Since many economic variables are non-stationary, a heavy burden is imposed among researchers to determine the stationarity of their data. One price to pay in using non-stationary data is that the conventional student t-test and F-distribution break down and become invalid. But the most serious problem faced when using non-stationary data, is the spurious regression problem. If at least one of the explanatory variables in a regression equation is non-stationary, it is very likely that the dependent variable in the equation will display a similar trend. When both dependent variable and regressor(s) in an equation are trend-dominated, we are likely to obtain highly 'significant' regression coefficients and high values for the coefficient of determination (R2), even if the variables are completely unrelated (Thomas 1996).