Revealing Volatility Dynamics in ARCH-GARCH Models

ARCH-GARCH models are notoriously difficult to interpret. Benton et al. (2026) show that interpreting the regressors in the heteroskedastic variance equation is particularly challenging and offer an approach to gaining inferences about these re- gressors beyond their sign and statistical significance. This manuscript implements their proposed strategy for dynamic inference through two commands: gbmake and gbplot. gbmake generates simulated parameter values from an estimated ARCH-GARCH model, using either a parametric- or residual-based bootstrap. gbplot produces dy- namic simulations using these bootstrapped values or through the delta method for approximating uncertainty. An applied example illustrates how these tools can help users better understand the effects of ARCH-GARCH model regressors on the condi- tional error variance.