Title: Nonparametric changepoint analysis of multiple time series
Authors: Elfred John Abacan - College of Arts and Sciences, University of the Philippines Visayas (Philippines) [presenting]
Erniel Barrios - University of the Philippines (Philippines)
Joseph Ryan Lansangan - University of the Philippines (Philippines)
Abstract: Analysis on changes in the level of time series helps in characterizing common components of multiple time series that helps identify the shared behavior of the data generating process with some known events that causes perturbations in the behavior of the time series. Change in variance of the error structure leads to model misspecification and more serious violation of other assumptions that facilitates model estimation. Volatility models are used to incorporate variance structure into the mean model, but this often suffers from overparameterization especially in multiple series data. A model for structural change in the variance component is estimated through the backfitting algorithm, this is used then as a reference for a test based on sieve bootstrap to detect changes in the variance of multiple time series. Simulation study shows the test performs well in terms of power and size.