The pairwise approach to model a large set of disaggregates with common trends

摘要: The objective of this paper is to model and forecast all the components a macro orbusiness variable. Our contribution concerns cases with large number (hundreds) ofcomponents where multivariate approaches are not feasible. We extend in several directions pairwise approach originally proposed by Espasa Mayo-Burgos(2013) study its statistical properties. consists on performing common features tests between N(N-1)/2 pairs series that exist group N them. Once done, groups share can be formed. Next, using single equation models include restrictions derived features. In we focus discovering trends. asymptotic properties procedure studied analytically. Monte Carlo evidence small samples performance provided correction designed. A comparison DFM alternative also carried out, results indicate dominates many empirically relevant situations. advantage it does need pervasive. strategy for dealing outliers breaks context designed Carlo. Results treatment these observations may considerably improve procedure's when 'contaminated'.