A1126
Title: QBIC of SEM for diffusion processes based on high-frequency data
Authors: Shogo Kusano - Kumamoto University (Japan) [presenting]
Masayuki Uchida - The University of Osaka (Japan)
Abstract: Structural equation modeling (SEM) is a statistical method to investigate relationships among latent variables. Since SEM is a confirmatory analysis method, the model needs to be specified in advance based on the theoretical framework of the respective research field. However, statisticians may often have several candidate models for SEM and must choose the optimal one from among them. To address this issue, various information criteria for SEM have been actively studied. In particular, the Akaike information criterion (AIC) and the Bayesian information criterion (BIC) are commonly used. Recently, SEM for diffusion processes based on high-frequency data has been studied. For model selection of the SEM, an AIC-type statistic based on the quasi-likelihood has been proposed. However, this criterion does not ensure model selection consistency. Therefore, BIC-type statistics are considered for SEM. The asymptotic expansion of the marginal quasi-log likelihood is first obtained. Based on this result, two types of quasi-BIC are proposed for SEM, and it is shown that the information criteria have model selection consistency. Furthermore, some examples are provided, and simulation studies are conducted.
