Forecasting Climatic Variables using Vector Autoregression (VAR) Model (Published)
Although it is a topic of global concern, climate change implementation is typically regional. Finding an adequate Vector Autoregression (VAR) model to predict temperature, rainfall, and cloud coverage for the Jessore region of Bangladesh was the goal of this research project. The stationarity of variables was determined by ADF, PP, and KPSS unit root tests. Granger causality test was used to verify the endogenity among the variables. Employing AIC, VAR (11) model found best. The parameters associated with the model were estimated using the ordinary least square approach. Forecast error variance decomposition and impulse response function were utilized to reveal structural analysis, and the outcome revealed endogenous in the future. The predicted value showed a trend toward increasing temperature and a trend toward decreasing rainfall and cloud coverage.
Keywords: Granger Causality, forecast error variance (FEV) decomposition, impulse response function, vector autoregression (VAR), white noise