Page 50 - Proceeding The 2nd International Seminar of Science and Technology : Accelerating Sustainable Innovation Towards Society 5.0
P. 50

nd
               The 2  International Seminar of Science and Technology
               “Accelerating Sustainable innovation towards Society 5.0”
               ISST 2022 FST UT 2022
               Universitas Terbuka
               reducing the error rate in calculations. The basic concept of GARCH
               is that variance is not only influenced by past residuals but also by the
               lag of the conditional variance itself [6].
               Thus, the conditional variance in the GARCH model consists of two
               components,  namely  the  past  component  of  the  squared  residual
               (denoted  by  degree  p)  and  the  past  component  of  the  conditional
               variance (denoted by degree q), in Equation (6).
                2                                2
               ó = ù + ∑     =1 á    2   −1  + ∑   =1 â ó    −1                         (6)
                                 
                                                 
                   
               2.3.1  Stages of GARCH Analysis
               The  steps  used  to  implement  the  GARCH  (Generalized
               Autoregressive Conditional Heteroskedasticity) model on Indonesian
               export value data for the period January 2014-February 2022, are as
               follows:

               1.  Perform stationarity test and normality test
               2.  Identify the ARIMA model based on the ACF and PACF charts
               3.  Determine the ARIMA model estimate
               4.  Test the significance of the parameters on the estimation results
                   of the ARIMA model
               5.  Test   assumptions   (normality,   heteroscedasticity,   and
                   autocorrelation)
               6.  Identify the GARCH model with the ARIMA model
               7.  Define the GARCH model
               8.  Test the significance of the parameters on the estimation results
                   of the GARCH model
               9.  Perform the LM-ARCH test
               10. Doing the forecasting stage
               2.4   Akaike dan Schwarz Information Criterion (AIC and SIC)
               Akaike and Schwarz criteria (AIC and SIC) in model selection can also
               be  done  using  Akaike  Information  Criterion  (AIC)  and  Schwarz
               Information Criterion (SIC) in Equation 7, and 8 [5], where    = 2.718,
                  = residual,     = number of estimation parameter variables,         =
                sum of squared residual, and    = number of observations (sample).
               The model chosen is the model that has the smallest AIC and SIC.

               ISST 2022 – FST Universitas Terbuka, Indonesia             29
               International Seminar of Science and Technology “Accelerating Sustainable
               Towards Society 5.0
   45   46   47   48   49   50   51   52   53   54   55