Author:
Guillermo Carlomagno: Central Bank of Chile and Center of Economic Research (CINVE)
Antoni Espasa: Department of Statistics and Instituto Flores de Lemus, University Carlos III of Madrid
Abstract
Macroeconomic variables are weighted averages of a large number of components. Our objective is to model and forecast all of the N components of a macro variable. The main feature of our proposal consists of discovering subsets of components sharing single common trends neither assuming pervasiveness of those trends, nor imposing special restrictions on the serial or cross-sectional idiosyncratic correlation. We adopt a pairwise approach and study its statistical properties. Our asymptotic theory works both, with xed N and T ! 1, and with [T;N] ! 1. An extension of our strategy allows a wide type of conditional and unconditional heteroskedasticity. The paper includes an application to the breakdown of the US CPI in 159 components.