Chasing Tails: How Do People Respond to Wait Time Distributions?

We use a series of pre-registered, incentive-compatible online experiments to investigate how people evaluate and choose among different waiting time distributions. Our main findings are threefold. First, consistent with prior literature, people show an aversion to both longer expected waits and higher variance. Second, and more surprisingly, moment-based utility models fail to capture preferences when distributions have thick-right tails: indeed, decision-makers strongly prefer distributions with long-right tails (where probability mass is more evenly distributed over a larger support set) relative to tails that exhibit a spike near the maximum possible value, even when controlling for mean, variance, and higher moments. Conditional Value at Risk (CVaR) utility models commonly used in portfolio theory predict these choices well. Third, when given a choice, decision-makers overwhelmingly seek information about right-tail outcomes. These results have practical implications for service operations: (1) service designs that create a spike in long waiting times (such as priority or dedicated queue designs) may be particularly aversive; (2) when informativeness is the goal, providers should prioritize sharing right-tail probabilities or percentiles; and (3) to increase service uptake, providers can strategically disclose (or withhold) distributional information depending on right-tail shape.

💡 Research Summary

This paper investigates how people evaluate and choose among waiting‑time distributions using a series of pre‑registered, incentive‑compatible online experiments. Across three studies involving 1,114 participants, the authors examine the roles of the first two moments (mean and variance), higher moments (skewness, kurtosis), and, crucially, the shape of the right tail of the distribution.

Study 1 establishes the baseline: participants dislike longer expected waits and higher variance, confirming prior work. Using Bernoulli, Uniform, and Exponential distributions, the authors find that a one‑minute increase in mean reduces willingness to wait by roughly $0.10, while a one‑minute increase in standard deviation reduces it by about $0.05. Moreover, even when mean and variance are held constant, choices differ across distributional forms, hinting that simple mean‑variance models are insufficient.

Study 2 delves deeper. In 2A, variance, skewness, and kurtosis are varied one at a time while keeping other moments fixed. Results show that higher variance or kurtosis does not consistently make a distribution less attractive, contradicting pure moment‑based predictions. In 2B, the authors contrast “thick‑right” tails (probability mass concentrated near the maximum possible wait) with “long‑right” tails (mass spread over many high‑wait outcomes). Participants robustly prefer long‑right tails, indicating a tolerance for rare but extreme delays versus a higher probability of moderately long delays. Study 2C tests whether this “thick‑right‑tail aversion” persists when information is incomplete. The ranking remains: known long‑right tail > unknown > known thick‑right tail.

Study 3 asks which part of a distribution people want to learn about. Using a rank‑order elicitation, participants overwhelmingly prioritize information about the right tail (the worst‑case waiting times). This aligns with the earlier finding that right‑tail shape drives choices.

To model behavior, the authors compare a conventional linear utility function based on mean and variance with a Conditional Value at Risk (CVaR) utility that explicitly penalizes adverse right‑tail outcomes. The CVaR model fits the data far better, especially capturing the aversion to thick‑right tails.

Practical implications follow. Service designs that merely equalize mean wait times but create a spike in long waits (e.g., priority queues) may deter customers. Providers should disclose right‑tail information—such as high‑percentile wait times—when aiming for transparency, but they should be cautious about revealing thick‑right‑tail risk, which can reduce demand. Incorporating CVaR‑type utilities into queueing models can change optimal design recommendations (e.g., pooled versus dedicated queues).

Overall, the paper contributes a novel insight: in time‑based decisions, people’s risk attitudes resemble those observed in financial contexts, with a pronounced sensitivity to the shape of the right tail. This bridges behavioral economics, operations management, and risk theory, and suggests that CVaR‑based utility functions are a promising tool for designing and communicating service systems.