A Simple Complete Conservative Sustainable Agent-Based Model Economy –

Some Hypotheses

Garvin H Boyle

(9th February 2013)

Orrery Software, PO Box 1149 (URL: modeco-software.webs.com)

Richmond, Ontario, Canada, K0A 2Z0 (e-mail: )

Abstract: As part of an enrichment program for highly gifted students in an Ontario high school, the ModEco application was conceived as a desk-top laboratory in which students could design and study very simple sustainable agent-based model economies. A ModEco-based economy has two interacting sub-systems: (i) at the biophysical level two codependent groups of agents (farmers and workers) exist in a world constrained by the laws of conservation of mass and energy. The agents live and evolve to become well-adapted agents competing for life-sustaining resources. They harvest food, eat it, produce waste, and use that waste to fertilize the fields for the next harvest. (ii) At the economic level all transfers of mass and energy (in the form of goods and services) must be reciprocated with a transfer of cash. However, no preexisting examples of sustainable agent-based model economies could be found. Economic sustainability is not easy to achieve, even in a simple economy. All but one ModEco-based economy quickly collapsed with the death of all agents due to economic failure. The one sustainable economy had severe constraints on prices. The author aims to encourage the serious study of the dynamics of sustainable economics using agent-based models by (i) making the ModEco code available from an online archive; (ii) making an ODD description of the model available; and (iii) tabling a set of testable hypotheses about the nature of sustainable model economies.

Keywords: agent-based computational economics, biophysical economics, conservative system, ecological economics, evolutionary system, maximum entropy production principle, maximum power, ModEco-based economy, steady-state economics, sustainable economics.

1 – INTRODUCTION

The software application that plays a key role in this paper, called ModEco, is designed with the intent to allow students of sustainable economies to design a model agent-based economy on a desktop computer, and see the economy develop in minutes as they watch. Inspired by an interesting book called “The Ecology of Commerce” by Paul Hawken [1], it was developed in conjunction with an enrichment program for exceptionally gifted students in an Ontario high-school. The most recent version of the software, together with some supporting documentation and derivative analyses, is available online. This includes a high-level design document, and an ODD model description, intended to make the model replicable on other platforms [2, 3]. The ultimate goal in the development of the ModEco application was to identify the necessary and sufficient components of a most simple but complete sustainable economy. That goal is unfortunately still a future hope.

A ModEco-based economy is a merger of two synchronized and closely linked dynamic sub-systems fashioned with a very simple post-barter agricultural economy in mind. An economic sub-system rides atop a biophysical sub-system. The agents come in two varieties playing two different roles: farmers and workers.

At the biophysical level, farmers and workers cooperate to harvest grain from fields and store it in the farmers’ inventory bins. They then carry the grain home and place it in their pantries, eat it, and take the waste to the recycling depot, from which farmers collect it and put it back on the field to start the cycle over again. Agents’ lives are controlled by biophysical parameters that determine things like food consumption rates, reproductive maturity, or death by starvation or old age. Agents compete for life-sustaining resources, and access to resources is mediated by genetic code. Those agents that have small genetic advantages tend to survive and reproduce offspring with small mutations, while those with small genetic disadvantages tend to die without offspring. This is the biophysical sub-system. It is based on a previously built agent-based sustainable biophysical model called PSoup that worked well, also available online [3].

Tightly intertwined with the biophysical sub-system is the economic sub-system. Farmers hire workers to work (with them) in the fields to harvest the grain. They then sell the grain to consumers. The consumers sell waste to a central manager who, in turn, sells the waste back to the farmers. A single good is bought and sold in its various forms (fresh grain, inventory, supplies, waste). A single service is bought and sold (harvesting grain).

Together, these two logical sub-systems are so tightly linked that they form a single dynamic system having both biophysical and economic characteristics. Within this system all mass and energy flows in one direction, mediated by transfers of cash in the opposite direction.

This economy is described as “complete” because (i) it includes both biophysical and economic dynamic sub-systems; and (ii) it models a complete cycle of mass and energy as it flows through both sub-systems, from farmer’s field, through consumers and back to farmer’s field. We describe it as “most simple” because it is difficult to conceive of an economy that is more

7

simple, but still complete in the sense described above.

Ideally a “sustainable” computer-based model economy would run forever without population collapse, as was possible with a purely biophysical agent-based model in PSoup. More pragmatically, a model is designated sustainable if it will run for one million cycles without collapse.

Regrettably, to date, a simple sustainable ModEco-based economy has been demonstrated only under very severe constraints. This highly idealized sustainable model economy has been dubbed the Perpetual Motion Machine (or PMM). The intellectual journey that has led to the successful design of this sustainable model economy has been difficult, and full of unexpected insight. The purpose of this article is to briefly share a variety of those insights, and to lay out a challenge to other modellers in the form of a set of testable hypotheses. It is sincerely hoped that those who have skills in developing agent-based models will find a satisfying challenge in testing these hypotheses, will successfully disprove them, will replace them with more qualified and elaborated hypotheses, and thereby improve our understanding of the dynamics of sustainable economies.

A model that is “closed” with respect to a conserved quantity is easy to debug, since, with every cycle of the economic engine, one can check the total amount of that quantity, and find the reason it is not conserved. The ModEco economic engine is closed with respect to the conserved quantities of mass and energy, and several thousand internal verification checks are made to ensure they are conserved in every step of the processing. Work is underway, but not completed, in opening up the models with respect to both energy (flowing from source to sink) and mass (flowing from low entropy endowments to high entropy landfill). The linear energy engine is available in prototype [3] as part of ModEco version 2.04A. Work on the linear mass engine has not yet started.

The remainder of this article is laid out as follows:

Section 2 summarizes the important technical characteristics of the closed cyclic agent-based model that is called the PMM, or Perpetual Motion Machine. The PMM is the only ModEco-based model economy produced to date that is sustainable.

Section 3 enumerates a number of insights and observations collected along the way as the search for a sustainable economy progressed.

Section 4 lists several hypotheses that are tabled therein as a challenge to other people with modelling skills and an interest in the dynamics of sustainable economics.

2 – TECHNICAL CHARACTERISTICS OF THE PMM

As mentioned above, a sustainable economy has been established only under severe constraints, and that constrained economy has been called the Perpetual Motion Machine, or PMM, due to the inspiration derived from the vision of an idealized friction-free bicycle wheel spinning freely forever. In the PMM there are quotas limiting the size of all commercial transactions, prices are precisely controlled, gene-based variations in agent abilities are neutralized, and central authorities redistribute wealth according to public policy. Clearly the PMM is an extremely constrained idealization of what most people would envisage as a “sustainable economy”. Even relative to the possible range of features available in ModEco, the PMM is a feature-poor economy. However, to date, all variations from this constrained model quickly collapse. The PMM, on the other hand, has run to an amazing thirty million ticks. The fact that the PMM is the only sustainable model economy achieved so far makes it of particular interest, and the focus of this article.

But before going into the details of the PMM, first one must understand the nature of a ModEco-based economy.

2.1 – Technical Characteristics of a ModEco-based Economy

In ModEco, different economies can be initialized by enabling options and changing parameters. Each such initialization is a demonstration of a different logical economy. They are viewed as demonstration economies, and not simulated economies, because they are very highly abstracted, and there is no real intent or effort to “simulate” any specific economy. Each forms a logical complex dynamic economy in its own right, and can be studied as such. They are all internally verified to be working as intended, with over 3,500 verification tests, but they are not validated with respect to real-world economies, as that is not their intent. They are nevertheless able to provide insight into the fundamental nature of real-world economies by way of analogy.

Each ModEco-based economy:

2.1.1 – Is agent-based – Two types of agents, farmers and workers, are co-dependent. That is to say, if one population collapses, so does the other. If one expands, so does the other.

2.1.2 – Has two spatial dimensions – The action in the model happens in a two-dimensional grid of squares called lots, and the rectangular collection of all lots is called the township. The lots wrap at the edges to form a logical toroid. Any agent can do business with any other suitable agents within a five-by-five square called its commuting area centered about each agent. For example, a worker can get a job with, or buy supplies from any farmer within its commuting area, but cannot conduct such business with farmers outside of its commuting area.

2.1.3 – Has a joint biophysical/economic transition rule – Time is measured in discrete units called ticks. In one tick of the ModEco clock the following ten functions happen in order: (i) setup; (ii) make job offers; (iii) move workers; (iv) sell inventory; (v) consume supplies; (vi) sell waste; (vii) buy recycled waste; (viii) reproduce agents; (ix) death of agents; and (x) cleanup. Together, these ten functions are called the ‘economic engine’, and it is executed once per tick of the ModEco clock, so completing one ‘cycle’ of biophysical/economic activity. Details about this process can be found in the very high-level design document. [2]

2.1.4 – Uses metabolism-based determination of value – In keeping with the philosophical perspective of biophysical economics [4] that views mass and energy flows and entropy production as fundamental economic processes, the biophysical needs of the agents are defined, and units of mass and energy are defined in terms of cyclic metabolic requirements, and the intrinsic value of mass and energy is calibrated using agents’ metabolic needs. Money has no intrinsic value, except as derived from its use in a transaction.

2.1.5 – Is conservative of key quantities – The conservative nature of a system can be a characteristic of individual transactions, or of the system as a whole. During each transaction between agents, biophysical quantities such as mass, energy and intrinsic value are conserved, as well as cash. At the system level, each economy is a closed mass/energy system with no mass, energy or intrinsic value flowing in or out. Cash is a special case for which the system may be open or closed, depending on a user-selected option.

2.1.6 – Is complete – In this context the word “complete” is meant to convey wholeness in two important dimensions. Firstly, the model is both biophysical and economic, with the two sub-systems in the model tightly synchronized. You can view this as complete from top-to-bottom. Secondly, the entire economy is demonstrated from end-to-end, as mass and energy flow from primary production on the farm, through harvest, to farm inventories, to consumer pantries, to waste collection facilities, and then back to the farm where the mass and energy are recycled for re-use. Sustainability is a characteristic of a complete economy, and not of a component of an economy, so a study of sustainability requires that the economy demonstrate complete mass and energy flows. There is a third possible meaning of the word “complete” that, unfortunately, is not included within the current scope of the design of the ModEco economic engines. People share the world with other species, and destruction of those species will lead to extinction of people as well. We can view the necessary sharing of the world with other species as a front-to-back completeness. Hall and Klitgaard make this point very clearly at [4], and Costanza et al. identify this as a key component of sustainability on page seven of the report at [5]. ModEco economies do not model any aspect of dependence on other species. That very important aspect of completeness was beyond the scope of ModEco at the time of inception.