Structural attribution of observed volatility clustering

Virtually all nonlinear economic models with independent, identically distributed stochastic shocks and time-invariant structural parameters will generate persistent, partially predictable heteroskedasticity (“volatility clustering”) in their key dependent variables. This paper offers some examples of this phenomenon, derives i.i.d. shock, time-invariant structural forms which generate various types of observed volatility clustering, and examines the modeling and forecasting implications of such “structural attribution.”