On December 6, 2021, the Luohan Academy convened its 8th Frontier Dialogue - Exploring Complexity and Complexity Economics. The event was moderated by Katherine Collins, Head of Sustainable Investing at Putnam Investments and Chair of the Santa Fe Institute’s Board of Trustees.
Advances in digital technologies have made the world more connected. Society’s actors are linked and affected by each other, making the world an increasingly complex system. Complexity studies probe how elements interacting in a system create patterns, and how these patterns, in turn, cause the elements to change or adapt in response. Whether it is cars in traffic reacting to adjacent cars, or cells in an immune system reacting to other cells and viruses, complexity asks how individual elements react to the current pattern they mutually create, and what patterns result. Panelists in this symposium discussed both theory and real-world applications.
Simon A. Levin, Princeton University Professor of Ecology and Evolutionary Biology, launched the discussion with several crucial points. First, many challenges in both ecological and socioeconomic systems are of complex adaptive systems (CAS). Second, these issues cross scales – macroscopic patterns come from microscopic interactions of heterogenous individual agents. Third, outcomes are often unpredictable with traditional linear tools because these systems have feedbacks, path dependence, hysteresis, etc. His own paper “Ecology for Bankers” (2008) used complex networks, inspired by ecological models, to show financial markets on the verge of collapse. While the paper was unfortunately proven prescient, it showed the capabilities of complexity models for seeing real world problems. Levin noted two important areas of work for complexity sciences. The first is to identify “critical transitions” or early warning indicators of systemic issues. The second is the research begun by Elinor Ostrom, the first female Nobel Laureate in economics, to study how pro-sociality behavior in small local groups can contribute to preserving the Commons globally. This work supports a polycentric governance model, in which multiple governing bodies interact to make and enforce rules in complex social-ecological systems, even without centralized global coordination. This is often a good way to achieve collect action in high uncertainty – key for environmental challenges such as climate change.
Santa Fe Institute External Faculty Member W. Brian Arthur followed by first explaining the history and impetus for his groundbreaking work in complexity economics, which began at the inaugural meeting of the SFI’s first research program – the Economy as an Evolving Complex System in 1987. This convened top thinkers in different disciplines, including Kenneth Arrow, Philip Anderson, Doyne Farmer, Tom Sargent, Larry Summers, John Holland. ‘Standard’ economics was trying to come up with models for problems that required well defined problems, which was often not possible. Holland, a computer scientist studying AI, suggested a more intuitive “agent based” approach where agents do not have perfect knowledge, but rather intuitions and rules that can learn and evolve. The application of computer simulation to these concepts allowed them to run models. Arthur related an early success simulating investors in a stock market. This model did not assume perfectly rational investors as in Robert Lucas’ standard model. While many results were in line with Lucas, additions were periods of bubbles and crashes, exactly like the real world. These early experiments showed the promise and much work has followed. Arthur sees complexity economics as a form of non-equilibrium economics that relaxes the overly strict assumptions of standard economics to improve on the latter.
University of Oxford mathematics professor J. Doyne Farmer described applications of complexity economics to addressing climate change, household savings, leverage in the financial system, economic impact of pandemic lockdown policies – all situations where the probabilities of events, or even the set of possible outcomes, are not known by individual agents. Farmer underscored the need for micro-level data collection to power the bottom-up models that he works with, which government and statistical agencies are not yet collecting. Finally, he touched on the persistence of mainstream resistance to complexity thinking, which challenges foundational assumptions and orthodoxy which have reigned over the past 150 years.
Yale economics professor John Geanakoplos served as discussant to review the presentations by Levin, Arthur, and Farmer. To Geanakoplos, complexity economics is drawing excitement and he cites two reasons for this. First, it is interdisciplinary and pulling in the brightest minds from various disciplines. Second, it arose by using modern computational methods, which has led to potentially powerful real-world results. Geanakoplos contrasts Farmer’s work accurately predicting Covid’s impact on employment, with the inability of many economists to predict the effect of Covid and stimulus on inflation, suggesting that complexity bottom-up models may have done better. Geanakoplos concludes by lamenting the loss of talented non-economists to the economic discipline. He calls for all economics, standard or complex, to be open and cross disciplines to absorb and germinate new ideas.
Opening the second session on operationalizing complexity, Stefan Thurner, a complex systems professor at the Medical University of Vienna, introduced real world research using complexity tools to quantify systemic risk in Austria’s financial system. Looking at data such as banks contract networks, books, and leverage ratios, Thurner was able to discover that even some very small banks can have outsized systemic importance. Actors who introduce more systemically risky transactions can eventually shoulder a higher price, perhaps by paying a systemic-risk tax. Based on his research, Thurner found that de-risking solutions to prevent cascading effects in a system can be accomplished without making it smaller or less efficient. These models could be applied globally as well to see which country defaults would be most significant from the perspective of global stability.
Asia Global Institute Distinguished Fellow and adviser to China’s banking regulator Andrew Sheng warned against the old reductionist ways of regulating specific products and within siloes, when modern innovations like platform economies and crypto-currencies are breaking down traditional barriers, supply chains, and networks. In a world where the markets are linked to the real economy, which is linked to foreign exchange and politics and national security, interdisciplinary complex thinking is required. Traditional remedies based on simple rules, as when governments flooded the world with liquidity in the aftermath of 2008, have only created asset bubbles that leave us with risks of instability, Sheng said.
Stanford Institute for Economic Policy Research scholar Alissa Kleinnijenhuis presented her research on bail-in mechanisms and their effect on the “too big to fail” problem. She used a complexity-based model because systemic risks arise from networks that are interactive and lead to amplification or dampening of crisis. Her research shows how that effective bail-in designs can break harmful deleveraging cycles, and provide recapitalization support for institutions to preserve banking system stability. This has critical impact on stability of the system. Modeling the European bail-in mechanism, the research suggests mechanism does dampen risks, but still far from optimal. Such research can help policymakers to improve on these mechanisms.
Long Chen, president of the Luohan Academy, closed the operational session as discussant by leveraging his deep experience with digital platform, presented modern firms as much more interconnected. Chen illustrated how digital technology facilitates close feedback loops with consumers and suppliers, allowing firms to adapt their behaviors in response to information. While scholars usually study networks after the fact, firms enabled by connected real-time information can design systems that react to change, to support positive outcomes or dampen negative ones. For example, algorithms used by digital lenders can dynamically adjust loan issuance to respond to shocks. The adaptability of firms to external environment is only growing more critical, especially in rising expectations of firms to be involved in environmental, social and governance (ESG) issues.
During the open-floor discussion, Harvard economics and mathematics professor Eric Maskin suggested that complexity economics remains outside of the mainstream because it needs not only to show that standard economics fails, but have in place a solid framework that performs better. Maskin pointed to prospect theory, advanced by behavioral economists, as an example.
Doyne Farmer agreed with Maskin that complexity economics should be challenged to show that they’ve been superior in understanding the real-world implications of policies. But Farmer believes complexity needs a “fair hearing” from mainstream journals, who have been less willing to publish the work of complexity economists.
Kleinnijhenhuis suggested that ideas from complexity economics have taken hold and are already well-understood within the profession especially in her focus of modeling financial crisis. Many mainstream papers use complexity models even if not explicitly referencing it. Columbia Business School professor Neng Wang noted that while he is very open to complexity economics, he still believes any form of economics needs to assume an optimizing rationality. Wang would like to see more precise definitions of complexity economics. That said, he notes Alissa’s point that mainstream economics is already incorporating complexity tools.
Peking University economics professor Chen Ping, one of China’s most followed economists, points to how mainstream economics continues to turn a blind eye to more atypical models because of its reliance on the idea of unlimited growth. In the current climate crisis, the need to recognize resource limits is putting pressure on economic foundations.
Finally, the Santa Fe Institute’s Arthur predicted that complexity economics will coexist with neoclassical economics in the same way that nonlinear mathematics never replaced linear mathematics. Just as relativity theory and quantum physics exists alongside Newtonian ideas, both economic schools are needed.
