Resumen de Efficient market hypothesis and anomalies: empirical studies on major financial markets
This dissertation is presented as three self-contained studies that concern market efficiency issues. Through these studies, we ares trying to help answer the questions whether the financial markets are efficient and why Contribution 1 tests the profitability of momentum strategies in the United Kingdom, Germany, Japan, and China over the period 1991 to 2006 and sub-periods. As a result, we find that the United Kingdom and Germany exhibit rather stable medium-term return continuation for both RSS and WRSS over the entire sample period and sub-periods, while Japan shows a medium-term return reversal over the sub-period 1991-1998. As for china, we report momentum profits over the period 1995-2006 and the sub-period 2001-2006 with RSS. Furthermore, we use the results of RSS to check the influence of risk factors and transaction costs on the momentum returns, as well as calendar effects and other characteristics of momentum portfolios reported in literature. With the results of WRSS, we examine the relative importance of time-series predictability and cross-sectional variation in accounting for the profits of momentum strategies.Contnbution 2 uses the cross-sectional variance of the betas to study herd behavior towards market index in major developed and emerging financia! markets (categorized as developed group, Asian group, and Latín American group). we propose a robust regression technique to calcúlate the betas of the CAPM and those of the Fama-Frenen three-factor model, with an intention to diminish the impact of multivariate outliers in return data. Through the estimated valúes obtained from a state space model, we examine the evolution of herding measures, especially their pattern around sudden events such as the 1997-1998 financial crises. This 1997-1998 turmoil turns out to have formed a turning point for most of the financial markets. we document a higher level of herding in emerging markets than in developed markets. We al so find that the correlation of herding between two markets from the same group is higher than that between two markets from different groups. in Contribution 3, we apply a GARCH model to examine the cross-autocorrelation pattern between daily returns of porfolios composed of dual-usted stocks in Chinese stock markets, before and after China opened its once foreign-exclusive B-share market. A lead-lag relationship between the A-share and B-share portfolio returns is identified during our sample periods, with the A-share portfolio leading the B-share portfolio. Upon the opening of B-share market, a change from underreaction to overreaction is found in the response pattern of B-share market, producing a rarely seen negative cross-autocorrelation. The results of two additional tests are reported. First, by decotnposing the portfolio return into portfolio-specific and market-wide returns, we find that the market-wide information contained in A-share portfolio return is strongly associated with the cross-autocorretlation structure. Second, we document a directional asymmetry in wihch B-share portfolio shows either slow or over response to bad, but not good, news of A-share portfolio. Using Granger Causal ity test, we observe a unidirectional volátil ity transmission from A-share market to B-share market before the opening of the B-share market, compared to a bidirectional transmission after the opening. we conjecture that information asymmetry alone is not enough to explain the empirical findings, and investor behavior must be taken into consideration
