Causality and Cointegration Analysis: Evidence from the Brazilian Stock Market

Abstract

Many studies have focused on examining the cointegration and causality between or among stock markets of different countries. This paper departs from these traditional inter-relationship studies through its investigation on the causality and cointegration between the Brazilian Stock Market (Bovespa) and a listed company (Petrobras) by employing the Granger causality test and error correction technique based on autoregressive distributed lag (ARDL) modelling approach to cointegration. We find empirical evidence of cointegration and that deviation from long-run equilibrium is corrected according to the speed of adjustment. In particular, a disequilibrium resulting from a shock to the stock market is corrected by 3.8% per week. Our findings also show a unidirectional causality running from Bovespa index to share price of Petrobras thus revealing the predictive power of the former. While our Granger causality finding is inconsistent with the preaching of efficient market hypothesis (EMH), it nonetheless fortifies the need for investors and financial analysts to closely monitor the movements of the Brazilian stock market index when investing (or analyzing changes) in Petrobras.

Keywords: Bovespa, Petrobras, Cointegration, Causality, Equilibrium, Short-run, Long-run