• Refine Query
  • Source
  • Publication year
  • to
  • Language
  • 1
  • Tagged with
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • 1
  • About
  • The Global ETD Search service is a free service for researchers to find electronic theses and dissertations. This service is provided by the Networked Digital Library of Theses and Dissertations.
    Our metadata is collected from universities around the world. If you manage a university/consortium/country archive and want to be added, details can be found on the NDLTD website.
1

[en] CAPITAL FLOWS TO EMERGING MARKETS: THE CASE OF BRAZIL / [pt] FLUXOS DE CAPITAIS PARA ECONOMIAS EMERGENTES: O CASO DO BRASIL

THIAGO GUEDES MORAIS 26 September 2022 (has links)
[pt] Motivados pela posição de destaque do real brasileiro entre uma das moedas mais depreciadas em relação aos seus pares emergentes em meados de 2020, potencialmente fomentada pela expressiva evasão de capitais observada no decorrer da pandemia COVID-19 que culminou com um déficit no mercado cambial, realizamos previsões um trimestre a frente para os fluxos de capitais líquidos para o Brasil através de técnicas de machine learning, utilizando modelos de regularização para seleção das variáveis importantes. Os fluxos são obtidos a partir de dados trimestrais do balanço de pagamentos, englobando 2004:T1 a 2021:T1. Os modelos propostos, tanto LASSO quanto adaLASSO + OLS, foram capazes de gerar previsões fora da amostra melhores que o modelo de benchmark, AR. Apesar disso, quando comparados entre si, não podemos rejeitar a hipótese nula de que os modelos propostos possuem a mesma precisão de previsão. / [en] Motivated by the prominent position of the Brazilian real among the most depreciated currencies in comparison with its emerging peers in mid-2020, potentially fueled by the significant capital outflow observed during the COVID19 pandemic that resulted in a deficit in the foreign exchange market, we make one quarter-ahead forecast for net capital flows to Brazil through machine learning techniques, using shrinkage methods to select important variables. These flows are computed from quarterly balance of payments data from 2004:Q1 to 2021:Q1. The proposed models, both LASSO and adaLASSO + OLS, were able to generate better out-of-sample forecasts than the benchmark model, AR. Nevertheless, when compared to each other, we cannot reject the null hypothesis that the proposed models have the same forecast accuracy.

Page generated in 0.0321 seconds