skip to main content
Idioma:

Contagion in economic networks: a data-driven machine learning approach

Silva, Michel Alexandre Da

Biblioteca Digital de Teses e Dissertações da USP; Universidade de São Paulo; Instituto de Ciências Matemáticas e de Computação 2022-05-27

Acesso online

  • Título:
    Contagion in economic networks: a data-driven machine learning approach
  • Autor: Silva, Michel Alexandre Da
  • Orientador: Rodrigues, Francisco Aparecido
  • Assuntos: Contágio; Economia; Redes Complexas; Risco Sistêmico; Complex Networks; Contagion; Economic System; Systemic Risk
  • Notas: Tese (Doutorado)
  • Descrição: Interconnectedness is pervasive in economic systems. This allows several economic issues to be analyzed through complex networks tools. Interconnectedness can be beneficial to economic agents through, for instance, risk-sharing in financial networks. However, the 2008 financial turmoil, whose main episode was the collapse of Lehman Brothers in September of that year, highlighted the importance of interconnectedness in the propagation of shocks i.e., contagion through economic systems. Despite its importance, there are still some open issues concerning contagion in economic networks, its consequences, and the processes governing its dynamic. In this thesis, we aim to shed some light on some of these open issues. To perform this task, we rely on tools suitable for the analysis of complex systems complex networks, machine learning (ML), and agent-based modeling , as well as several unique Brazilian databases. Our contributions address three broad questions: i) the identification of systemically relevant economic agents (banks, firms, and assets), ii) the dynamics of monetary policy shocks propagation and its interplay with the financial network topology, and iii) the impact of heterogeneous loss distribution mechanisms on systemic risk (SR). Our main conclusions are the following: i) interest rate shocks affect financial stability in a non-linear way and this effect is stronger in periods of monetary policy tightening, ii) ML techniques can successfully identify drivers of SR among financial and topological variables, iii) the adoption of a heterogeneous loss distribution rule significantly increases SR, iv) topological features of the bank-firm credit network are significantly affected by shocks to the policy interest rate, and v) the newly developed centrality measure, the risk-dependent centrality, captures better the dynamics of the external risk level than other centrality measures.
  • DOI: 10.11606/T.55.2022.tde-13072022-134420
  • Editor: Biblioteca Digital de Teses e Dissertações da USP; Universidade de São Paulo; Instituto de Ciências Matemáticas e de Computação
  • Data de criação/publicação: 2022-05-27
  • Formato: Adobe PDF
  • Idioma: Inglês
Links
Voltar para lista de resultados

  • Realização: ABCD-USP USP   Logos de Redes Sociais: Freepik

Buscando em bases de dados remotas. Favor aguardar.