CMStatistics 2017: Start Registration
View Submission - CFE
A1661
Title: Spatio-tempral autoregressive conditional heteroskedasticity model Authors:  Takaki Sato - Tohoku University (Japan) [presenting]
Yasumasa Matsuda - Tohoku University (Japan)
Abstract: A spatio-temporal extension of time series autoregressive conditional heteroskedasticity (ARCH) models is proposed. We call the spatio-temporally extended ARCH models as spatio-temporal ARCH (ST-ARCH) models. ST-ARCH models specify conditional variances given same time surrounding observations and past time observations, which constitutes a good contrast with time series ARCH models that specify conditional variances given past observations. A spatial weight matrix which quantify the closeness between observations is used to express effects from surrounding observations. We estimate the parameters of ST-ARCH models by a two-step procedure of least squares and the quasi maximum likelihood estimation to avoid bias of least squares estimators. We demonstrate the empirical properties by real data analysis of stock price data in the Japanese market to show the relation between volatility of a particular stock and change rates of same time and past time other stock prices.