Teste da hipótese de mercados adaptativos para o Brasil

Abstract

This dissertation aims to analyze the Brazilian stock market efficiency using Brazilian Stock Index (Ibovespa) daily data, from January 1995 to December 2012. Based on Efficient Market Hypothesis (EMH), used to analyze the predictability of stock returns, it was applied the tests of automatic variance ratio with wild bootstrap, developed for linear dependence, and the generalized spectral shape, used for non-linear conditions. When applying these tests, it was created fixed length moving subsample windows, testing the martingale for each move. With the results, it was possible to verify if the market efficiency depends on its moment conditions, as the Adaptive Markets Hypothesis (AMH) shows. As a conclusion, it was not possible to reject the HME, because, even with some periods statistically predictable, in the most of the time, the Brazilian stock market was efficient. This interchange of predictable and unpredictable periods also do not allowed the rejection of AMH, that was supported by the unimportant correlation existent for the predictability tests and the volatility of the stock returns volatility, discarding, this way, any state that the risk is responsible for the predictability existence.