Investment-Saving Comovement and Capital Mobility : Evidence from Century Long U.S. Time Series

This paper makes three contributions. First, I construct annual time series of gross domestic investment and national saving in the United States for the 1897-1949 period using historical component series. I compare the qualitative and quantitative properties of the newly constructed series with the properties of four existing alternative series constructed by the Bureau of Economic Analysis, the Commerce Department, Kuznets, and Kendrick. Second, I combine the newly constructed data with the Bureau of Economic Analysis' 1929-1989 period data, and the resulting time series are used to reexamine and document the long-run bivariate relationship between the time series of investment and saving. Third, I also examine the short-run as well as the cyclical relationships between the time series of investment and saving. The results reported in this paper indicate that there is a strong long run and cyclical relationship between investment and saving, and this relationship seems to be independent of the time period considered. Furthermore, I find that during the postwar period the investment-saving comovement is strong and significant also in the short run. However, this is not true during the prewar period. Quantitatively, I find that the investment-saving relationship is stronger during the postwar period than during the prewar period. Feldstein and his coauthors have argued that the high investment-saving correlation reflects imperfect capital mobility. This view, however, is hard to reconcile with the finding that the correlation increased during a period in which it is largely believed that capital markets have become more open and integrated. I conclude, therefore, that long-term capital mobility tests based on investment-saving correlation analysis are not likely to provide an accurate measure of capital mobility