|
Abstract:
This paper provides a filtering methodology for discretely observed, continuous-time jump-diffusion models. Our approach applies particle filtering methods to a system with data points simulated between observation times. We demonstrate the approach's accuracy and we address three empirical issues related to jumps and stochastic volatility. First, we analyze how sampling frequency affects the volatility and jump estimation. Second, we use the filter to separate the effects of jumps and stochastic volatility, and show how the estimates depend on the given model. Finally, we filter volatility and jumps jointly from index returns and option prices, to quantify the information embedded in option prices. Joint work with Michael Johannes from Columbia University and Nicholas Polson from the University of Chicago. |
©
2003 Purdue University Last Update: Feb 5, 2004 Please send comments and suggestions to the Webmaster. |