Equilibrium methods for Resource Allocations and Dynamic Pricing

Project Information

EQUIPRICE

Grant agreement ID: 866274

DOI

10.3030/866274

Project closed

EC signature date 13 February 2020

Start date 1 June 2020

End date 31 May 2026

Funded under

• EXCELLENT SCIENCE - European Research Council (ERC)

Total cost

€ 1 809 595,00

EU contribution

€ 1 809 595,00

Coordinated by

FONDATION NATIONALE DES SCIENCES POLITIQUES
France

Periodic Reporting for period 3 - EQUIPRICE (Equilibrium methods for Resource Allocations and Dynamic Pricing)

Reporting period: 2023-06-01 to 2024-11-30

Summary of the context and overall objectives of the project

The EQUIPRICE project seeks to build an innovative economic toolbox (ranging from modelling, computation, inference, and empirical applications) for the study of equilibrium models with gross substitutes, with applications to models of matching with or without transfers, trade flows on networks, multinomial choice models, as well as hedonic and dynamic pricing models.

These topics are at the core of the algorithms that are routinely used by platforms. Given that the regulatory pressure on platforms is ever increasing with the various regulations currently prepared by the European Union (Digital Markets Act; Digital Services Act; IA Act; Data Act / GDPR), it is ever as crucial to increase the technical knowledge on the topics and it’s dissemination to an audience as wide as possible.

By developing open source algorithms, and widely disseminating the ideas and methods, in particular through the master classes and scientific articles, the Equiprice project aims at responding to these challenges.

Work performed from the beginning of the project to the end of the period covered by the report and main results achieved so far

The ‘equiprice’ project started in June 2020, and has already achieved a number of successes, in spite of a slow down due to Covid, which has impacted in particular in-person synergetic activities.

Since the start of the project, several theoretical successes have been obtained:

• A unifying notion of gross substitutes (unified gross substitutes) that applies to functions and correspondences alike

• The reformulation of many problems in matching and in network flows in the language of generalized linear models, allowing to leverage the scalability of modern machine learning libraries

• A proof of existence of matching function equilibria under full assignment with general normalization

• The SISTA algorithm, which allows to do model selection in inverse optimal transport

• A complete toolbox of open-access software has been made available as part of the math+econ+code masterclasses.

A dynamic team was assembled, of post-doctoral, pre-doctoral and doctoral researchers, and several research assistants, with various backgrounds in economics, mathematics and computer science.

All synergetic activities were moved online or hybrid during covid and remained that way. This includes five online masterclasses (in January and June each year since June 2020); and one hybrid workshop in October 2021. The lunch seminar series were started in fall 2020, and were followed by the math-econ-code monthly seminar.

The number of papers accepted or published in academic journals (both in economics and in mathematics) speak for the promising nature of the research agenda, and its initial success. Since the start of the project, the PI has received a number of accolades such as being elected a Fellow of the Econometric Society in 2020, being invited semi-plenary speaker at the world Congress of the Econometric Society in 2020, and delivering the Sir John Hicks lecture at the SAET conference in 2020.

Progress beyond the state of the art and expected potential impact (including the socio-economic impact and the wider societal implications of the project so far)

Consistent with the methodology sketched in the scientific proposal, the project is developing new mathematical results for the use of readily available toolboxes, used in particular in the field of machine learning, for the problems under investigation. When existing toolboxes are not available, the projects seeks to build them, and this is how M-correspondences, the equilibrium flow problem, the theory of matrons have been introduced. The intersection between Data science, machine learning, and economics is surprisingly underdeveloped and the current project bridges the gap between these disciplines.

In addition to “classical“ scientific activities such as scientific publication, open-source coding, masterclasses, conferences, similar presentations, etc, the PI is currently working an original and novel dissemination project, which consist of a coding competition. As a follow-up to the math+econ+code masterclasses, a platform is currently being put together what participants will code dynamic matching and pricing algorithms which will solve the problem of a car passenger service. This will allow participants to put in practice to techniques and knowledge that they have learned during the masterclasses. The PI is currently working with a team of programmers, and with research assistance on the platform’s core engine and on the microsimulation of the drivers and passengers behaviors.

erc-equiprice.jpg