Description

In this paper we investigate the possible presence of nonlinear dynamics for stock index returns and trading volume at the Chilean Stock Market. To capture any nonlinear behavior in the series we estimate Smooth Transition Autoregressive (STAR) models and test them against the linear alternatives. As a complement to this univariate approach, we use Markov-Switching Vector Autoregressive (MS-VAR) models to investigate the empirical relationship between both variables. The results clearly show that the Chilean Stock Market is characterized by the presence of nonlinear patters in both series (trading volume and stock returns) as well as in their joint relationship. The presence of nonlinearities is a key issue in testing the Efficient Market Hypothesis (EMH), according to which stock returns and trading volume should be not related. Previous researches on the efficiency using data from the Chilean stock market, using linear models, support the hypothesis. However, the nonlinear patters we found in the data are a clear signal of misspecification problems in a testing procedure based on a linear approach.