Evaluating VAR models

After fitting a VAR model, the next step is to evaluate how well the model captures the interactions and dynamic relationships between the different endogenous variables (multiple time series). These evaluations provide insights into causality, variable influence, and how shocks to one variable propagate through the system.In this recipe, you will continue where you left off from the previous recipe, Forecasting multivariate time series data using VAR, by exploring various diagnostic tools to deepen our understanding of the VAR model’s performance. Specifically, you will do the following:

• Test for Granger causality to determine whether changes in one variable help predict changes in another
• Analyze the residual ACF plots to check whether residuals are uncorrelated (white noise)
• Use the Impulse Response Function (IRF) to measure how shocks to one variable affect others over time
• Perform FEVD to quantify each variable’s contribution to forecast error...