Implementation with ex post hidden actions

We study implementation in settings where agents take strategic actions that influence preferences over mechanism outcomes and yet are hidden from the mechanism designer. We show that such settings can arise in entry auctions for markets, and that the Vickery-Clarke-Groves mechanism is not necessarily truthful. In this paper we first formalize so-called ex post hidden actions, we then characterize social choice functions that can be implemented in a way that is robust with respect to ex post hidden actions, and finally we propose a mechanism to do so. The model allows agents to have multi-dimensional types and to have quasi-linear utilities in money. We showcase these results by identifying social choice functions that can and cannot be implemented in entry auctions for Cournot competition models.

Position Ranking and Auctions for Online Marketplaces

Online e-commerce platforms, such as Amazon and Taobao, connect thousands of sellers and consumers every day. In this work, we study how such platforms should rank products displayed to consumers and utilize the top and most salient slots. We present a model that considers consumers’ search costs and the externalities sellers impose on each other. This model allows us to study a multiobjective optimization, whose objective includes consumer and seller surplus as well as the sales revenue, and derive the optimal ranking decision. In addition, we propose a surplus-ordered ranking mechanism for selling some of the top slots. This mechanism is motivated in part by Amazon’s sponsored search program. We show that the Vickrey–Clarke–Groves mechanism would not be applicable to our setting and propose a new mechanism. This mechanism is near optimal, performing significantly better than those that do not incentivize sellers to reveal their private information regarding each consumer purchase, such as their profit. Moreover, we generalize our model to settings in which platforms can provide partial information about the products and facilitate the consumer search and show the robustness of our findings. This paper was accepted by David Simchi-Levi, operations management.

Agent-based mechanism design – investigating bounded rationality concepts in a budgeting context

Purpose This study aims to demonstrate how agent-based simulation (ABS) may provide a computational testbed for mechanism design using concepts of bounded rationality (BR). ABS can be used to systematically derive and formalize different models of BR. This allows us to identify the cognitive preconditions for behavior intended by the mechanism and thereby to derive implications for the design of mechanisms. Design/methodology/approach Based on an analysis of the requirements of the decision context, the authors describe a systematic way of incorporating different BR concepts into an agent learning model. The approach is illustrated by analyzing an incentive scheme suggested for truthful reporting in budgeting contexts, which is an adapted Groves mechanism scheme. Findings The study describes systematic ways in which to derive BR agents for research questions where behavioral aspects might matter. The authors show that BR concepts may lead to other outcomes than the intended truth-inducing effect. A modification of the mechanism to more distinguishable levels of payments improves the results in terms of the intended effect. Research limitations/implications The presented BR concepts as simulated by agent models cannot model human behavior in its full complexity. The simplification of complex human behavior is a useful analytical construct for the controlled analysis of a few aspects and an understanding of the potential consequences of those aspects of human behavior for mechanism design. Originality/value The paper specifies the idea of a computational testbed for mechanism design based on BR concepts. Beyond this, a systematic and stepwise approach is shown to formalize bounded rational behavior by agents based on a requirements analysis, including benchmark models for the comparison and evaluation of BR concepts.

Conditional Sojourn times and the volatility of payment schemes for bandwidth sharing in packet networks

Predictability of revenue and costs to both operators and users is critical for payment schemes. We study the issue of the design of payment schemes in networks with bandwidth sharing. The model we consider is a processor sharing system that is accessed by various classes of users with different processing requirements or file sizes. The users are charged according to a Vickrey–Clarke–Groves mechanism because of its efficiency and fairness when logarithmic utility functions are involved. Subject to a given mean revenue for the operator, we study whether it is preferable for a user to pay upon arrival, depending on the congestion level, or whether the user should opt to pay at the end. This leads to a study of the volatility of payment schemes and we show that opting for prepayment is preferable from a user point of view. The analysis yields new results on the asymptotic behavior of conditional response times for processor sharing systems and connections to associated orthogonal polynomials.