MET Seminar - Andrew Ellis (LSE)
| Dates: | 10 May 2023 |
|---|
| Times: | 17:00 - 18:00 |
|---|
| What is it: | Seminar |
|---|
| Organiser: | School of Social Sciences |
|---|
| Who is it for: | University staff, Current University students |
|---|
|
Title: Equilibrium Effects of Machine Learning (joint with Michele Piccione and Shengxing Zhang)
Abstract: We introduce a tractable framework for studying the equilibrium effects of machine learning. Agents process information using a Chow-Liu (1968) tree, a widely-used machine learning procedure that admits a closed-form solution. We apply the model to several settings, including auctions, bilateral monopoly, and an asset market with dispersed information based on Hellwig (1980). In the market setting, ex ante identical traders' equilibrium beliefs may be based on different trees. This leads to asymmetric reactions to the same information. With low noise, most agents learn only from the (endogenous) price. This bounds the equilibrium prices away from informational efficiency. With high-dimensional public information, agents update their beliefs in response to price changes, even though prices are a garbling of that information.
MET full schedule: http://daviddelacretaz.net/seminars/
Travel and Contact Information
Find event
Room G.020
Arthur Lewis Building
Manchester
