B1558
Title: Maximum-likelihood estimation for jump-diffusion processes with nonsynchronous observations
Authors: Teppei Ogihara - University of Tokyo (Japan) [presenting]
Abstract: For two-dimensional jump-diffusion processes, the properties of maximum likelihood type estimators with nonsynchronous observations are examined. Nonsynchronous observations are a fundamental issue in high-frequency data in financial markets, as stock prices are observed when transactions occur, leading to the problem where observation times do not necessarily align across different securities. Additionally, jump-diffusion processes are used to model sudden fluctuations in stock prices and serve as models for insurance companies' stock price movements and asset transitions. The threshold for jump detection by another study is used to distinguish between the jump and continuous parts and apply the asynchronous Gaussian likelihood function of a prior study to the continuous part to construct a quasi-log-likelihood function and propose a maximum-likelihood-type estimator. Asymptotic properties, such as the consistency and asymptotic normality of the estimator, are demonstrated and its optimality is discussed in terms of asymptotic efficiency.
