Non-stationary dynamic factor models

• Autores: Francisco Corona
• Directores de la Tesis: Esther Ruiz Ortega (dir. tes.) , Pilar Poncela Blanco (codir. tes.)
• Lectura: En la Universidad Carlos III de Madrid ( España ) en 2017
• Idioma: español
• Tribunal Calificador de la Tesis: Maria Helena Lopes Moreira da Veiga (presid.) , María Jesús Sánchez Naranjo (secret.) , Javier José Peréz García (voc.)
• Enlaces
  • Tesis en acceso abierto en: e-Archivo
• Resumen

  • This dissertation focuses on studying two topics of large non-stationary Dynamic Factor Models (DFMs). A very common practice when extracting factors from non-stationary multivariate time series is to differentiate each variable in the system. As a consequence, the ratio between variances and the dynamic dependence of the common and idiosyncratic differentiated components may change with respect to the original components. In the first step, we analyze the effects of these changes on the finite sample properties of several procedures to determine the number of factors. In particular, we consider the information criteria of Bai and Ng (2002), the edge distribution of Onatski (2010) and the ratios of eigenvalues proposed by Ahn and Horenstein (2013). The performance of these procedures when implemented to differentiated variables depends on both the ratios between variances and dependencies of the differentiated factor and idiosyncratic noises. Furthermore, we also analyze the role of the number of factors in the original non-stationary system as well as of its temporal and cross-sectional dimensions. Finally, we implement the different procedures to determine the number of common factors in a system of inflation rates in 15 euro area countries.

    In the second step, we analyze and compare the finite sample properties of alternative factor extraction procedures in the context of non-stationary DFMs. On top of considering procedures already available in the literature, we extend the hybrid method based on the combination of Principal Components and Kalman filter and smoothing algorithms to non-stationary models. We show that, unless the idiosyncratic noise is non-stationary, procedures based on extracting the factors using the non-stationary original series work better than those based on differenced variables. The results are illustrated in an empirical application fitting non-stationary DFM to aggregate GDP and consumption of the set of 21 OECD industrialized countries. The goal is to check international risk sharing is a short or long-run issue.