Zhenyuan Zhao

There is widespread agreement that carbon emissions need to be reduced but there is little agreement on how this should be done. Aside from the long-term goal of creating new emissions-friendly technologies, the immediate issue concerns how to globally control existing emissions. Despite many international summits on global warming and its high profile in the media, there is very limited quantitative understanding of the extent to which institutions or governments can in principle control total emissions without having to continually intervene to micro-manage daily quotas, and hence lose their free-market ethos. By building on a novel theoretical Complex Systems framework developed in part by my supervisor, my project will address this issue. In this article I describe this approach and some preliminary results already obtained.

The purpose of this project is to explore the extent to which free competition, linked with minimal global control, can lead to a self-organized capping of the global emissions. Via computer simulations I will study a model in which a population of competing, adaptive emitters make decisions on when to emit based solely on the behavior of some shared public information. My preliminary work shows that within this simple framework, the emitters can organize themselves in order to collectively hit their emissions target at the expense of some quantifiable fluctuations in the total volume emitted. Most importantly, they can achieve this without the need for any external regulation or manipulation of the market.

Specifically, our model considers an ecology of companies who are continually trying to outguess each other in such a way that they end up emitting at the right time. Our model is a generalization of both the so-called “El Farol Bar Problem” and the “Minority Game,” in which agents repeatedly compete for some limited resource [1, 2, 3, 4]. The companies constitute a heterogeneous population with possibly quite different strategies but similar capabilities and who make their respective decisions about emitting based on some knowledge of past history or limited public information. In the El Farol Bar Problem [1], agents decide whether to visit a bar with a limited seating; correct (incorrect) decisions correspond to visiting an undercrowded (overcrowded) bar or not visiting an overcrowded (undercrowded) bar. In the context of the carbon market, our model assumes that the goal of the government is that the companies collectively emit no more than some predetermined total of carbon pollutants each month. If this limit is exceeded then the amount of carbon emitted into the atmosphere is too high, but if the aggregated emissions are too low then this suggests some wasted production capacity. The only information given to the companies after each day is whether or not the actual emissions level exceeded or fell below the average daily value of the monthly cap. Each company makes its decisions based on the strategies it holds, with the best-performing strategy being used at any given moment. In an ideal world, all companies want to be operating (and hence emitting) every day. But we assume that any given company will be sanctioned by the government or the national press if it emits on an overcrowded day (i.e. it emits on a day when too many others are also emitting). Likewise, the company will be sanctioned by its stockholders or customers if it fails to emit on an undercrowded day (i.e. it fails to emit on a day when few others are emitting) since this would represent a wasted opportunity. The model allows us to explore the consequences of many different forms of penalty-reward structure.

The net performance of the overall system is assessed through an analysis of the mean and maximum aggregated emissions over a fixed period of time, and the standard deviation of this aggregated emission about the mean. Based on preliminary simulations we expect the results to show that within the basic constraints of the model, companies are able to organize themselves to hit their collective monthly emissions target with relatively minor fluctuations in the aggregated emissions each month and in the absence of any external regulator controlling the market. We will explore the extent to which companies react to changes in the monthly emission limit, and the difference between this behavioral change for both an incremental and a sudden cap reduction. This will provide insight into the most efficient method for reducing the emissions cap within the carbon market.

As documented in Johnson et al. [5], we know that the underlying model concept works well for financial exchange markets and regular stock markets, that is, it reproduces quantitatively the fat-tail distributions, clustered volatility, and bursty behavior typical of markets. We are now applying this to emissions markets. Both the regular and emissions markets have the same human aspect of yes/no decisions in response to limited global information and a maximum global capacity, so it is likely to be a good first approximation in terms of emissions markets. As emissions markets are established in the next few years, we will be able to test out their behaviors in terms of our common model of collective competition, and hence refine the model according to specific regulations, etc.

Although this model setup is not unique and arguably leaves out many possible complications, we believe that it does indeed incorporate the essential ingredients and hence provides a potentially useful laboratory for exploring the dynamical behavior of future carbon emissions.

[1] W.B. Arthur. “Inductive reasoning and bounded rationality (the El Farol problem).” Amer. Econ. Assoc. Papers. Proc., 84:405, 1994.

[2] N.F. Johnson, et al. “Volatility and agent adaptability in a self-organizing market.” Physica A, 258:230, 1998.

[3] D. Challet and Y.C. Zhang. “Emergence of cooperation and organization in an evolutionary game.” Physica A, 246:407, 1997

[4] N.F. Johnson, P.M. Hui, D. Zheng, and Tai C.W. “Minority game with arbitrary cutoffs.” Physica A, 269:493-502, 1999.

[5] N. F. Johnson, P. Jefferies, and P. M. Hui. Financial Market Complexity (Oxford University Press, 2003).