Economics Letters 39 (1992) 55-58 North-Holland 55 Are saving and investment cointegrated? Another look at the data 0. David Gulley * zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA Bentley College, W altham MA, USA Received 29 August 1991 Accepted 24 March 1992 In his paper ‘Are saving and investment co-integrated’ (Economics Letters 27, pp. 31-34), Miller finds that U.S. saving and investment rates were cointegrated during the post-WWII fied exchange-rate regime and not cointegrated during the flexible exchange-rate regime. Miller did not take into account the fact that both of these rates have non-zero means. Once this is taken into account, I find that his results no longer hold. I also find that the levels of saving and investment are not cointegrated during either the fixed or flexible exchange-rate regimes. These findings cast doubt on whether or not this technique can shed light on international capital mobility. 1. Introduction Over the last decade, numerous papers have appeared that attempt to measure the degree of international capital mobility. Feldstein and Horioka (19801, Feldstein (1983), Wong (19901, and Frankel (1991) are just a few of the researchers who have looked at this question. As have previous authors, Miller (1988) examines the relationship of domestic saving and investment to assess the mobility of international capital across the borders of a given country. Using U.S. data, he applies the recently developed technique of cointegration to this question. He notes that many of the previous studies are open to criticism of the econometric techniques used. He argues that testing domestic saving and investment for a long-run relationship during different exchange-rate regimes can shed light on the mobility of international capital. He concludes that the gross national saving rate, which is gross national saving divided by GNP (denoted GNSR), had a long-run relationship with the gross private domestic investment rate (GPDIR) using data for the period 1946 : 1 to 1971: 2. However, using data for the 1971: 3-1987 : 3 period, these two rates did not have a long-run relationship. Miller argues that this could be the result of increased capital mobility across U.S. borders, which would cut the link between domestic saving and investment that must exist in a closed economy. I find that Miller’s initial tests of the time-series properties of the saving and investment rate are not done properly. I also find that the levels of saving and investment are non-stationary, but not Correspondence to: 0. David Gulley, Department of Economics, Bentley College, Waltham, MA 02154, USA. * Thanks to Stephen Grubaugh and Stephen Miller for providing helpful comments. 0165-1765/92/$05.00 0 1992 - Elsevier Science Publishers B.V. All rights reserved