Does the Feldstein-Horioka Puzzle Exist? Theoretical Analysis and Empirical Evidence

In 1980 Feldstein and Horioka (F-H) found high saving-investment cross-section correlation among OECD countries and interpreted this result as evidence of low capital mobility. Such a inconsistency between the conventional wisdom about capital mobility in the last decades and its empirical measurement posed an embarrassing puzzle in international macroeconomics and stimulated a vigorous debate. Since the saving-investment association has been largely confirmed in many subsequent studies, economists have responded to the F-H conclusion by suggesting several ways to reconcile this empirical finding with the "observed" high capital mobility. In this paper we first review the extensive literature on this issue. We then present a new approach to the paradox focusing the attention on the dynamics of the current account. Starting from Obstfeld-Rogoff intertemporal equilibrium model, which contains transport costs in the real markets and no impediments to financial capital flows, we detect a non-linear relationship between the current account and the real interest rate. Our contention is that a current account adjustment mechanism exists; this generates small net external imbalances relative to total saving and investment. The F-H findings would be a natural empirical consequence, without any implication about the degree of gross capital mobility. In order to verify the predictions of the model, we employ an original econometric and statistical procedure emphasizing the non-linearity of the current account-real interest rate relationship. As a first step. we apply a set of non-linearity and non-linear co-trending tests on both the current account and the real interest rate series and on their relationship. Finally we estimate a non-linear VAR model including twentieth order Chebishev polynomials. We find that the relationship between the variables exists, is non-linear, and works in two ways to provide dynamic adjustment in the series