Abstract

In this paper, we apply a nonstructural approach to analyze coffee price returns behavior over time. In particular, we use the Hinich portmanteau statistic test (H) to detect nonlinear dependence of the International Colombian Arabica Coffee price return for the period of June 29, 1990 to July 1, 2010 (5,219 observations). Our results sheds light that in 10 out of 173 episodes, a standard volatility model like GARCH for instance, does not entirely capture all the nonlinearity patterns that are present in the data. And we report some events whose arrival might have induced nonlinear dynamics. Our findings also suggest that the weak form of the efficient markets hypothesis (EMH) cannot be supported in this market.

Key words: Colombian Arabica coffee price, Hinich portmanteau statistic test, nonlinear dependence.