Trading Dynamics in the Foreign Exchange Market: A Latent Factor Panel Intensity Approach

Abstract

We develop a panel intensity framework for the analysis of complex trading activity datasets containing detailed information on individual trading actions in different securities for a set of investors. A feature of the model is the presence of a time varying latent factor, which captures the influence of unobserved time effects and allows for correlation across individuals. We show how to estimate the model parameters using a simulated maximum likelihood technique adopting the efficient importance sampling approach of Richard & Zhang (2007).Apart from the innovative methodology enabling detailed characterization of a complex dynamic trading process, we contribute to the literature on behavioral finance by providing new results on behavioral biases such as the disposition effect and investor overconfidence. These new insights are made possible by the joint characterization of not only the decision to close (exit) a position, usually considered in isolation in the behavioral finance literature, but also the decision to open (enter) a position, which together describe the trading process in its entirety. While the disposition effect is defined with respect to the willingness to realize profits/losses with respect to the performance of the position under consideration, we find that the performance of the total portfolio of positions is an additional factor influencing trading decisions and can reinforce or dampen the standard disposition effect. Moreover, the proposed methodology allows the investigation of the severity of various behavioral biases for different groups of investors ranging from small retail investors to professional and institutional investors.