Information and noise in speculative markets: Two essays.

Abstract

The study makes three major contributions towards understanding the role of asymmetric information and noise in speculative markets. First, a model of informed speculation is developed in which the informed agent trades strategically so as to withhold part of the information. This leads to a model of imperfect competition which avoids the unsatisfactory "schizophrenia" problem associated with the competitive models in which each trader takes the equilibrium price as given despite the fact that he influences the price. The model yields predictions about the adverse selection component of the bid-ask spread, volatility of prices and the autocorrelation structure of returns. The determinants of the bid-ask spread is shown to depend on the fundamental parameters governing the information process and the noise in the system. Second, an empirical methodology to test the predictions of the model is developed. This is achieved by postulating that the stock price dynamics follow a jump diffusion process where it is assumed that the diffusion component is induced by noise, and the jump component is due to the impact of significant information. Parameter estimates of the stochastic model are used as surrogates to test the predictions. The empirical results confirm the predictions of the model. Finally, an event-study methodology is introduced that is based on the generalized Poisson jump diffusion model for the stock price dynamics. The model consists of a Weiner process which captures the normal fluctuations in stock prices, and an independent compound event process which models the price reaction to events. The structure added by separating the event process from the non-event process leads to a cumulative event return estimator that is more efficient and has higher power than the traditional multi-day abnormal return estimator for multiple events with event day uncertainty. A maximum likelihood technique is implemented to estimate the parameters of the model, and simulations confirm the higher power and efficiency of the estimator. The method is applied to study the impact of greenmail on stockholders and it is found that shareholders earn a positive return for the overall period including the blockholding and repurchase.