Animal spirits and stock market dynamics

Abstract

This dissertation consists of two independent studies-which are closely related-that build agent-based computational stock market models. The main objective of these models is to investigate the impacts of animal-spirit shocks on fundamental values as well as the impacts of heuristic trading rules on stock market dynamics within a new computational framework distinct from mainstream neoclassical economic models.The second core chapter develops an agent-based model of a dynamic investment economy to examine the role of animal-spirit shocks in the determination of firm fundamental values. The economy is populated by traders with intertemporal utility objectives who engage in consumption, labor, and asset investment activities in an attempt to increase their utility over time, and by a corporate firm with an intertemporal profit objective that engages in R\&D in an attempt to increase its profits over time. It is shown that a one-time animal-spirit shock, modeled as an abrupt purchase of additional IPO stock shares by one of the traders, can have persistent effects on the determination of firm fundamental values, measured as earnings per share, as well as on other critical system outcomes. Moreover, these effects can be amplified or contracted depending on the connectivity of this animal-spirit trader within a social network, and the extent to which traders desire to conform to the behaviors of other traders within this social network.The third chapter develops a computational stock market model in which each trader's buying and selling decisions are endogenously determined by multiple factors: namely, firm profitability, past stock price movement, and imitation of other traders. Each trader can switch from being a buyer to a seller, and vice versa, depending on market conditions. Simulation findings demonstrate that the model can generate excess volatility, a fat-tail property, and the ARCH effect in stock returns.The results also suggest the importance of trader memory length for determining the stability of stock prices in response to dividend shocks.