EcoSta 2019: Start Registration
View Submission - EcoSta2019
Title: Realized jump beta: Evidence from high-frequency data on Tokyo stock exchange Authors:  Masato Ubukata - Meiji Gakuin University (Japan) [presenting]
Abstract: The focus is on jump betas of some sector portfolios in the Japanese stock market and to assess the dynamics in realized jump betas estimated from the high-frequency data. We test the null hypothesis of constant jump betas over years and months. The results show that annual constant jump betas are always rejected at 1\% significance level, while monthly constant jump betas are not rejected for about a half of the monthly samples. Given the worldwide evidence of fractional integration in realized variance and covariances, the estimation results under the assumption of a pure fractional noise process of the monthly jump betas indicate a small average degree of integration, namely ARFIMA(0, 0.2, 0). However, a further analysis implies that the monthly realized jump betas are arguably modeled as stationary I(0) processes.