Dynamic Factors of Macroeconomic Data

Abstract

This article gives a double-cycle algorithm to estimate the parameters of the dynamic factor model with given number of factors and order of the autoregressive process of the factors. In the inner cycle compromise factor decomposition, a generalization of the eigenvalue-eigenvector decomposition of principal component analysis is used to nd extrema of sums of heterogeneous quadratic forms that has not been used before. Application of the algorithm for macroeconomic indicators of the Hungarian economy since the 1990's is also discussed.