Is the Feldstein-Horioka Puzzle Really a Puzzle?*

The empirical evidence that national saving and domestic investment are correlated, also known as Feldstein-Horioka (1980) Puzzle, has received significant attention in the literature since Feldstein and Horioka interpret it as an evidence of low international capital mobility. This paper argues that there is nothing mysterious in the finding that investment and saving are correlated. Since the neoclassical growth theory predicts that in the steady state investment and saving will be proportional to output and therefore will grow at the same rate, it would be surprising if we did not find high long-run correlation between investment and saving. Using a dynamic intertemporal optimization model of open economy, it is shown that long-run investment saving correlation directly follows also from the dynamic budget constraint of open economy and this does not depend on the degree of international capital mobility. Therefore, unless the dynamic budget constraint is violated, the time series of investment and saving should be cointegrated, and this would be true for any degree of capital mobility. Using an improved econometric technique which encompasses the tests used by previous authors, I show that their conflicting findings can be explained by a simple, but important, omitted variables problem. Using annual and quarterly postwar U.S. data, I find that investment and saving are cointegrated in levels as well as in ratios (as a fraction of output) regardless of the time period covered, as predicted by the neoclassical growth theory. This indicates that the U.S. economy is solvent in the sense that it does not violate its dynamic budget constraint. Therefore, I conclude that the observed investment-saving correlation cannot be useful in measuring capital mobility.