Model Markov Switching Autoregressive pada Data Covid-19 di Indonesia

Abstrak

The Covid-19 pandemic has had a very influential impact on socio-economic conditions in Indonesia. Forecasting the number of Covid-19 cases is needed to support taking preventive action. The method that can be used to determine the number of Covid-19 cases is a forecasting method using the Markov Switching Autoregressive (MSAR) time series data model as an alternative for analyzing structural change data. This research uses Covid-19 confirmation data in Indonesia for the period March 2020-June 2021, with the aim of designing an MSAR model and calculating the magnitude of the transition opportunity in each state in the Covid-19 confirmation data in Indonesia. The MSAR model begins by describing the data and checking the stationarity of the data. After that, Box-Jenkins modeling was carried out to test heteroskedasticity and structural changes. Next, the MSAR model parameters were estimated and the transition matrix was formed. This research shows that the best MSAR model formed is the MS (2)-AR (5) model, with a static transition probability value in state 1 of 0.981330. However, it appears that there is a chance of 0.018670 for the Covid-19 confirmation condition to move to state 2. Testing in the case of state 2 produces a transition chance of 0.980991 in state 2, with a transition chance of Covid-19 confirmation changing to state 1 of 0.019009.