Saving-Investment Correlations and Capital Mobility: On the Evidence from Annual Data

This paper evaluates the use of saving-investment correlations to measure capital mobility. It is argued that the intertemporal approach to the balance of payments predicts that the practice of calculating the correlation coefficient using long-term averages of saving and investment shares biases the result toward accepting the hypothesis of capital immobility. Calculations based on annual data show that the correlation coefficient is much lower and that is varies considerably. Estimates of intranational capital mobility based on saving and investment shares of U.S. federal states show that the saving and investment link is much looser within a nation than among nations.