Comparativo de metodologias de mensuração de VAR para o mercado financeiro brasileiro

• Ano de defesa: 2007
• Autor(a) principal: Chun, Liu Yuan
• Orientador(a): Matone, Ricardo
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Palavras-chave em Português: Gerenciamento de Risco; Metodologias de mensuração de risco de mercado; Avaliação de metodologias de mensuração de risco de mercado
• Palavras-chave em Inglês: Risk management; VAR (Value at risk); Methodologies to measure market risk
• Link de acesso: http://hdl.handle.net/10438/2089

Many methodologies to measure market risk have been developed and improved in the last few decades. While some methodologies are non-parametric, others are parametric. Some methodologies are theoretical, while others are more practical. While some methodologies are original, others are hybrid. In this work, we compared methodologies to measure market risk in the Brazilian market. We evaluated non-parametric and parametric methodologies to measure VaR in a fixed income, equities and a mixture between fixed income and equities’ portfolios between 2000 and 2006. Non-parametric methodologies evaluated were: Fixed-Weight Historical Simulation, Fixed-Weight Antithetic Historical Simulation, Exponential Historical Simulation and Scenario Analysis. Parametric methodologies evaluated were: FixedWeight VaR Delta-Normal, Exponential VaR Delta-Normal (EWMA), Fixed-Weight Monte Carlo Simulation and Exponential Monte Carlo Simulation. Based on statistical measures of conservatism, accuracy and efficiency, we compared the methodologies.