| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 2.05 MB | Adobe PDF |
Orientador(es)
Resumo(s)
É amplamente reconhecido que os valuation ratios fornecem, in-sample, indicações relevantes sobre os retornos futuros de ativos. Este padrão de previsibilidade é comum a uma larga maioria de mercados. Nesta dissertação, avaliamos a capacidade de certos valuation ratios para prever, out-of-sample, os retornos agregados para o mercado de ações e para o mercado imobiliário, nos E.U.A.. Aplicamos modelos lineares e filtros multivariados para gerar as previsões e utilizamos "poderosos" testes out-of-sample para fazer inferência estatística. Verificamos que existe evidência estatística que suporta a passagem dos resultados in-sample para um contexto out-of-sample. O rácio dividendo-preço e o rácio renda-preço apresentam uma capacidade significativa para prever os retornos de ações e imóveis, respetivamente. Notamos, contudo, que estes resultados podem depender da amostra. Sobretudo para o mercado de ações, o final da amostra (que inclui a recente crise financeira) pode ser o responsável pelos bons resultados.
It is well established that valuation ratios provide, in-sample, relevant signals regarding future returns on assets. This pattern of predictability is pervasive across financial markets. In this dissertation we assess the ability of valuation ratios to predict out-of-sample aggregate returns for the stock and the housing markets in the U.S.. We apply linear models and multivariate filters to produce the forecasts and employ powerful out-of-sample tests for inference. We find that there is statistical evidence supporting the extension of the in-sample results to an out-of-sample framework. The dividend-price ratio and the rent-price ratio display a significant ability for predicting stock and housing returns, respectively. Nevertheless, we note that these findings may be sample dependent. Especially for the stock market, the end of the sample, including the recent financial crisis, may be responsible for the good results.
It is well established that valuation ratios provide, in-sample, relevant signals regarding future returns on assets. This pattern of predictability is pervasive across financial markets. In this dissertation we assess the ability of valuation ratios to predict out-of-sample aggregate returns for the stock and the housing markets in the U.S.. We apply linear models and multivariate filters to produce the forecasts and employ powerful out-of-sample tests for inference. We find that there is statistical evidence supporting the extension of the in-sample results to an out-of-sample framework. The dividend-price ratio and the rent-price ratio display a significant ability for predicting stock and housing returns, respectively. Nevertheless, we note that these findings may be sample dependent. Especially for the stock market, the end of the sample, including the recent financial crisis, may be responsible for the good results.
Descrição
Mestrado em Econometria Aplicada e Previsão
Palavras-chave
Previsão Out-of-sample Retornos agregados Mercados Financeiros Valuation-ratios Forecasting Aggregate returns Financial Markets
Contexto Educativo
Citação
Sequeira, Ana Carla Natal da Silva (2012). "Predicting aggregate returns using valuation ratios out-of-sample". Dissertação de Mestrado, Universidade Técnica de Lisboa. Instituto Superior de Economia e Gestão.
Editora
Instituto Superior de Economia e Gestão
