Implications Of Mixed Results With Cointegration Models In International Studies

Abstract

While cointegration models have been used extensively in many fields (such as in testing for stock market predictability), there have been mixed results from using cointegration models in international studies.  For example, one study attempted to regress the levels of the general stock market price indexes of Norway, Mexico, Venezuela, and of Oman on the levels of interest rate and oil prices variables during the period 1992 to 1999.  Attempts to explain mixed results have included distinguishing developed markets versus less developed markets and differences in the volatilities of markets. Another approach has been attempts to improve cointegration methods.  Some studies have included both bivariate and multivariate methods of cointegration.  Others have employed augmented Dickey-Fuller unit root tests of the residuals.  The extent of mixed results in international studies has implications for earlier studies in single countries.  It is possible that many earlier studies in business and in economics will need to be replicated using newer techniques developed by international researchers.  The purpose of this study is to explain the attempts to improve cointegration methods, the mixed results in international studies, and the implications for earlier studies relying upon cointegration models.