Posted Pricing and Competition in Large Markets

José Correa, Vasilis Livanos, Dana Pizarro, Victor VerdugoarXiv 2025 · DOI · arXiv

Introduction

In mechanism design, a fundamental tension exists between optimal mechanisms—complex auctions designed to maximize welfare—and simple mechanisms, such as posting a single fixed price. While complex auctions are theoretically superior, modern online marketplaces like eBay and Amazon favor simplicity for its lower overhead and better user experience. Traditional "prophet inequalities" often provide pessimistic worst-case bounds for simple mechanisms (e.g., a fixed price capturing $\approx 0.632$ of optimal welfare). This paper investigates whether the massive scale of modern markets provides a "large market benefit" that improves these guarantees.

Theoretical Framework: Extreme Value Theory

To analyze behavior in large markets, the researchers employ Extreme Value Theory (EVT). EVT is typically used to model rare events by examining the distribution of the maximum of a large number of independent random variables. In this context, it models the maximum valuation among $n$ bidders as $n \to \infty$.

The authors categorize buyer valuations into three distinct families based on their tail behavior:

  1. Fréchet: Heavy-tailed distributions, such as the Pareto distribution.
  2. Gumbel: Distributions with exponential-like tails, including Normal and Exponential distributions.
  3. Reversed Weibull: Distributions with a finite upper bound, such as the Uniform distribution.
Classification of extreme value distributions and their limit properties.

By applying this framework, the authors derive tighter approximation guarantees for both single-item and multi-unit ($k$-unit) sales.

Key Results and Competition Complexity

The study yields several significant improvements over classical bounds. Most notably, in the challenging case of heavy-tailed (Fréchet) distributions, the welfare guarantee for a single item increases from the classic 0.632 to 0.712 in large markets. For Gumbel and Reversed Weibull distributions, fixed pricing achieves a perfect 1.0 welfare guarantee as the market grows.

The paper also addresses "competition complexity"—the question of how many additional bidders are required for a simple mechanism to match the performance of an optimal auction. Previous research suggested this factor could be infinite. However, this work proves that in large markets, the competition complexity is a small constant. For Gumbel distributions, this factor is approximately 1.781.

Summary of welfare guarantees and competition complexity across distribution types.

Empirical Validation

The theoretical findings were validated using real-world data from 97 eBay Cartier watch auctions. By treating the watches as substitutes and focusing on unique bidder valuations, the authors created a dataset of 509 valuations. Fitting this data to a Fréchet distribution ($\alpha \approx 2.24$), they found that a fixed-price policy achieved 73.3% of the optimal welfare, exceeding their theoretical lower bound of 71.2%.

Conclusion and Future Directions

This research demonstrates that market scale significantly mitigates the inefficiencies of simple pricing mechanisms. While the advantage of scale is most pronounced for single-item sales and diminishes as the number of units ($k$) increases, the results provide a strong theoretical justification for the use of posted prices in high-volume environments.

Future work may address the limitations of the current model, such as the assumption of Independent and Identically Distributed (i.i.d.) valuations, and explore dynamic market settings where bidders arrive and depart over time.

Summary updated

BibTeX
@misc{correa2025posted,
  author = {José Correa and Vasilis Livanos and Dana Pizarro and Victor Verdugo},
  title = {{Posted Pricing and Competition in Large Markets}},
  year = {2025},
  eprint = {2505.18061},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2505.18061},
}
Figure