Electronic International Standard Serial Number (EISSN)
Dynamic factor models provide a useful way to model large sets of time series. Thesedata often have heterogeneity and cluster structure and the formulation and estimationof dynamic factor models should be adapted to these features. This article presents aprocedure to fit Dynamic Factor Models with Cluster Structure (DFMCS), where someof the factors are global and others group-specific, to heterogeneous data that mayincludemultivariateadditiveoutliersandlevelshifts.Theprocedurestartswithaninitialcleaning of the times series from outlying effects. Then a first estimation of the possiblefactors is applied to the cleaned data and these factors are used to build the commoncomponent of each series. The groups are found by studying the joint dependency ofthese common components. Then, additional factors are estimated by using the seriesin each cluster and, finally, all the factors found are classified as global or group-specific.We show in a Monte Carlo study that the procedure works well and seems to be betterthan other alternatives in terms of estimation of factors and loadings as well as in termsof misclassification rates for the series. An example of an electricity market is presentedto illustrate the advantages of cleaning for outliers and taking into account the clusterstructure for understanding and forecasting.
clustering time series; dependency measures; multivariate additive outliers; multivariate level shifts; principal components