Factor models with vector autoregressive dynamics

• Grant number: 23/11547-2
• Support Opportunities: Scholarships in Brazil - Doctorate (Direct)
• Start date: September 01, 2023
• End date: January 31, 2028
• Field of knowledge: Physical Sciences and Mathematics - Probability and Statistics - Statistics
• Principal Investigator: Pedro Alberto Morettin
• Grantee: Davi Oliveira Chaves
• Host Institution: Instituto de Matemática e Estatística (IME). Universidade de São Paulo (USP). São Paulo , SP, Brazil
• Associated research grant: 23/02538-0 - Time series, wavelets, high dimensional data and applications, AP.TEM
• Associated scholarship(s): 25/25711-4 - Quasi-Maximum Likelihood Estimation of Dynamic Factor Models with Time-Varying Loadings, BE.EP.DD

Abstract

Factor Models are extensively used as a resource to reduce the dimensionality (number of variables) and to explain the common variability in multivariate data sets. The idea, like in Principal Component analysis, is to consider a small set of factors, which explain most of the variance in the data. In the context of multivariate time series, the analysis of the covariance structure becomes more difficult, especially when the time series are non-stationary, show co-movements, time transitions and structural changes. In this Project we propose a factor model where the factors follow a vector autoregressive dynamics and the coefficients are time-varying and, as a consequence, the variances are also time-varying. Relevance: This research line lies in the state-of-art in problems involving multivariate non-stationary time series, with applications in finance, macroeconomic data in presence of transitions between economic recessions or the impact of political crises. (AU)