Dynamic Modeling Data Time Series By Using Constant Conditional Correlation-Generalized Autoregressive Conditional Heteroscedasticity

Abstract

Abstract The Constant Conditional Correlation-Generalized Autoregressive Conditional Heteroscedasticity (CCC-Garch) model as one of the multivariate time series models is used to model economic variables, especially in stock price data with high volatility characteristics that result in heterogeneous variations. The higher the volatility, the higher the level of uncertainty of the stock returns that can be obtained. The CCC-Garch Multivariate model is the simplest model in its class. The principle of this model is to decompose the conditional covariance matrix into conditional standard deviation and correlation. In this study, we will discuss and determine the best model that can describe the relationship between two vector data timeseries, namely stock return data for companies engaged in mining and construction in Indonesia, namely United Tractor Tbk (UNTR) and Petrosea Tbk (PTRO) where the data is the daily stock return data for the period July 2015 to August 2020. Several models that involve modeling the mean and variance with CCC-GARCH parameterization are applied to data such as the VAR (1)-Garch (1,1), VAR (2)-Garch (1) model., 1), VAR (3) Garch (1,1) and VAR (4)-Garch (1,1). The result was that the VAR (1)-Garch (1,1) model was selected as the best model with the criteria for selecting the AICC, SBC, AIC and HQC models. The dynamic behavior of both UNTR and PTRO stock return variables is explained by Granger Causality and Impulse Response. Furthermore, the forecasting of this data was carried out for some time in which the VAR (1)-Garch (1,1) model which was selected as the best model was only suitable for forecasting in a short time.