Exploring The Methods of Cointegration Procedures Using Stock Prices

Abstract

Stationary models are essential class of stochastic models for describing time series data which have received a great attention. In reality, however, business and economic data are non-stationary multivariate time series that are often better understood by cointegration analysis. This study investigates the cointegration testing methods of Engle-Granger two-step estimation technique, Phillip-Ouliaris test and Johansen's multivariate test. The stock prices of selected companies in Nigeria from 2008-2014 are used in the study. Findings revealed that the three techniques produced different results and that the Johansen's method and Engle-Granger two steps procedure exhibits higher efficiencies than Phillips-Ouliaris methods but their efficiency is dependent on the number of variables and correct selection.