Information model inference from asset price dynamics

Most previous studies of information based market microstructure models concentrate on the consequences of single items of asymmetric information on asset markets. However, as episodes of asymmetric information are difficult if not impossible for researchers to identify, a model that incorporates asymmetric information's flow into a model of its microstructure consequences may significantly increase the number of testable hypotheses. In Chapter 1, we delineate the link between the time series properties of information flow and the time series properties of return volatility and trade volume. Less temporal aggregation causes returns and trade volume to reflect the microstructure effects of learning. Greater temporal aggregation causes returns and trade volume to reflect the time series characteristics of the underlying information process in a way that is similar to the dependence suggested by the mixture of distribution hypothesis (MDH). However, by acknowledging facets of information typically ignored by MDH models, this work offers an explanation for the documented non-proportionality between serial correlation of trade volume and squared returns that cannot be accounted for by MDH models.; In Chapter 2, a microstructure model based on asymmetric information is developed that, in contrast to previous models, produces stationary asset price and trading volume series. The model naturally captures both public and private information and parameterizes the leakage of information around earnings announcements. Moreover, the structure of the model leads to econometric estimation that explicitly recognizes that the econometrician has less information than the uninformed market participants.; Inference about latent episodes of asymmetric information relies on observed departures in trading behavior from typical trading patterns. In order for inference to be most effective, a clear understanding of what constitutes typical trading behavior is required. In Chapter 3 we document the existence of high-frequency cyclic patterns in the volume and volatility for the stocks that comprise the Dow Jones Industrial Average. A low order polynomial does a surprisingly good job of explaining the intraday patterns of 5-minute aggregates. We also show how the previous close and the current open affect the intraday pattern.