![IconGems-Print](images/IconGems-Print.png)
288 – Economic and Business Applications in High-Dimensional and Big Data Contexts
Inherently High-Dimensional Analysis with Indicator Saturation
Neil R. Ericsson
Federal Reserve Board
Indicator saturation is a generic approach to robust estimation, and to the detection and quantification of structural breaks. Saturation techniques are inherently high-dimensional and require automated model selection with non-standard inference. This paper characterizes several roles for saturation techniques and proposes extensions of impulse indicator saturation that have greater power to detect empirically common structural breaks. A model for the Brazilian inflation rate illustrates saturation techniques.