Workshop

Economics often defines and measures discrimination as disparities stemming from direct effects of group identity. We develop new tools to model and measure systemic discrimination, defined as disparities stemming from differences in non-group characteristics. Systemic discrimination can arise from differences in signaling technologies and opportunities for skill development. We propose a measure based on a decomposition of total discrimination into direct and systemic components. The measure is illustrated in a series of hiring experiments and a novel Iterated Audit experimental paradigm with real hiring managers. Results highlight how direct discrimination in one domain can drive systemic discrimination in other domains.

The interdependent values model [Milgrom Weber '82] captures when one buyer's private information about an item impacts how much another buyer is willing to pay for it, and that without this information, a buyer may not know their own value.  While this model is more realistic than those which assume each buyer's value is fully known and independent from other buyers, it still assumes that the way in which each buyer aggregates the information (signals) about the item into a numerical value (their valuation function) is public information.

I will present recent work in the model where both a buyer's signal and valuation function are considered to be private information. In this setting, satisfying incentive-compatibility is not only challenging, but extremely non-trivial to do so while obtaining any sort of welfare-approximation.  I will discuss how the Submodularity-over-Signals (SOS) condition on valuations can be leveraged to obtain a constant-factor approximation to the optimal welfare.

Based on joint work with Alon Eden, Michal Feldman, Simon Mauras, and Divyarthi Mohan.

Multi-item mechanisms can be very complex offering many different bundles to the buyer that could even be randomized. Such complexity is thought to be necessary as the revenue gaps between randomized and deterministic mechanisms, or deterministic and simple mechanisms are huge even for simple classes of valuations. We challenge this conventional belief by showing that these large gaps can only happen in situations where buyers' actions are severely restricted. These are situations where the mechanism sells a bundle of items at a higher price than the sum of the prices of the constituent items and buyers wanting to purchase such a bundle must pay this premium as they are not allowed to purchase the constituents separately.

We accordingly propose a new class of mechanisms that we call buy-many mechanisms wherein the buyer is allowed to interact with the mechanism multiple times and purchase as many (randomized) bundles as he pleases. Most real-world mechanisms are buy-many. We show that optimal buy-many mechanisms satisfy many nice properties that general mechanisms do not: they are approximable by simple mechanisms; their revenue is a smooth function of the buyer's value distribution; and they have bounded description complexity.

This talk is based on joint work with Rojin Rezvan, Yifeng Teng, and Christos Tzamos.

We consider auctions where buyers can acquire costly information about their valuations and those of others, and investigate how competition between buyers shapes their learning incentives. In equilibrium, buyers find it cost-efficient to acquire some information about their competitors so as to only learn their valuations when they have a fair chance of winning. We show that such learning incentives make competition between buyers less effective: losing buyers often fail to learn their valuations precisely and, as a result, compete less aggressively for the good. This depresses revenue, which remains bounded away from what the standard model with exogenous information predicts, even when information costs are negligible. Finally, we examine the implications for auction design. First, setting an optimal reserve price is more valuable than attracting an extra buyer, which contrasts with the seminal result of Bulow and Klemperer (1996). Second, the seller can incentivize buyers to learn their valuations, hence restoring effective competition, by maintaining uncertainty over the set of auction participants.