Robust Estimation of Clustered Multiple Time Series Model with Structural Change Resi O. Olivares A Thesis Presented to the Faculty of the University of the Philippines in Candidacy for the Degree of Master of Science in Statistics Recommended for Acceptance by the School of Statistics Adviser: Dr. Erniel B. Barrios March 2014 ABSTRACT We postulate a model for a clustered multiple time series where individual and clusters-specific effects were represented by random components. To induce robustness during episodes of temporary structural change, we use the forward search and bootstrap methods in the backfitting algorithm to estimate the parameters. Simulation studies show that the resulting models possess high predictive ability specially in long time series where structural change usually occur. Keywords: multiple time series, cluster, structural change, backfitting, additive model, forward search bootstrap