Prize Lecture: Paul R. Milgrom, The Sveriges Riksbank Prize in Economic Sciences 2020

Paul Milgrom's Journey Paul Milgrom, a renowned economist, started his career as an actuary before returning to university to pursue higher education. His doctoral dissertation on auction theory revolutionized the field and set the foundation for his extensive contributions in economics.

"Auction Theory Evolving" "Auction Theory Evolving" is Paul Milgrom's talk where he discusses new aspects and questions related to auction theory. He highlights that most of his work involves creating mathematical models and proving their properties through theorems. Additionally, he mentions engaging in experiments or simulations to understand how different approaches perform. The applications of auction theory have significant financial implications.

The speaker expresses gratitude towards influential figures in their career, including Bob Wilson, David Kreps, and J. Michael Harrison. They also mention the inspiring young faculty at Northwestern University.

Understanding Auctions Early in my career, I studied auctions and integrated the insights of William Vickery and Bob Wilson. Vickery believed that bidders had an idea of what something was worth to them before bidding. He introduced game theory methods and Nash equilibrium to study auction strategies. On the other hand, Wilson focused on auctions for mineral rights where bidders have different information about the value, leading to a problem called 'the winner's curse.' The winner's curse occurs when winning bids are more likely due to overestimating value rather than underestimating it.

'The Winner's Curse' Problem 'The winner's curse' is a concern in auctions where winning bidders tend to overestimate the value of what they bid for. This leads them to win more often but potentially pay too much because their estimates were higher than necessary. Different auction designs can exacerbate or mitigate this problem, raising questions about how auction rules affect revenue generation.

Auction Design Impact In our research model developed by Bob Weber and me, we considered both private and common values at play simultaneously in an auction setting. We found that using an English ascending auction resulted in the highest expected price based on Nash equilibrium analysis while a Vickery second-price auction came next followed by a standard sealed tender (first-price) option with lowest expected price outcome.The sharing of relevant information by sellers also increased the expected price compared tounshared information.This led us further into exploring whether differentauction rules impact efficiency.

The No Trade Theorem In this chapter, the concept of the no trade theorem is introduced. It states that in a common value setting where buyers and sellers have different information but no fundamental reason for having different values, there will be no trade regardless of the rules used. This was surprising at the time because it challenged the idea that trading based on differences in information could lead to effective trade.

Market Makers and Price Formation "Market makers" are discussed in this chapter as individuals who set prices for securities based on competition with other market makers. They quote bid prices (for buying) and ask prices (for selling), which are conditional expectations taking into account potential losses from traders with superior information. The transactions price reflects these quotes, resulting in movement of prices reflecting available information. Prices tend to fluctuate around a certain level without systematic drift.

In the early days of internet advertising, ads were similar to those in newspapers. However, with the rise of performance advertisers, targeting specific individuals became important. This posed a problem for traditional display advertisers as they risked losing high-income customers. To address this issue, we studied and designed auctions that protected against adverse selection and ensured fairness. Our auction design considered both common value (attributes everyone cares about) and match value (suitability between advertiser and customer). We created a one-parameter family of auctions where performance advertisers won based on bid ratios exceeding a parameter alpha.

Radio Spectrum and Invention In 1993, Bob Wilson and I created an invention for the Federal Communications Commission (FCC) to solve a problem faced by Pacific Bell. They needed to acquire licenses covering northern and southern California but didn't know which ones were cheapest. We proposed a simultaneous multiple round auction with bidding rules that eliminated bid sniping. This design was adopted by the FCC and has been used worldwide, generating billions of dollars in revenue.

"Matching with Contracts" Model We discovered intriguing similarities between Galen Shapley's mechanism for matching without money involved and Vickery's price-based auction mechanism. Both mechanisms have dominant strategies for bidders to make truthful offers, resulting in core allocations if offers are substitutes. To explore this connection further, John Hatfield introduced our model of matching with contracts where offers are accompanied by terms or conditions. This model encompasses various market designs including auctions.

Radio spectrum allocation is a complex process that involves bidding for licenses to implement business plans. Bidders often need specific combinations of licenses to make their plans viable, and the risk they face is winning only part of what they need. To address this, bidders can bid for combinations of licenses in spectrum auctions. Peter Crampton, Larry Osibel, Bob Day, and I developed a method to provide guidance on prices before the final round of bidding. We also identified pricing rules that minimize incentives for dishonest bidding while ensuring competitive revenue.

Package Bidding: A Challenging Problem Package bidding is a complex problem due to the large number of packages available for bidders. For example, in an auction with 100 licenses, there are 2^100 different combinations possible. This makes it impractical for bidders to consider all options. However, researchers have found that if bidders bid up to their limits on relevant packages, efficient and core allocations can be achieved.

"Can We Have Good Outcomes?" [John Cagle](image) and [Yen Chen Lean](image), along with the speaker's guidance as a professor at Ohio State University, explored strategies for achieving favorable outcomes in package bidding scenarios. They discovered that by encouraging bidders to bid up to their limits on relevant packages and using simulations with auto-bidders, one could predict whether desirable allocations would emerge.

The recent incentive auction, known as the broadcast incentive auction, aimed to repack TV broadcasters into a smaller set of channels and create mobile broadband licenses from what remains.

The challenge of assigning TV broadcast channels to broadcasters without interference is an NP-hard problem. Our team at Auctionomics, along with colleagues from Stanford Economics Department and the University of British Columbia, developed new algorithms and auction designs to address this problem. These designs accommodate computational limits while ensuring properties like strategy-proofness, budget compliance, and privacy preservation.

Working Papers and Co-Authors The speaker discusses their current working papers with various co-authors, including Muhammad Akbar, Sheng Wu Lee, Scott Commenters, Martin Bickler, Gregor Schwartz, Josh Mullner, Neil Newman and Mitch Watt. These papers focus on investment incentives in auction problems and related topics in game theory.

New Generation of Students 'm proud to introduce a new group of students who will make a significant impact on the future of auction and market design. This diverse group includes talented individuals like Got In Pearl Evan Ziang Lucas Kevin. Their contributions are highly anticipated.