Inference in high-dimensional time series models

Luca Margaritella

    Research output: ThesisDoctoral ThesisInternal

    119 Downloads (Pure)

    Abstract

    Today’s world provides us with great potential in terms of data availability: “big data” is a term that very much circulates and many came across with. While having loads of data is a great opportunity to better understand the complexity of the real world, designing reliable statistical inference in such data-dense contexts requires careful modelling. Furthermore, when data such as time series is considered, the matter gets further complicated, given the inherent time dependency one needs to account for. This research develops statistical techniques aimed at both testing causal hypothesis and obtain forecasts in high-dimensional time series models. Applications of these techniques are provided in both finance, macroeconomics and climate econometrics, thus demonstrating the relevance of such tools across various sub-disciplines.
    Original languageEnglish
    Awarding Institution
    • Maastricht University
    Supervisors/Advisors
    • Smeekes, Stephan, Supervisor
    • Hecq, Alain, Supervisor
    Award date8 Nov 2021
    Place of PublicationMaastricht
    Publisher
    Print ISBNs9789464235104
    DOIs
    Publication statusPublished - 2021

    Keywords

    • high-dimensional Inference
    • time series models
    • Granger Causality

    Cite this