Common Stochastic Trends in International Stock Markets: Testing in an Integrated Framework
23 Pages Posted: 20 Nov 2000
Date Written: October 2000
In this study we explore the implications for the identification of common stochastic trends among stock price indices of using data transformed in a "real dollar" basis. By applying a "general" VAR model where all the relevant variables (stock indices, consumer priced indices and the exchange rate) appear, we show that the expected results from the cointegration analysis differ substantially. In particular, the use of the "transformed" data pre-supposes that the Purchasing Power Parity condition has been imposed. If this is not the case then the adoption of the "transformed" data leads to an entirely different economic identification of the model where demand shocks play a crucial role and the coefficients of the cointegrating vectors should satisfy different restrictions. Other contributions of this paper is that it studies the problem under the light of recent developments in cointegration theory which allows us to implement our tests in the presence of 1(2) variables. Furthermore we fill a gap in previous studies by testing for the temporal stability of the cointegration results. Our results, on data for the USA and the UK, validate our approach since they make clear that the necessary restrictions for the Purchasing Power Parity to hold are not satisfied and moreover the results are sample dependent.
Keywords: International equity markets, 1(2) analysis, cointegration, identification, purchasing power parity, temporal stability
JEL Classification: G15, G12, G32
Suggested Citation: Suggested Citation