example, overconfidence provides a plausible behavioral explanation for the empirical findings of patients' expected waiting times being consistently shorter than their actual waiting times in health care service systems," Zhang said [10]. The word "plausible" is chosen with care. The hospital finding is a gap between what patients expected and what they got. Overprecision fits that gap without being shown to cause it. The published account does not describe an experiment, a sample of customers, or a measured size for either the bias or the price gap it produces [13].
I think the hidden-queue result is the part an operator can act on, because it points one way. On this analysis, a price for an unseen line set from a classical model sits below the revenue-maximizing price [3]. That holds only if the customers in that queue are overprecise, and only to a degree someone has measured.
What to watch
- Whether the full paper reports how large the price gap is under realistic levels of overprecision, which would let the hidden-queue result be sized.
- Field or lab measurements of how far customers' estimates of service-time variability fall short, for example in the health care waits Zhang cites.
- Evidence from operators of queues customers cannot see on whether prices or wait-time disclosure behave as the model predicts.