摘要：Knowledge of dependence pattern in stock market has paramount importance for both theoretical and practical in financial markets. Their usefulness is wide, can be used in portfolio predictability (of portfolio) and risk management. The aim of this paper is to investigate the autoregressive dependence under the alternative perspective of quantile regression. Our study investigates a period from 2001 until 2012 daily returns of twenty stock markets in Latin America, Europe, USA and Asia-Pacific. Our results emphasize that the estimates obtained by quantile regression are different and more consistent than those by AR-GARCH. We conclude also that there is an asymmetric behavior of the investor, in association the quantiles with bear and bull markets.
