An Agent-Based Computational Model for the Evolution of Trade Networks

David McFadzean - Kumo Software Corporation and Leigh Tesfatsion - Iowa State University

This paper describes an agent-based computational model for studying the formation and evolution of trade networks in decentralized market economies. A virtual economic world is constructed, populated by heterogeneous endogenously interacting traders with internalized data and modes of behavior. This virtual world can be used to study the formation and evolution of trade networks under alternatively specified market structures at three different levels of analysis: individual trade behavior; trade interaction patterns; and social welfare.

More precisely, agent-based computational economics (ACE) is the computational study of economies modelled as evolving decentralized systems of autonomous interacting agents. A key concern of ACE researchers such as Epstein and Axtell [2] is to understand the apparently spontaneous appearance of global regularities in economic processes, such as the unplanned coordination of trading activities in decentralized market economies that economists associate with Adam Smith's invisible hand. The challenge is to explain how these global regularities arise from the local interactions of autonomous agents channeled through actual or potential economic institutions rather than through fictious coordinating mechanisms such as imposed equilibrium conditions.

The present paper, a summary version of [3], discusses the C++ implementation of a particular ACE model developed by Tesfatsion [5] to study the endogenous formation and evolution of trade networks. This model, referred to as the trade network game (TNG), extends to an economic setting an earlier model ([1],[4]) combining evolutionary game play with endogenous partner selection. In the TNG, successive generations of resource-constrained traders choose and refuse trade partners on the basis of continuously updated expected payoffs, engage in risky trades modelled as two-person games, and evolve their trade behavior over time.

In particular, traders are instantiated as autonomous endogenously interacting software agents ("tradebots") with internal behavioral functions and with internally stored information that includes identifiers for other tradebots. The tradebots can therefore display anticipatory behavior (expectation formation), and they can communicate with each other at event-triggered times. The tradebots use these and other capabilities to determine their trade partners and to evolve their trade behavior. The modular design of the TNG program permits experimentation with alternative specifications for market structure, trade partner matching, trading, expectation formation, and trade behavior evolution. All of these specifications can potentially be grounded in tradebot-initiated activities. The TNG program thus facilitates the general ACE study of trade networks.

Source code and executable files for the TNG program can be downloaded from the website of the second author (see above).

Scheduled for Session 5.6 Computer Languages - I