An Agent-Based Model of Dealership Markets

This paper describes an agent-based model of financial markets with monopolistic or competitive market-makers and analyzes some of the emergent properties of these markets, including time series properties. The artificial markets we discuss utilize models of “informed” trading agents who decide to trade based on received signals of the true or fundamental value of the stock, and “uninformed” trading agents (sometimes called liquidity traders) who trade for exogenous reasons and are modeled as buying and selling stock randomly. These simple models of traders, combined with more complex market-making agents who function as price-setters and inventory-holders in the market, lead to a rich array of market properties, many of which qualitatively replicate properties observed in real financial markets. For example, the bid-ask spread increases in response to uncertainty about the true value of a stock, average spreads tend to be higher in more volatile markets, and market-makers with lower average spreads perform better in competitive environments. The time series data generated by our market models demonstrate phenomena like volatility clustering and the fat-tailed nature of return distributions, without the need to specify explicit models for opinion propagation and herd behavior in the trading crowd. Our models of trading are simpler than most models in the literature that succeed in demonstrating such properties.