• Date: Friday, November 22, 2019.
• Location: Kellogg Global Hub 5101 (map), Northwestern U, Evanston, IL 60208.
• Transit: Noyes St. Purple Line (map).
• Parking: Validation for North Campus Parking Garage (map) available at workshop.
• Registration: None required.  Please bring your own name badge from past conference.

Schedule

• 8:30-9:00: Continental Breakfast
• 9:00-9:05: Opening Remarks
• 9:05-9:45: Shaddin Dughmi (USC):
  Algorithmic Bayesian Persuasion
• 9:45-9:50: Shaddin Dughmi Q/A
• 9:50-10:30: Laurent Mathevet (NYU):
  Optimal Information Hierarchies
• 10:30-10:35: Laurent Mathevet Q/A
• 10:35-11:05: Coffee Break
• 11:05-11:45: Stephen Morris (MIT):
  Information Design with Adversarial Equilibrium Selection
• 11:45-11:50: Stephen Morris Q/A
• 11:50-12:30: Rachel Cummings (Georgia Tech):
  Algorithmic Price Discrimination
• 12:30-12:35: Rachel Cummings Q/A
• 12:35-1:30: Lunch

Kellogg Global Hub 4130:

• 2:00-4:00: Student meetings with speakers

Titles and Abstracts

Speaker: Shaddin Dughmi
Title: Algorithmic Bayesian Persuasion
Abstract: 
Persuasion, defined as the act of exploiting an informational advantage in order to effect the decisions of others, is ubiquitous. Indeed, persuasive communication has been estimated to account for almost a third of all economic activity in the US. This paper examines persuasion through a computational lens, focusing on what is perhaps the most basic and fundamental model in this space: the celebrated Bayesian persuasion model of Kamenica and Gentzkow. Here there are two players, a sender and a receiver. The receiver must take one of a number of actions with a-priori unknown payoff, and the sender has access to additional information regarding the payoffs. The sender can commit to revealing a noisy signal regarding the realization of the payoffs of various actions, and would like to do so as to maximize her own payoff assuming a perfectly rational receiver.
We examine the sender’s optimization task in three of the most natural input models for this problem, and essentially pin down its computational complexity in each. Examining persuasion through this computational lens leads to novel structural insights. Notably, en-route to some of our results we draw a connection between the space of persuasion schemes and Border’s characterization of the space of reduced-forms for single-item auctions.

Abstract: The way in which information is organized within a group matters for how the transmission of that information can be implemented. We introduce the concept of information hierarchies as organizational forms of distributed knowledge: In an information hierarchy, agents are partitioned into groups such that all members of a group are equally informed, and more informed (in a strong sense) than all members of lower groups. Information hierarchies can be implemented by either select public meetings or by delegated information transmission. We show that, in a general class of problems involving network interactions with complementarities, there always is an information hierarchy among the optimal information schemes. Therefore, although the environment might display no hierarchical structuring of agents (their influences and dependencies on the state and each other might be only partially ordered), optimal design treats them hierarchically with totally ordered information. Remarkably, information hierarchies are optimal in standard design (where the best equilibrium is selected) as well as in robust design (where the worst equilibrium is selected).

Joint work with Ina Taneva.

Speaker: Stephen Morris
Title: Information Design with Adversarial Equilibrium Selection
Abstract: When an information designer gets to pick both the information structure and the equilibrium played, the set of implementable outcomes is characterized by a linear program. We provide a linear algebraic characterization of implementable outcomes under adversarial equilibrium in binary action supermodular games. The result exploits an existing literature on the robustness of equilibria in games.

Speaker: Rachel Cummings
Title: Algorithmic Price Discrimination
Abstract: We consider a generalization of the third degree price discrimination problem studied in Bergemann et al. 2015, where an intermediary between the buyer and the seller can design market segments to maximize any linear combination of consumer surplus and seller revenue. Unlike in Bergemann et al. 2015, we assume that the intermediary only has partial information about the buyer’s value. We consider three different models of information, with increasing order of difficulty. In the first model, we assume that the intermediary’s information allows him to construct a probability distribution of the buyer’s value. Next we consider the sample complexity model, where we assume that the intermediary only sees samples from this distribution. Finally, we consider a bandit online learning model, where the intermediary can only observe past purchasing decisions of the buyer, rather than her exact value. For each of these models, we present algorithms to compute optimal or near optimal market segmentation.

Joint work with Nikhil Devanur, Zhiyi Huang, and Xiangning Wang.