A1858
Title: SEM for jump-diffusion processes based on high-frequency data
Authors: Shogo Kusano - Kumamoto University (Japan) [presenting]
Masayuki Uchida - The University of Osaka and University of Tokyo (Japan)
Abstract: Structural equation modeling (SEM) for diffusion processes with jumps based on high-frequency data is considered. In recent years, high-frequency data have become easily available due to advances in information technology. To conduct SEM based on high-frequency data, SEM for diffusion processes has been previously proposed. However, in many applications, discontinuous sample paths are often observed. Since SEM for diffusion processes assumes continuous paths, it is not suitable for analyzing such data. To address this issue, SEM for diffusion processes with jumps is proposed. By using a threshold method based on increments, a quasi-likelihood function for the SEM is constructed. The quasi-maximum likelihood estimator is shown to have consistency and asymptotic normality. Furthermore, to examine whether a specified parametric model is correctly specified, a quasi-likelihood ratio test statistic is developed and its asymptotic distribution is derived. The proposed framework enables analysis of relationships between latent processes based on high-frequency data with jumps. Finally, numerical simulations are conducted to illustrate the finite-sample performance of the estimators.
