Four Week Retreat Attracts World's Leading Complexity Economists

• Month-long Retreat at Santa Fe Institute attracts leading thinkers in Complexity Economics;
• Unique format allows deeper thinking on expanding discipline;
• Workshop provided a rare opportunity to make progress on the deepest questions in Economics, Modelling and Complex Systems.

A four-week Complexity Economics Retreat held at the Santa Fe Institute over the summer provided a unique opportunity to bring together the leading scientists tackling economic problems from Europe, America and Asia.

Organised by Christian Diem, Jose Moran and J. Doyne Farmer from the Complexity Economics Programme at INET Oxford, the workshop ‘Economic Agent-Based Models: Crossing Over the Tipping Point’ ran throughout the month of August, beginning on Monday 4th and finishing on Friday 29th.

The event attracted some of the big names in economics and complexity science, including Michael Woodford (Columbia), James Evans (U Chicago), Vasco Carvalho (Cambridge), Chris Caroll (Johns Hopkins), John Genakoplos (Yale), Rob Axtell (George Mason) and up and coming researchers like Marta Grzeskiewicz (Cambridge), Takahiro Yabe (NYU), Sara Constantino (Stanford), Marco Pangallo (CENTAI) and Anton Pichler (WU Vienna).

A unique format of just one talk a day, with the remainder left to focused discussions and work, provided a rare opportunity to make progress on the deepest questions in economics, modelling and complex systems.  Joint breakfast, lunches and dinners made for even more room to exchange ideas and discuss hot topics in research and global affairs, broadening the horizon of the meeting. 

Talks were given by Yukiko Saito (Waseda), Suresh Naidu (Columbia), Vasco Carvalho (Cambridge), Marco Pangallo (CENTAI), Aldo Glielmo (Bank of Italy), Arnau Quera-Bofarull (Macrocosm), Luca Mungo (Macrocosm), Jordan Kemp (Oxford), Takahiro Yabe (NYU), Michael Woodford (Columbia), Doyne Farmer (Oxford), Hiroyasu Inoue (Hyogo), Sara Constantino (Stanford), John Geanakoplos (Yale), Marta Grzeskiewicz (Cambridge), James Evans (U. of Chicago), François Geerolf (OFCE) and Rob Axtell (George Mason).

The workshopprompted deep thinking on a series of questions:

• How reinforcement learning can improve ABMs and how the resulting optimizing behaviour can break common ground with older modelling traditions;
• Whether rational expectations is backing economic models too strongly into a corner and thereby yielding inflexible models and hitting computational limits;
• The failure of many key theoretical models to address stylized facts, while at the same time forming the core of curricula that are taught in economics and finance around the world;
• The problem of modelling inflation without modelling money in the economic models used for forecasting inflation;
• How differentiable ABMs are pushing us close to one of the tipping points in agent based modelling: the ability to properly fit the parameters of complicated ABMs to data.

Reflecting on the workshop, organiser Christian Diem said that meetings featured vibrant discussion on solving problems in economics, the newest methodologies, including how advances in AI can be utilized.

"There was plenty of focus on identifying the big open problems in economics and economic modelling and how agent based models can contribute to solving them.

"The open format considered questions such as: 'how do economic models need to change?', 'What can agent based models bring to the table' and what will lead to them being more widely adopted? 

"After four weeks of deep thought and discussion, participants have gone away with renewed ideas and vigour”.

Director of the Complexity Economics Programme at INET Oxford J. Doyne Farmer said “Working in a relaxed and beautiful setting while exchanging ideas with really knowledgeable, smart people is a great way to accomplish big things – and it is really fun."