Agent-based modelling and market microstructure

This thesis advances the literature by applying agent-based simulation to market microstructure issues. Although most of the agent-based literature focuses only on generating market dynamics that best resemble those in the real world, a trading structure is presumed to be a Walrasian auction in most of the published literature. This thesis argues that the institutional details of trading are relevant, and need to be reflected when modeling agent behavior in trading models. Chapter 2 compares the convergence property in a double-auction market versus a Walrasian market. This chapter constructs an artificial equity market in which agents trade a risky asset that pays a stochastic dividend each period. In some of the experiments, market dynamics under double-auction converge to the Rational Expectation Equilibrium. However, this convergence is sensitive to deviations from rationality. In experiments where we introduce noise trading, convergence becomes unattainable. Minimal rationality is not sufficient to generate convergence in a double-auction market when the market price is endogenous. Chapter 3 constructs a three-stage model of a dealership market. After trading with the customer, the dealer chooses between trading in a quote-driven inter-dealer market or in an order-driven brokered inter-dealer market. We show that the dealers choice of inter dealer markets depends on the number of dealers available to make the market. This condition determines which inter-dealer market will prevail. We also demonstrate the conditions for the existence of the equilibrium number of dealers that enter the market making industry. At equilibrium, an increase in risk-aversion, the volatility of the customer order flow and asset shifts inter-dealer trading from the direct market to the broker. Moreover, customers are better off when dealers have a choice between two inter-dealer trading venues. Building on this theoretical framework, an agent-based simulation is employed to investigate the welfare effects of heterogeneous dealers under different transparency regimes. Chapter 4 examines the factors that affect liquidity-motivated investors preference between a dealership market and a limit-order book market. This study builds on the theoretical market microstructure literature and uses an agent-based computational approach that allows us to examine the equilibrium properties of models that are richer than the extant literature in terms of the heterogeneity of trading and customer attributes. We find that an increase in the thickness of the market, correlated customer order flow and a decrease in the number of dealers cause a rise in the probability of an order book system prevailing. We also find that large-size customer orders and an increase in the degree of dealer heterogeneity are more likely to shift trades to a dealership system.