Last update: March 10th, 1998
| 28. | Name: | James G. Uber | |
| Organiz.: | University of Cincinnati; Dept. of Civil & Environmental Eng. | ||
| Co-Authors: | Marco Propato and Ali A. Mina | ||
| JELC: | ??? | ||
| Keys: | Agent-based computational economics; space-time economy; self-organiza | ||
| Title: | Evolution of Production, Consumption, and Resource Use in a Space-time Economy | ||
| INTRODUCTION AND MOTIVATION A deeper understanding of the self-organizing economy has been of considerable interest to economists, but has also been a goal of geographers, demographers, anthropologists, and urban planners. Significant progress has been made in understanding particular pieces that fit into an overall understanding of the space-time economy, such as economists' interest in specialized markets, demographers' interest in migration, and the new economic geography, which explores the explicit ties between demographics and economics. Relatively less progress, however, has been made in formulating and solving research questions about the evolution of the whole economy. Such questions would contemplate the aggregate spatial and temporal distribution of heterogeneous producers (corporations) and consumers (individuals) that arise from dynamic interactions between corporations, individuals, and the spatial distribution of natural resources (some renewable, some not) that provide basic inputs for economic production. Indeed, while popular phrases such as ``sustainability'' or ``sustainable development'' evoke images of important connections between economy, population, and valuable resources (even those that literally sustain life), these images seem far from clear at the present time. We believe agent-based models can be used to formulate and investigate issues such as the global interactions between producers and consumers and the supply of basic resources, in order to obtain insights about the controls (or, lack thereof) on space-time evolution of the economy and long-term ``sustainability.'' Unlike the conventional dynamic systems models (e.g., Forrester's ``world dynamics''), an agent-based approach allows the explicit inclusion of insights from fields like sociology or psychology into large-scale models, thereby relating ``micromotives'' and ``macrobehavior'', to use Schelling's eloquent terminology. BRIEF MODEL DESCRIPTION In approaching the types of problems outlined above, we have found it important to consider explicitly a hierarchy of economic agents, and the euclidean space in which they exist. We model a simple, spatially distributed economy populated by consumer and corporate agents. These two agent types mimic the two symmetric components of classical general equilibrium economics: corporations supply goods to the market according to profit maximizing behavior, and in the process create a demand for labor and other production inputs (e.g., land and natural resources); consumers supply factors of production and, based on utility maximizing behavior, create a demand for goods in the market. We are interested in the ways that these dynamics play out in space, creating and reacting to changes in demographics and resource availability. Especially, our ultimate interests lie in understanding the influence of spatio-temporal resources that flow on a directional network connecting all spatial locations (notably, as does water in a stream channel network), and thus impart an asymmetric interaction between economic activities at different spatial locations. Our initial model is quite simplified. We consider corporate agents that produce for two different economic sectors: agriculture and manufacturing. Each corporate agent can occupy a single site on a fixed 2-D spatial lattice representing the physical space. Production at that location is determined by a production function that depends on labor and resource inputs (other inputs, such as land and capital, are obviously important, but ignored in the present model). The produced goods are shipped to particular lattice sites, where they are sold in a local market for a price that reflects both unit production and transportation costs, the latter depending on distance between the production and market sites. Each consumer agent also has a location on the same 2-D lattice, and is associated with a particular corporate agent at the same site. Consumer agents purchase a product mix from the local market, at local prices, such that utility is maximized subject to a budget constraint. This budget constraint depends on wages received from their associated corporate agent. Consumer agents are able to change their location according to spatial differences in utility caused by spatial differences in prices and wages. Therefore every location is, in general, occupied by many consumers at any given time. At any step, corporate agents are (potentially) required to decide: 1) The production output level, 2) The production factor mix to achieve the desired output level, 3) The selling price of the good, 4) where the good should be sold (in which markets), and 5) the labor wage rate. In order to make such decisions, the corporate agents may rely on market signals such as historic demands and associated prices, prices of similar goods produced by other corporate agents, demographic and employment data, and local resource availability. Similarly, consumer agents may be required to decide: 1) the good mix (based on utility maximization and budget and inventory constraints), 2) the particular corporate agents from which to purchase each good (selected from those selling goods in the local market), and the best location to live. In arriving at such decisions consumer agents may rely on market prices, labor wage rates, their history of interaction with corporate agents, and information supplied by other agents. We are currently engaged in hypothesizing and testing various agent representations and adaptive schemes for improved decision-making over time. Insights obtained through this work will be discussed in our paper. DISCUSSION The model framework we describe is certainly a simplified version of a real worl economy; it does, however, incorporate two main factors that agent-based economic models often omit. The first is the explicit modeling of corporate and consumer agents in a hierarchy. The two types differ in their goals and actions, but are linked by an inexorable cycle of production and consumption. Thus, the beneficiaries of a firm's sales and the consumers of its products are the {\it same} individuals whose labor produced the goods. This captures the often paradoxical dependencies that exist between individuals and corporate employers in the real world, and are manifested in labor disputes, downsizing, and the economic cycle itself. Our model is clearly too simplistic for such details, but does represent progress towards the inclusion of these issues. The second significant aspect of our model is the explicit inclusion of space. While many game-theoretic models also do so, their dynamics is typically rather simple, and involves space only in the definition of neighborhoods. In fact, spatial factors are important for economic models in several ways, two of which are of direct relevance to our model: (1) Different locations confer different degrees of access to and control of resources (such as water). For example, a river network creates a highly nonuniform spatial distribution of water, and the flow of water within this network produces directional dependencies between the locations; (2) A spatially extended system implies inescapable limitations, costs and delays associated with transporting agents, resources, products, and information. An important consequence of spatial distribution and its costs is the limitations it places on the rationality of agents. Both corporate and individual agents only have information from a limited number of sites in making their decisions, and these sites are chosen primarily on the basis of spatial proximity or prior interaction. The result is the emergence of ``information networks'' and ``social networks'' which connect agents to different subsets of available information, and sustain the essential heterogeneity of the system. It is precisely this lack of omniscience in agents which drives the process of adaptation, and provides room for it. From this perspective the model we propose includes the basic features of a real devoloping society, and simulating its dynamics should be useful for identifying some general rules of the evolving economy. | |||